Milan Milanović | May 28, 2020

How T-Mobile’s Merger with Sprint is Changing the Game for 5G

After almost two years and numerous legal battles, T-Mobile and Sprint are finally one. Equipped with a vast arsenal of spectrum assets, the new company aims to create an unparalleled 5G experience from urban areas to underserved rural communities across America. The biggest prize of this merger is 150 MHz of Sprint’s 2.5 GHz spectrum across top 100 markets. Combining Sprint and T-Mobile assets, T-Mobile now controls 319 MHz of sub-6GHz spectrum nationwide, and they have already started injecting some of this valuable capacity into the network and making it available to Americans.

The Philadelphia experiment

Because of the leasing and permitting work done in Philadelphia months ahead of the April 1, 2020 merger approval, T-Mobile was able to activate a number of 2.5 GHz (n41) sites on day one, followed by the official market launch on April 21. Massive MIMO radios were overlaid on T-Mobile’s existing cell sites, adding 60 MHz of spectrum in the 2.5 GHz band. Since Philadelphia wasn’t one of the nine markets where Sprint launched 5G, the fallow 2.5 GHz spectrum was readily available for use. T-Mobile had already covered Philadelphia with 5G in the 600 MHz band (n71), and the 2.5 GHz layer should provide a much-needed capacity boost. We can already see the increase in performance over the past four months.
t-mobile_5G_performance_philadelphia_0520-1

An analysis of all 5G tests taken on T-Mobile’s network in Philadelphia shows mean download speeds doubling during the month of April. A huge jump from 60.40 Mbps to 119.82 Mbps coincides with the rollout of 5G on the fresh mid-band spectrum. During the same month, we observed peak 5G download speeds exceeding 700 Mbps. Mean latency also fell from 30 ms to 21 ms over the period. Unlike the millimeter wave frequencies, which could offer greater speeds in a very small footprint, the 2.5 GHz band provides a good balance between coverage and capacity for mobile use.

T-Mobile’s layer cake in Manhattan

On May 5, T-Mobile launched its second mid-band 5G market in New York City. This launch marked the world’s first 5G network built on low, mid and high frequency bands. Paired with the multi-gigabit fiber backhaul, this 5G network is designed to deliver uncompromised performance. T-Mobile calls this the “Layer Cake 5G strategy.”
t-mobile_5G_performance_nyc_0520-1

Mean download speeds of T-Mobile’s 5G network in New York City increased from 79.18 Mbps in January 2020 to 98.96 Mbps in May (as seen through the 15th of the month). This represents a 25.0% jump across all three 5G layers. Mean upload speeds have also improved 11.0% during the same period.
t-mobile_5G_performance_NYC_Philadelphia_0520-1

Analyzing week-over-week performance, Speedtest Intelligence^® offers unique insights on the upward-trending download speeds on T-Mobile’s 5G across these two markets. A combination of additional 600 MHz spectrum explained in our previous article and the new 2.5 GHz 5G layer were likely the main drivers behind this improved performance.

Field testing the layers of 5G

Users are typically not aware of the layers of technologies that they’re utilizing when they connect to a 5G network. To better understand the performance and capabilities of T-Mobile’s layer cake 5G network, I looked at the performance of each individual 5G layer separately. This process involved running over 300 tests on the ground in New York City between May 5 and 19 (while fully complying with the social distancing and health guidelines) and logging modem diagnostic messages. This allowed me to observe individual throughput contributions across three different 5G layers on two different devices, the Samsung Galaxy S20 Ultra 5G and the OnePlus 8 Pro. While both devices are powered by Qualcomm’s latest Snapdragon 865 5G Mobile Platform, only the Samsung Galaxy S20 Ultra 5G (and S20+ 5G) has the required mmWave hardware components to fully access the 5G spectrum layer cake. The open-market OnePlus 8 Pro could only access sub-6 GHz 5G.

According to T-Mobile’s published map of projected 2.5 GHz coverage, the newly launched mid-band 5G layer is mostly available in the eastern parts of Manhattan. That meant a quick ride to the areas east of Herald Square in Midtown as well as parts of East Village. Keep in mind that the walk tests were collected on a recently launched and lightly loaded mid-band 5G network. This report samples what this network is capable of and should not be considered predictive of future performance under normal use.

LAA boosts speeds

License Assisted Access (LAA) is LTE technology that leverages the unlicensed 5 GHz spectrum. While often overlooked, this technology has been deployed on T-Mobile sites in New York City since 2017, and adds ample downlink capacity (40-60MHz) using Ericsson’s micro Radio 2205, “assisted” by the anchor in the licensed LTE mid-band. This can significantly boost download speeds on capable devices, while at the same time offloading traffic from the licensed spectrum and improving the LTE experience for all users served by that cell.

In Manhattan, many low-build rooftop sites with Ericsson AIR 5121 mmWave radios originally received the LAA treatment and are excellent candidates to receive (or have already received) the 2.5 GHz 5G overlay. This means these NYC sites would gain 40-60 MHz of LAA downlink capacity in addition to the full three layer 5G deployed in the 600 MHz low-band, 2.5 GHz mid-band, the 28 GHz mmWave high-band spectrum and LTE. This would open up a variety of options in terms of network scheduling and capacity management.

What each NR layer contributes to download speeds

Based on personal field test data logged over the course of the last eight months, the observed peak download contribution of the NR physical layer (PHY) on Sprint’s (now decommissioned) 2.5 GHz 5G network using Nokia radio and 40 MHz bandwidth was 367 Mbps, achieved on Sprint’s 5G launch day in New York City, August 27, 2019.

T-Mobile’s new 2.5 GHz deployment — leveraging the same amount of spectrum, Ericsson AIR 6488 radio access, under similar network conditions — produced a peak downlink NR contribution in excess of 541 Mbps, an improvement in spectral efficiency of 47 percent. The downlink NR contribution measures the portion of the download speed arriving at the device only from the 5G layer, the remainder of the download speed comes from LTE.

Furthermore, on May 19, the NR bandwidth on some sites was increased from 40 MHz to 60 MHz, and I’ve since observed a peak NR downlink contribution approaching 900 Mbps. The increased bandwidth was observed on three locations in the East Village, while the majority of 2.5 GHz sites are still at 40 MHz. We should stress again that these peak 5G speeds are meant to show the full potential of this new network deployment and have been achieved in good outdoor signal conditions, on a lightly loaded 5G network.

When taking into account the 3:1 TDD frame configuration (75% downlink 25% uplink), the effective peak downlink spectral efficiency of the 2.5 GHz NR layer looks like this:
peak-spectral-efficiency_0520-2

Lower spectral efficiency on Sprint’s legacy network could be attributed to the lower distribution of 256 QAM and MIMO (Rank 3, 4). Granted, there could have been a potential performance penalty associated with an earlier software load on Nokia’s massive MIMO radios running in split-mode. These radios were logically partitioned for both n41 and LTE B41 concurrent operation, while pushing carrier aggregation across three LTE CCs and using 100-120MHz of spectrum in total. In contrast, T-Mobile’s existing deployment consists of Ericsson AIR 6488 radios delivering only n41, while LTE mid-band anchor bands are delivered from separate radios (B2, B66).

40 MHz of n41 delivered NR upload speeds of 33 Mbps on T-Mobile vs. 23 Mbps on Sprint, a 43.5% increase. Tests from May 19 using 60 MHz wide NR channel produced consistent speeds of 50 Mbps arriving from the NR layer.

peak-upload-contributions_0520-2

The benefits for both users and the operator are apparent as the aggregate (LTE B66 + NR n41) upload speeds achieved on capable user devices now well exceed 100 Mbps.

In addition to the new mid-band 5G (n41), T-Mobile’s low-band 5G (n71), which launched in early December, is providing the foundational layer for the future SA (Standalone) NR. The initial 5 MHz FDD deployment was, from my first-hand experience, capable of boosting the downlink performance at the capable user terminals with an additional 42 Mbps. As of the middle of March, that allocation was increased to 10 MHz, now delivering peak n71 speeds of roughly 100 Mbps. With wider NR channels comes an improved spectral efficiency.

Millimeter wave adds capacity

T-Mobile’s millimeter wave (mmWave) NR layer (n261) has been available in New York City since June 2019. Leveraging the existing macro grid, T-Mobile’s Manhattan mmWave overlay, in terms of cell site density, is one of the most impressive in the world. This layer leverages 100 MHz of spectrum (2×50 MHz) in the 28 GHz band (n261), adding peak n261 speeds of over 520 Mbps in addition to the LTE anchor bands. In comparison, T-Mobile’s mid-band 2.5 GHz NR layer is able to deliver similar capacity out of just 40 MHz of spectrum. This jump in spectral efficiency is attributed to the use of up to four data streams and 256 QAM modulation in the mid-band, something the current generation of mmWave chipsets can’t do.

During my testing, the Samsung Galaxy S20 Ultra 5G was able to switch between all three NR spectrum bands seamlessly and deliver consistent performance at hundreds of megabits per second. While 100 MHz of mmWave can deliver a speed boost of a few hundred megabits per second, the full potential of this spectrum can be unlocked by deploying much wider 400 MHz or 800 MHz blocks of mmWave spectrum, which can produce speeds of multiple gigabits per second, just on that layer alone. T-Mobile now controls 1,160 MHz of mmWave spectrum nationwide, including significant amounts of not-yet-deployed 24 GHz and 47 GHz licenses won in recent auctions.

Device limitations (and opportunities)

T-Mobile’s 5G network uses a mechanism called the EUTRA-NR Dual Connectivity (EN-DC) split bearer, which allows for combining of data traffic from LTE (master node) and 5G (secondary node) both on the downlink (sub-6 and mmWave) as well as on the uplink (sub-6). This improves user experience by enabling utilization of the maximum amount of data streams across two technologies supported by capable 5G devices. Both devices are capable of aggregating two mid-band LTE anchor bands (B2, B66) with sub-6 NR, but only the Samsung Galaxy S20 Ultra 5G has the ability to leverage up to four spatial streams on the two mid-band LTE bands and NR (n41) at the same time. The OnePlus 8 Pro supports EN-DC combinations with two streams per the mid-band LTE anchors paired with four streams on n41. This is likely related to the RF Front End design limitation on the OnePlus device.
field-test-t-mobile_5G_device_nyc_0520-1

In my field tests, both devices delivered excellent speeds, leveraging 40 MHz of 2.5 GHz 5G (n41) with two LTE mid-band component carriers (CC) on the downlink, while the OnePlus 8 Pro had an additional help coming from the LAA CCs. On the uplink, devices utilized dual connectivity combining one LTE CC with NR. The latency was also similar between devices, although I did observe latency as low as 7 ms.

What’s missing from the Galaxy S20 Ultra 5G — and what the OnePlus 8 Pro fully capitalizes on — is the ability to aggregate mid-band LTE with LAA and mid-band NR. As we’ve described, the existing Manhattan LAA cell sites in some areas conveniently located on every two to three city block corners are the obvious candidates for the 5G mid/high-band overlay, which is exactly what I observed on the streets of the East Village. The combination of 60 MHz of LAA, 30 MHz of mid-band LTE (B66+B2) and 60 MHz NR produced an outstanding user experience, delivering mean download speeds of 989 Mbps and mean upload speeds of 107 Mbps. During May 19 testing, I recorded peak speeds in excess of 1.2 Gbps with the NR contribution of 900 Mbps. This particular test utilized only a single LTE CC (B66), three LAA CCs and 60 MHz of NR.

field-test_Oneplus-8-pro-performance_0520-1

Considering that Samsung’s Galaxy S20 lineup uses state-of-the-art RF Front End components and is built to provide uncompromised user experience, we really hope to see LAA EN-DC support added via the future software updates for both sub-6 and mmWave, which would unlock the device’s full potential. The business case for operators is very appealing: the ability to free up the licensed LTE spectrum in densely populated areas while at the same time providing a significant boost in user experience, as shown in this report. It’s also worth noting that the branded Samsung Galaxy S20 lineup does support the use of higher order MIMO on both LTE mid-band and n41 concurrently, which could prove to be an advantage over the OnePlus device outside of the dense urban areas where LAA isn’t available.

How unique device features boost performance

While most smartphones today can’t aggregate multiple low-band component carriers across multiple low-band frequencies, T-Mobile’s branded devices powered by Qualcomm’s Snapdragon X55 5G Modem-RF System are capable of aggregating multiple low-band channels within the same frequency band (600 MHz). In other words, these smartphones (including the Samsung Galaxy S20 Ultra 5G, and the OnePlus 8 Pro) are able to leverage all of the deployed downlink 600 MHz capacity by way of EN-DC, which allows for concurrent use of LTE band 71 and n71, in addition to the use of mid-band LTE anchors (Band 2 and Band 66). These devices also support uplink carrier aggregation of the LTE anchor band and sub-6 NR.

Because the OnePlus 7T Pro 5G McLaren Edition was the first device to receive this feature, we analyzed its performance in the U.S. as a whole during the first four months of 2020. The software update on February 13 resulted in an improvement of over 30% in mean download speeds from those seen in January 2020. The concurrent use of multiple 600 MHz channels produced mean download speeds of 82.35 Mbps and a latency of 32 ms in April.
t-mobile_5G_performance_oneplus_0520

What the future holds

In markets like New York City, Sprint’s legacy 2.5 GHz LTE network (for the most part) operates on a 60 MHz contiguous chunk of spectrum (3CC LTE carrier aggregation) adjacent to T-Mobile’s 40 MHz (in some parts 60 MHz) NR. In addition to this 100 MHz contiguous spectrum block, Sprint uses a separate 20 MHz slice for its small cells applications, for a total of 80 MHz of deployed 2.5 GHz LTE capacity. Repurposing 20 MHz (or more) of LTE channels and widening the 2.5 GHz NR capacity could easily provide a quick incremental capacity gain for T-Mobile, as already seen on some sites in Manhattan.

Furthermore, the existing Nokia 64T64R massive MIMO radios used on Sprint’s sites still operate in split-mode (32T32R for LTE, 32T32R for NR), but the NR logical partition is not in use. These could have the NR portion of antenna elements remapped for 64T64R LTE which should, in theory, further enhance the LTE coverage through the use of beamforming and deliver an improved LTE sector capacity for users with legacy Sprint LTE B41 devices. When we consider that over 80% of Sprint’s postpaid subscribers own smartphones capable of accessing the T-Mobile network, and 10 million unique Sprint subscribers already utilize T-Mobile’s LTE network on a weekly basis, the accelerated refarm of the 2.5 GHz spectrum makes even more sense.

While T-Mobile could easily add LTE B41 capability to its n41 sites, most (if not all) T-Mobile branded devices aren’t currently capable of aggregating LTE B41 channels with all of T-Mobile’s existing LTE spectrum bands. Therefore, adding LTE B41 wouldn’t provide any immediate capacity gain to T-Mobile’s user base. To achieve that capacity gain, Class II Permissive Change would have to be filed with the FCC for the entire portfolio of T-Mobile’s branded devices and then T-Mobile would have to perform a large-scale software update push, which could be a massive undertaking, and frankly isn’t necessary.

As we described back in August integrating a portion of Sprint’s PCS spectrum onto T-Mobile’s network and widening the existing PCS channels to 15 or 20 MHz could further improve user experience and overall network efficiency for the combined network. This could be highly beneficial in markets with a high uptake of Sprint customers where Sprint’s PCS spectrum holdings sit adjacent to T-Mobile’s. Alternatively, PCS spectrum swap with other operators could produce similar results.

Top-Largest-Markets_TMobile-Sprint-03

Over the past few years, Sprint has done a lot of work to densify its network using small cells and distributed antenna systems (DAS). These nodes provide added coverage and capacity, but often operate on their own separate 2.5 GHz frequency block, different from the blocks used on Sprint’s macro sites. Repurposing Sprint’s network of small cells and outdoor DAS (oDAS) and retrofitting for T-Mobile’s LTE/NR at some point in the future could provide additional capacity in urban and suburban areas.

Standalone 5G should improve performance and coverage

Finally, the launch of the Standalone (SA) 5G slated for later this year will offer many core benefits, such as reduced latency, improved speeds and network slicing, to name a few. With the introduction of VoNR (Voice over NR), SA 5G should provide an improved 5G coverage, especially for customers in rural America, as the need for mid-band LTE anchors would be removed.

The existing portfolio of branded devices powered by the 2nd generation Qualcomm Snapdragon X55 5G Modem-RF System has the support for Standalone 5G, but will require a software update. Recent announcements of SA 5G lab and field testing using a commercial OnePlus 8 device indicate that T-Mobile is inching closer to commercializing this service. The upcoming 3rd generation Qualcomm Snapdragon X60 Modem-RF System will fully set the stage for standalone 5G deployments around the world by enabling mmWave-sub6 aggregation, as well as sub-6 carrier aggregation across FDD and TDD bands. This means that the upcoming portfolio of T-Mobile’s 5G devices will be able to combine n71 with n41, n261 with n41, and so on. The Dynamic Spectrum Sharing (DSS) feature, which enables coexistence of LTE and 5G in the same frequency channel at the same time, will likely find its purpose in T-Mobile’s toolbox by enabling the use of n2 and n66 frequency bands.

Two weeks of testing T-Mobile’s 5G in New York City delivered an impressive user experience, and it was just an early taste of a 5G layer cake that could continue to improve. Low-band 5G provides the foundational layer of the cake, the mid-band filling is already thickening and the mmWave at the moment is the cherry on top, with the potential to become a much bigger layer. We’re excited to see how other operators can learn from this approach and improve 5G performance across the globe.

Ookla retains ownership of this article including all of the intellectual property rights, data, content graphs and analysis. This article may not be quoted, reproduced, distributed or published for any commercial purpose without prior consent. Members of the press and others using the findings in this article for non-commercial purposes are welcome to publicly share and link to report information with attribution to Ookla.

About the Author

Milan Milanović

| More Articles

Ookla Research™ Articles

Milan Milanović | November 16, 2020

Rakuten is Leveraging O-RAN to Reimagine Mobile and it Could Change the Way We Use Our Phones Forever

日本語

Japanese companies have often been at the forefront of applying innovative new technology to improve the everyday lives of their customers. For example, the world’s very first 1G phone service was pioneered and launched in Japan in 1979. So it’s no surprise that the most cutting-edge approach to mobile is already underway in Japan as Rakuten embraces the Open Radio Access Network (O-RAN) standard to build a cloud-native mobile network, Rakuten Mobile. This fully virtual mobile network allows Rakuten to reduce the costs associated with building and upgrading a network and also offers Rakuten Mobile’s customers a complete beginning-to-end user experience for their online activities.

Rakuten disrupts a well-established market

EN-Rakuten-Diagram

Prior to Rakuten Mobile’s entry, the Japanese mobile market was dominated by three major operators: NTT Docomo, au by KDDI and Softbank Mobile. All three have been in operation for over a decade and have deployed 5G throughout Japan.

NTT DoCoMo was spun off from Nippon Telegraph and Telephone (NTT) with 2G service in 1991. Over the next decade, the company transitioned to 3G. It currently offers LTE service on a combination of 700 MHz, 800 MHz, 1.5 GHz, 1.8 GHz, 2.1 GHz and 3.5 GHz bands. Recently the company has launched 5G service on 3.6GHz, 4.5GHz and mmWave.

KDDI was created in 2000 as a result of a merger of three companies: KDD, DDI and IDO. The company used CDMA technology for its 2G and 3G services, and currently offers LTE services using 700 MHz, 800 MHz, 1.4 GHz, 2.1 GHz and 3.5 GHz. Recently launched 5G service is rolled out on 3.8 GHz, 4.1 GHz and mmWave.

Softbank Mobile, the third major operator in Japan, entered the Japanese market in 2006 with the purchase of Vodafone Japan. Softbank Mobile’s LTE frequency bands span from 700 MHz to 3.5 GHz, and 5G is rolled out on 3.9 GHz and mmWave.

Rakuten is new to mobile, but this is only one part of a well-established business. Founded in 1997 as an e-commerce company with only six employees and one server, the company now has a global reach with a portfolio of over 70 companies and more than a billion membership and loyalty program customers worldwide. The enterprise now covers a wide variety of industries including banking, travel, online shopping, advertising, global messaging, video delivery, data analysis and now wireless, just to name a few.

What’s extremely interesting about Rakuten as a wireless operator is the company’s culture and decades of data-centric expertise in the cloud space. Offering end-to-end user experience and controlling the entire customers’ online journey certainly brings tremendous value to the company, but it also opens the idea of other tech giants like Amazon, Google and Apple entering the mobile telecom space. With the promise of running fully virtualized network functions, the costs associated with network rollouts and operational expenditure may be significantly reduced, and any new entrants to the telecom space may bring fresh perspectives on running and operating a mobile network.

Rakuten leverages the cloud to break the chains of infrastructure

Rakuten opted for a cloud-native mobile network approach using O-RAN instead of relying entirely on traditional network infrastructure equipment vendors for core and radio access technology. This means Rakuten is decoupling radio access from baseband processing and virtualizing a large amount of network functions. This approach relies on data centers and the cloud, which are Rakuten’s strengths.

In the traditional telco world, infrastructure vendors’ solutions typically provide an end-to-end vertical stack, which includes everything from radio access and baseband processing to the transport and core. This means that when operators lock into a contract with infrastructure giants like Huawei, Nokia and Ericsson, they are locked into a proprietary closed ecosystem of software and hardware.

Once hardware and software are disaggregated, operators will be free to deploy any software-defined functions offered by any company, small or large, following “white-box” O-RAN guidelines. This approach, called “virtualization,” can be run on any off-the-shelf computer server. Virtualization reduces the importance of traditional telco hardware, speeds up rollouts, drives innovation and significantly reduces the operator’s capital expenditures. Virtualization can quickly and easily disrupt the economics of traditional operators, paving the way for fresh opportunities both for new entrants and the industry as a whole. With O-RAN, operators can work with major infrastructure vendors like Nokia, NEC, Fujitsu, Cisco and Intel and smaller companies like Airspan and Altiostar to choose custom solutions.

This also means that multi-vendor interoperability and a multitude of innovations like virtualization and automation will largely depend on a vendor’s nimbleness in the research and development space.

The innovations behind O-RAN

O-RAN architecture is a concept designed to address the future needs of mobile network providers based on openness, interoperability and industry collaboration as a whole. Traditional network deployments rely on 3GPP-compliant hardware, typically designed by a single vendor providing a closed-box software and hardware solution as well as security aspects of the networks. On the other hand, the O-RAN Alliance and its members (which include leading telecom, cloud and chipset suppliers, as well as network operators) have been leading an effort to standardize virtual RAN functions and open interfaces to enable cloud-native deployments and multi-vendor interoperability.

O-RAN opens the door to new solutions

One of the goals of this initiative is to maximize the openness and use of common off-the-shelf hardware while minimizing the use of the proprietary telecom hardware. Creating a standardized multi-vendor virtual RAN environment allows third parties to access what used to be closed vendor RAN data and create a multitude of innovative services and applications.

O-RAN promises open interfaces that enable smaller vendors to introduce their products and services in ways that are tailored to suit operators’ unique needs. This should foster vendor diversity and infrastructure robustness while allowing the operators to minimize security risks by controlling all aspects of the network.

By using commercial off-the-shelf (COTS) hardware with virtualization software in the form of containers and virtual machines (VM), the Radio Access Network basically becomes the first app in the O-RAN app store.

Improving network security and agility

Another major aspect of O-RAN is the unparalleled view it provides into the security apparatus of the network, allowing the operator to have full end-to-end visibility and transparency into the entire stack — and to control the entire supply chain for each and every component. Furthermore, because of the software-based nature of O-RAN, the operator has the ability to add, inspect or replace network functions much faster than with traditional deployments which often require physical visits for upgrades. In this way O-RAN improves the efficacy and agility of the network.

This is potentially the first time in the history of the mobile industry where operators will be given the voice and the ability to custom-build networks. From an engineering perspective, moving hardware-based baseband to the virtual world is challenging, but also fun. With recent advancements in software-defined networks and virtualization, LTE and 5G networks have the potential to behave like Wi-Fi networks from installation through integration. The concept of a telecom app store will continue driving innovation, collaboration and industry support, while both private and government sectors will need to be engaged to drive this ongoing transformation in the industry.

Allowing applications to program the network

During the last decade, progress has been made on building applications that can run on top of the network. O-RAN promises to take this to the next level by allowing applications to actually program and run the network itself, which introduces the importance of artificial intelligence (AI). The use of AI to program a network could dramatically simplify the rollout and operation of 5G networks.

Leveraging automation, virtualization and artificial intelligence could also enable self-driving network operations and reduce operating expense. Future deployments could (ideally) depend on virtualized network elements and a standardized white-box approach, which would foster a multi-vendor, interoperable and autonomous RAN.

One of the main propositions of O-RAN architecture is to extend the software-defined network (SDN) concept of decoupling the control-plane (CP) from the user-plane (UP) into RAN, while bringing in embedded intelligence. Separating UP from CP allows for more scalability and cost-effectiveness on the UP. According to the O-RAN Alliance White Paper, this new architecture introduces a set of key interfaces between multiple decoupled RAN components. These include enhanced 3GPP interfaces (F1, W1, E1, X2, Xn) to enable multi-vendor interoperability. In addition to the proposed white-box hardware, many software components will be delivered on an open-source basis like the RAN intelligent controller, protocol stack, PHY layer processing and virtualization platform.

How Rakuten made the world’s first Open Radio Access Network

As a brand new entrant in the Japanese mobile space, Rakuten took its cloud-centric approach and virtualized everything from radio to the core with hundreds of virtual functions implemented. These functions are delivered from multiple computing centers leveraging COTS hardware.

Rakuten has also fostered innovation on the radio access side by having infrastructure giants like Nokia agree to open up the radio, which is typically a black box. This allows Rakuten to control all the hardware that goes into the network, the supply chain, the ecosystem as well as the security aspect of this platform. Rakuten prides itself in the “zero touch provisioning” which breaks free from traditional network rollouts, drastically speeds up the deployment process and requirements, and redefines the way telecom operators can architect, provision and deploy the network.

A company like Rakuten — with decades of experience running server-centric IP-based services — already has an advantage over traditional telcos because of the hundreds of edge data centers they operate. At this time, Rakuten has an infrastructure of nearly 300 edge data centers that can bring content geographically closer to the user and drastically reduce latency. The CP workloads are handled by central data servers, and the entire operation is managed by the horizontal cloud. Everything from radio access, transport and network functions are virtualized and orchestrated by the Rakuten Mobile operating system.
Mobile-Architecture_1120_en-1

The process of activating and provisioning a new site takes minutes, rather than hours or days, bringing significant savings in capital and operational expenditures. For the field technicians this means all they have to do is hang radio units, connect the fiber and power, and the new cell site will be up and running.

Rakuten began initial trials late last year. Using an app-based approach, they onboarded 5,000 customers to test and validate the world’s first end-to-end cloud-native architecture. The official April 2020 launch was on LTE only, and the initial network service area covers Tokyo, Nagoya City and Osaka City. Outside of the native coverage area, users will have the ability to roam on KDDI and Okinawa Cellular. Since the operator owns the end-to-end orchestration and automation of eNodeB as a virtual network function, and both hardware and software is 5G-ready from the get-go. This also means that new network feature rollouts and fixes are significantly easier than they are for a traditional telco. On September 30, Rakuten rolled out a commercial 5G network which will be the test of how easy this flexible software-based architecture makes rolling out core and network functions.

Rakuten Mobile is performing well so far

As revolutionary as Rakuten’s approach is, end users are more likely to care about performance than how the network is configured. We took a look at Rakuten’s performance over 4G LTE and 5G in Tokyo during Q2-Q3 2020. Data for 5G is limited to Q3 as Rakuten’s 5G is only newly commercially available.

Median Speeds for Rakuten Mobile in Tokyo

Speedtest Intelligence^® | Q2-Q3 2020

	Download (Mbps)	Upload (Mbps)
Q2 LTE	38.05	18.28
Q3 LTE	31.68	19.51
Q3 5G	101.33	18.78

Rakuten delivered a median LTE download speed of 38.05 Mbps in Q2 2020. However, LTE download speed was 16.7% lower in Q3 2020. Upload speed over LTE increased slightly. Users with access to the 5G network layer that launched at the end of Q3 2020 experienced a median download speed of 101.33 Mbps, 166.3% faster than the median download speed over LTE in Q2 and 219.9% faster than median download speed over LTE in Q3. The median upload speed on 5G was slower than on LTE.

What operators can learn from the Rakuten model

The Rakuten model is a very early dive into the world of O-RAN architecture. It went from a proof-of-concept to a fully fledged commercial mobile network serving millions of customers in what felt like the blink of an eye. There are still many unknowns, in particular in terms of future scalability and security, but the operator seems to be very confident with its own Rakuten Mobile Platform (RMP). In fact, that same platform is being offered as a turn-key solution to existing mobile operators and new entrants.

This model promises a significant cost reduction in both capital and operating expense, including head count (due to the data-centric nature and the level of automation). This approach seems to pose a threat to the traditional telecom giants like Ericsson, Nokia and Huawei in terms of revenue. However, these companies’ willingness to open their radio components and allow customization for Rakuten’s specific needs implies that broader change is coming in the telecom world.
dish_spectrum_map_en-01

An operator that could benefit from the extension of this experiment is DISH, a recent entrant to the U.S. mobile market. Over the past few decades, DISH has acquired significant amounts of spectrum that spans from 600 MHz all the way up to the mmWave, including the potential acquisition of Sprint’s 800 MHz spectrum assets. Similar to Rakuten Mobile, DISH could potentially combine fallow spectrum and the promise of building and deploying a facilities-based O-RAN 5G network. Deploying a lean network could, at least in theory, enable a rapid deployment of DISH’s assets, which could create many new jobs while leveraging American companies that are eager to innovate and disrupt, such as Cisco, Altiostar, Mavenir, Qualcomm, Intel and Airspan (to name a few).

A recent announcement reveals VMWare as a strategic partner in providing a cloud-based abstraction layer, which is essentially a 5G operating system that will hold the silicon, software and cloud together, and will enable a hyper-scale of public cloud capacity where needed.

O-RAN would allow DISH and other mobile operators to expand beyond the constraints of vertical solutions provided by traditional telco vendors like Ericsson, Huawei and Nokia. The O-RAN concept would simplify and fully automate the network deployment using DISH’s greenfield environment. Nokia has been the most receptive to the O-RAN approach, agreeing to provide 5G core applications including subscriber management, device management and integration services (among others). DISH has recently signed a deal with Japanese giant Fujitsu, securing a large order of O-RAN-compliant radios while Altiostar and Mavenir will also be providing O-RAN software.

While all eyes will be immediately focused on Rakuten Mobile and its cloud-native O-RAN 5G network, operators and infrastructure vendors around the world are working together to improve the O-RAN concept. From a financial and operational standpoint, the O-RAN model makes a lot of sense, and a few operators are already looking into leveraging at least some tools from the O-RAN toolbox. India’s Reliance Jio has plans for a deployment similar to Rakuten’s, while U.S. telcos Verizon and AT&T are already making steps to allow for multi-vendor 5G interoperability in some markets. Recently, Telefónica Spain and STC in Saudi Arabia have shown interest in Rakuten’s mobile platform.

We are likely to see a lot more O-RAN announcements in the coming year, and we at Ookla will continue monitoring Rakuten’s network performance, so stay tuned.

楽天がO-RANの活用でモバイルを一新、携帯電話の今後のありかたをまったく変える可能性を示す

日本の企業はこれまでに何度も、顧客の生活を豊かにする革新的な新技術導入の最前線にいました。たとえば、世界初の1Gの通話サービスは1979年に日本で初めて始まりました。そのため、楽天がOpen Radio Access Network（O-RAN）標準を採用し、クラウドネイティブなモバイルネットワークとして楽天モバイルを構築することで、携帯通信への最先端のアプローチがすでに日本で始まっていることも驚きではありません。楽天は、この完全に仮想的なモバイルネットワークにより、ネットワークの構築とアップグレードにかかるコストを削減し、さらに、楽天モバイルの顧客にオンラインでのアクティビティにおける一貫したユーザーエクスペリエンスを提供できるようになりました。

楽天が確立された市場を打破

JP-Rakuten-Diagram

楽天モバイルの参入以前、日本の携帯通信市場はNTT Docomo、au by KDDI、Softbankモバイルの3社の主要な事業者によって支配されていました。これら3社すべては10年以上運営されており、日本全国で5Gを展開しています。

NTT DoCoMoは日本電信電話（NTT）から派生し、1991年に2Gサービスを開始しました。その後10年間で、同社は3Gに移行しています。現在は、700MHz、800MHz、1.5GHz、1.8GHz、2.1GHz、3.5GHzのバンドでLTEサービスを提供しています。現在、同社は3.6GHz、4.5GHz、ミリ波で5Gサービスを開始しています。

KDDIは、KDD、DDI、IDOの3社が合併して2000年に設立されました。同社は2Gと3GサービスにCDMA技術を使用し、現在では700MHz、800MHz、1.4GHz、2.1GHz、3.5GHzを使用してLTEサービスを提供しています。最近開始された5Gサービスは、3.8GHz、4.1GHz、ミリ波で展開されています。

日本の3つめのキャリアであるSoftbankモバイルは、Vodafone Japanを買収して2006年に日本市場に参入しました。SoftbankモバイルのLTE周波数バンドは700MHzから3.5GHzで、5Gは3.9GHzとミリ波で展開されています。

楽天は携帯通信事業に初参入ですが、同事業は確立されたビジネスの一部となっています。1997年にわずか6名の従業員と1台のサーバーでEコマース企業として設立された同社は、現在では70社以上の企業のポートフォリオと、10億人以上の会員とポイントプログラムの顧客を世界中に抱え、グローバルに展開しています。同社は現在幅広い業界に展開しており、いくつか例を挙げると、銀行、旅行、オンラインショッピング、広告、グローバルメッセージ機能、動画配信、データ解析、そして携帯通信などです。

楽天が携帯通信事業者として非常に興味深い点は、同社の文化とクラウド事業における数十年分のデータ中心のノウハウです。エンドツーエンドのユーザーエクスペリエンスの提供と顧客のオンラインジャーニー全体の制御は、確かに多大な価値をもたらしていますが、同時に、Amazon、Google、Appleなどの情報技術系最大手企業が携帯通信事業に参入したことを連想させます。完全に仮想化されたネットワーク機能を実行するという約束のもと、ネットワークの展開と運用費用に関連するコストが大幅に削減され、新規通信事業者が増えることでモバイルネットワークの実行と運用に新しい展望をもたらすでしょう。

楽天はクラウドを活用してインフラストラクチャの鎖を解く

楽天は、核となる技術と無線アクセス技術について従来のネットワークインフラストラクチャの設備ベンダーに完全に依存するのではなく、O-RANを使用してクラウドネイティブなモバイルネットワークへのアプローチを採用しました。これは、楽天が無線アクセスをベースバンド処理から分離させ、多数のネットワーク機能を仮想化していることを意味します。このアプローチでは、楽天の強みであるデータセンターとクラウドを基盤としています。

従来の携帯通信業界では、インフラストラクチャベンダーのソリューションがエンドツーエンドの垂直スタックを提供することが一般的でした。これには、無線アクセスとベースバンド処理から、転送とコアに至るまでのすべてが含まれています。これは、事業者がHuawei、Nokia、Ericssonなどのインフラストラクチャの大手企業との契約に縛られると、ソフトウェアとハードウェアの私的に閉じられたエコシステムに閉じ込められることを意味します。

ハードウェアとソフトウェアが分離されると、事業者は大小あらゆる企業によって提供される、あらゆるソフトウェア定義機能を自由に展開できます。これは「ホワイトボックス化」されたO-RANのガイドラインに従っています。「仮想化」と呼ばれるこのアプローチは、あらゆる既成のコンピューターサーバー上で実行可能です。仮想化は従来の携帯通信ハードウェアの重要性を低減し、展開を高速化し、イノベーションを促進して事業者の資本支出を大幅に減らします。仮想化は従来の事業者の経済を一瞬で簡単に破壊させ得るため、新たな参入企業と業界全体の両方で新鮮な機会を創出しています。O-RANを使用することで、事業者は、Nokia、NEC、富士通、Cisco、Intelなどの大手インフラストラクチャベンダー、およびAirspanやAltiostarなどの小規模な企業と提携し、カスタムのソリューションを選択できます。

これは同時に、マルチベンダーの相互運用性と、仮想化や自動化などのイノベーションの数が、研究と開発分野におけるベンダーの俊敏性に大きく依存することも意味します。

O-RANの背景にあるイノベーション

O-RANアーキテクチャは、モバイルネットワークプロバイダーの将来のニーズに対応するために設計されたコンセプトで、オープン性、相互運用性、業界全体の協業性に基づいています。従来のネットワーク展開は3GPP準拠のハードウェアに依存しており、これは通常、クローズボックスのソフトウェアとハードウェアソリューション、およびネットワークのセキュリティ面を提供する単一のベンダーにより設計されています。一方で、O-RANアライアンスとそのメンバー（大手通信企業、クラウドおよびチップセットのサプライヤー、携帯通信事業者を含む）は、仮想RAN機能とオープンインフラストラクチャを標準化し、クラウドネイティブな展開とマルチベンダーの相互運用性を実現するための取り組みを進めています。

新たなソリューションへの扉を開くO-RAN

このイニシアチブの目標の1つに、私的な携帯通信ハードウェアの使用を最小限にしながら、オープン性と一般的な既成ハードウェアの使用を最大限にすることが挙げられます。標準化されたマルチベンダーの仮想RAN環境を作り上げることにより、サードパーティがかつてはクローズドだったベンダーのRANデータにアクセスし、多くの革新的なサービスとアプリケーションを作成できるようになります。

O-RANは、小規模なベンダーが、事業者固有のニーズに合うようにカスタマイズされた、自社の製品とサービスを導入できるようにするオープンなインターフェイスを確約します。これにより、事業者がネットワークのすべての側面を制御することでセキュリティのリスクを最小限に抑えながら、ベンダーの多様性を広げ、インフラストラクチャの堅牢性を強化することにつながります。

商用の既成（COTS）ハードウェアを、コンテナと仮想マシン（VM）の形で仮想化ソフトウェアとともに使用することで、基本的に無線アクセスネットワークがO-RANのアプリストア内の最初のアプリとなります。

ネットワークのセキュリティと俊敏性を向上

O-RANの他の大きな側面として、ネットワークのセキュリティ組織にもたらされる他にはない視点があります。これにより、事業者はスタック全体に完全なエンドツーエンドの視認性と透明性を手に入れることができます。そして、各コンポーネントすべてに対するサプライチェーン全体を管理できます。さらに、O-RANのソフトウェアベースの性質によって、アップグレードのために物理的に訪れる必要があった従来の展開よりもはるかに高速にネットワーク機能を追加、点検、交換することができます。このようにして、O-RANはネットワークの有効性と俊敏性を高めることができます。

事業者がカスタムに構築されたネットワークを表明し実行できるのは、携帯通信業界においてこれが初めてかもしれません。エンジニアリングの観点からは、ハードウェアベースのベースバンドを仮想世界に移行するのは困難ですが、同時に楽しさもあります。最近のソフトウェア定義ネットワークと仮想化の進歩により、LTEおよび5Gネットワークは、設置から統合までWi-Fiネットワークのように機能する可能性を秘めています。携帯通信アプリストアのコンセプトはこれからもイノベーション、コラボレーション、業界のサポートを推進しつづけます。一方で一般部門と政府部門の両方が業界におけるこの継続的な転換の促進に従事する必要があります。

アプリケーションによるネットワークのプログラミングを実現

ここ10年間で、ネットワークの上層で実行可能なアプリケーションの構築は進歩しています。O-RANは、アプリケーションが実際にネットワークそのものをプログラミングおよび実行できるようにすることで、これを次のレベルへと高めます。これは、人工知能（AI）の重要性につながります。ネットワークのプログラミングにAIを使用することで、5Gネットワークの展開と運用が劇的に簡素化できます。

自動化、仮想化、人工知能の活用により、自己駆動形のネットワーク運用が実現し、運用コストを削減できます。将来の展開は（理想的には）、仮想化ネットワーク要素と標準化されたホワイトボックスのアプローチに基づく可能性があり、これにより、マルチベンダー化、相互運用性、自律型RANが推進されます。

O-RANアーキテクチャの主な課題の1つに、埋め込みインテリジェンスをもたらしながら、同時に制御プレーン（CP）をユーザープレーン（UP）からRANに分離させるという、ソフトウェア定義ネットワーク（SDN）の概念を拡大することが挙げられます。UPをCPから分離させることにより、UPでのスケーラビリティとコスト効率性が高まります。O-RANアライアンスのホワイトペーパーによると、この新たなアーキテクチャによって、複数の分離されたRANコンポーネント間の一連の主なインターフェイスが導入されます。これには、マルチベンダーの相互運用性を実現するための3GPPインターフェイス（F1、W1、E1、X2、Xn）が含まれます。提唱されたホワイトボックスのハードウェアに加えて、多くのソフトウェアコンポーネントは、RANインテリジェントコントローラー、プロトコルスタック、PHYレイヤー処理、仮想化プラットフォームのように、オープンソースベースで提供されます。

楽天はいかにして世界初のオープン無線アクセスネットワークを作り上げたか

日本の携帯通信事業への新たな参入企業として、楽天はそのクラウド中心のアプローチを採用し、数百の仮想化機能を実装して無線からコアにいたるまでのすべてを仮想化しました。これらの機能は、COTSハードウェアを活用した複数のコンピューティングセンターから提供されます。

また楽天は、Nokiaのようなインフラストラクチャ大手企業の周波数を利用可能にすることについて合意を得ることで、無線アクセスにおける革新を促進しました。これは、通常はブラックボックスとなっています。これにより、楽天はネットワーク、サプライチェーン、エコシステム、そしてこのプラットフォームのセキュリティ面に使用されるすべてのハードウェアを制御できます。楽天は、従来のネットワーク展開から解放し、展開プロセスと要件を劇的に高速化させ、携帯通信事業者がネットワークを設計、プロビジョニング、展開する方法を再定義する「ゼロタッチプロビジョニング」に自信を持っています。

サーバー中心のIPベースのサービスを運用してきた数十年の経験を持つ楽天のような企業は、同社が運用する数百のエッジデータセンターにより、従来の携帯通信よりもすでに優位に立っています。現時点で、楽天は300か所近いエッジデータセンターのインフラストラクチャを有しており、地理的にユーザーに近い場所にコンテンツを配信することで、遅延を劇的に低減しています。CPのワークロードは中央データサーバーによって処理され、運用全体は水平的なクラウドによって管理されます。無線アクセス、転送、ネットワーク機能のすべては仮想化され、楽天モバイルのオペレーティングシステムによってオーケストレーションされます。
Mobile-Architecture_1120_jp

新たなサイトの有効化とプロビジョニングのプロセスは、数時間も数日もかかることなく、数分で完了し、資本コストおよび運用コストを大幅に削減できます。現場技術者にとってこれは、無線ユニットを吊るし、ケーブルと電源を接続するだけで、新たなセルサイトを稼動できることを意味します。

楽天は初期テストを昨年後半に開始しています。アプリベースのアプローチを使用することで、5,000人の顧客を招待して世界初のエンドツーエンドのクラウドネイティブなアーキテクチャをテストおよび検証しました。2020年4月に公式に運用開始したのはLTEのみで、当初のネットワークサービスでは東京、名古屋市、大阪市のエリアをカバーしています。ネイティブにカバーされているエリアの外では、ユーザーはKDDIおよび沖縄セルラーからローミングできます。事業者は、eNodeBのエンドツーエンドのオーケストレーションと自動化を仮想ネットワーク機能として所有しており、ハードウェアとソフトウェアの両方が最初から5Gに対応しています。これは、新たなネットワーク機能の展開と修正が従来の携帯通信よりも大幅に簡単であることも意味します。9月30日に、楽天は商用の5Gネットワークを展開しました。これは、この柔軟なソフトウェアベースのアーキテクチャによって、コアとネットワーク機能の展開がどれくらい簡単になるかのテストです。

楽天モバイルのパフォーマンスは順調

楽天のアプローチは革命的ですが、エンドユーザーはネットワーク構成の仕組みよりもパフォーマンスを気にします。2020年第2四半期から第3四半期の、東京における4G LTEと5Gの楽天のパフォーマンスを確認してみました。楽天の5Gは商用的に開始されたばかりのため、5Gのデータは第3四半期のみとなっています。

東京における楽天モバイルの中央速度

Speedtest Intelligence^® | 2020年第2四半期から第3四半期

	下り（Mbps）	上り（Mbps）
第2四半期LTE	38.05	18.28
第3四半期LTE	31.68	19.51
第3四半期5G	101.33	18.78

楽天は、2020年第2四半期のLTEの中央速度が下り38.05Mbpsでした。しかし、2020年第3四半期にはLTEの下り速度は16.7%落ちています。LTEでの上り速度はわずかに上昇しています。2020年第3四半期の終わりに開始された、5Gネットワークレイヤーにアクセスできるユーザーは、101.33Mbpsの下り中央速度で利用しています。これは、第2四半期のLTEでの下り中央速度よりも166.3%速く、さらに第3四半期の下り中央速度よりも219.9%速くなっています。5Gの上り中央速度はLTEよりも低速でした。

事業者が楽天モデルから学べること

楽天モデルでは、O-RANアーキテクチャの世界へかなり早い段階で踏み入りました。そこから概念実証が始まり、あっという間に数百万人のユーザーにサービスを提供する、完全に実用的な商用モバイルネットワークになりました。特に将来のスケーラビリティとセキュリティに関しては、まだ数多くの未知の部分があるものの、楽天は独自の楽天モバイルプラットフォーム（RMP）に非常に信頼を置いているように見えます。実際、この同じプラットフォームが既存の携帯通信事業者と新たな参入企業に対して、ターンキーソリューションとして提供されています。

このモデルでは、社員数を含む、資本コストと運用コストの両方の大幅な削減が確約されています（データ中心の性質と自動化のレベルによるものです）。このアプローチは、Ericsson、Nokia、Huaweiのような従来の携帯通信大手企業にとって、収益の観点からは脅威となる可能性を秘めているでしょう。ただし、これらの企業が無線コンポーネントを開放し、楽天固有のニーズに対するカスタマイズを快く許可したことは、携帯通信業界において大きな変化が訪れることを示唆しています。
dish_spectrum_map_jp-01

この試みが広がることでメリットを享受できる可能性のある事業者にDISHが挙げられます。同社は、米国の携帯通信市場に最近参入しました。過去数十年間で、DISHは600MHzからミリ波にいたるまでの十分な周波数を獲得しています。また、Sprintの800MHz周波数アセットも獲得予定です。楽天モバイルと同様に、DISHでは、使用されていない周波数と、施設ベースのO-RAN 5Gネットワークを構築および展開するという契約を組みわせることが検討されています。無駄のないネットワークの展開によって、少なくとも理論的には、DISHのアセットの迅速な展開が可能になります。これにより、Cisco、Altiostar、Mavenir、Qualcomm、Intel、Airspanなど（ごく一部の例ですが）の、革新と打破に意欲的な米国企業を活用しながら、多くの業務が新たに創出される可能性があります。

最近の発表では、VMwareが、シリコン、ソフトウェア、クラウドをまとめて有する本質的な5Gオペレーティングシステムであり、必要な場合にパブリッククラウドのキャパシティを拡大できるようにする、クラウドベースの抽象化レイヤーの提供における戦略的パートナーとなっています。

O-RANにより、DISHなどの携帯通信事業者が、Ericsson、Huawei、Nokiaなどの従来の携帯通信ベンダーにより提供されている垂直的なソリューションの制約を超えて拡張することができるようになります。O-RANの概念は、DISHの未開発環境を使用したネットワーク展開を簡素化し、完全に自動化するものになるでしょう。NokiaはO-RANアプローチを最も受け入れている企業であり、とりわけ登録者管理、デバイス管理、統合サービスなどの5Gコアアプリケーションの提供に合意しています。DISHは最近、O-RANに準拠した大量の無線を有している日本の大手企業である富士通との契約に署名しています。一方、AltiostarとMavenirもO-RANソフトウェアを提供予定です。

楽天モバイルとそのクラウドネイティブなO-RAN 5Gネットワークに多くの注目が集まる一方で、世界中の事業者とインフラストラクチャベンダーが協力してO-RANの概念を向上させています。経済的および運用的な観点からは、O-RANモデルは大変理にかなっており、小数の事業者はすでにO-RANツールボックスの少なくとも一部のツールを活用することを検討しています。インドのReliance Jioでは楽天に似た展開を計画しており、米国の携帯通信事業者であるVerizonとAT&Tではすでに一部の市場でマルチベンダーの5G相互運用性を許可するための手順を踏んでいます。最近では、サウジアラビアのTelefónica SpainおよびSTCが楽天のモバイルプラットフォームに興味を示しています。

来年にはさらに多くのO-RAN関連の発表があると思われます。Ooklaでは楽天のネットワークのパフォーマンスを引き続き注視していく予定です。

About the Author

Milan Milanović

| More Articles

Ookla Research™ Articles

Maria Gómez | December 4, 2019

Borrowed Networks: National Roaming in Mexico

Leer en español

Increased competition and new roaming agreements have significantly changed the telecommunications landscape in Mexico in the past decade such that subscribers of some operators often have cellular connectivity even if their own provider does not have coverage in a certain area. We considered Speedtest® data from Android devices during Q3 2019 to investigate where subscribers experience roaming, how much time they are spending roaming on 4G and how roaming affects mobile speeds.

Roaming is common in Mexico, but differs widely by operator and location

Ookla_Mexico_ATT_On-network_Roaming_map_1219_en
Ookla_Mexico_Movistar_On-network_Roaming_map_1219_en
Ookla_Mexico_Telcel_On-network_Roaming_map_1219_en

AT&T, Movistar and Telcel all demonstrated a majority of on-network activity in major cities like Mexico City, Guadalajara, Mérida, Monterrey and Tijuana. However, Telcel showed substantially more on-network activity in areas beyond large cities than competitors, including up and down Mexico’s western coast from Tapachula in Chiapas to Guaymas in Sonora. Both AT&T and Movistar showed more roaming in this same area.

It’s interesting to note that on-network and roaming activity in Mexico often occur within close proximity, even in major cities. For example, the map shows most of AT&T’s on-network activity concentrated in central Mexico. The same was true of AT&T’s roaming activity. We have considered AT&T and subsidiary Unefon jointly throughout this article.

A similar overlap can be seen in Movistar’s on-network and roaming activity, although Movistar demonstrated roaming in fewer locations than we saw on AT&T. Telcel showed virtually no roaming.

It will be interesting to see how the recent deal between AT&T and Telefonica (Movistar’s parent company) affects the roaming experience for subscribers.

AT&T shows the most Time Spent on 4G on and off network

Time-Spent-On-Net-Roam-in-Mexico-ENG-2

We used Speedtest data on Time Spent to measure how often, on average, consumers were able to connect to 4G LTE on their own network and while roaming in Mexico during Q3 2019. Reported 4G roaming includes “emergency calls only” modes (which do not have normal voice or data service).

Mobile users in Mexico were able to connect on-network to 4G LTE 69.9% of the time on average throughout the country. Separating results by provider, AT&T had the highest Time Spent on-network on 4G at 74.4%. Telcel followed with 70.4% and Movistar was third with 58.1%.

Time Spent roaming on 4G in Mexico was a fraction of what we saw looking at Time Spent on 4G while on-network with users spending an average of 0.7% of the time roaming on 4G during Q3 2019. At the operator level, AT&T subscribers showed the most Time Spent roaming on 4G at 1.4%. Movistar followed with subscribers spending 0.7% of the time roaming on 4G. Telcel subscribers had the least Time Spent roaming on 4G at 0.3%.

Telcel was the fastest operator for on network use

On-net-Roaming-Mobile-Speeds-in-Mexico-ENG-1

At 24.18 Mbps, mean download speed on-network in Mexico was 173.5% faster than mean download speed while roaming during Q3 2019. We use Speed Score™ to measure operator performance as it incorporates measures of download and upload speed. Telcel was the fastest operator with an on-network Speed Score of 28.80. Movistar showed the second fastest on-network performance with a Speed Score of 17.15. AT&T ranked third for on-network performance with a Speed Score of 9.87, 65.7% slower than Telcel.

AT&T had the fastest Speed Score while roaming with a Speed Score of 5.18, 47.5% slower than their on-network measure. Movistar had the second fastest Speed Score while roaming at 3.02, 82.4% less than on-network. Telcel showed virtually no roaming.

We will continue to analyze emerging trends and their effects on internet performance as telecommunications in Mexico continue to evolve. To find out more about internet speeds in Mexico or other parts of the world, contact us.

Editor’s note: The maps and speed data in this article have been revised to further reflect the experience of consumers using a single SIM.

Redes prestadas: Itinerancia nacional en México

El aumento de la competencia y los nuevos acuerdos de itinerancia han cambiado significativamente el panorama de las telecomunicaciones en México en la última década, de tal modo que suscriptores de algunos proveedores frequentemente tienen conectividad celular incluso si su propio proveedor no contaba con cobertura en un área determinada. Tomamos en cuenta datos de Speedtest® de dispositivos Android durante el tercer trimestre de 2019 para investigar dónde experimentan itinerancia los suscriptores, durante cuánto tiempo usan la itinerancia en 4G y de qué manera la itinerancia afecta las velocidades móviles.

La itinerancia es común en México, pero difiere ampliamente por operador y ubicación

Ookla_Mexico_ATT_On-network_Roaming_map_1219_es
Ookla_Mexico_Movistar_On-network_Roaming_map_1219_es
Ookla_Mexico_Telcel_On-network_Roaming_map_1219_es

AT&T, Movistar y Telcel demostraron una mayoría de actividad en la red en grandes ciudades como ciudad de México, Guadalajara, Mérida, Monterrey y Tijuana. Sin embargo, Telcel exhibió una actividad en la red en áreas alejadas de las grandes ciudades sustancialmente mayor que la competencia, incluida la costa oeste de México, desde Tapachula en Chiapas hasta Guaymas en Sonora. Tanto AT&T como Movistar exhibieron mayor itinerancia en la misma área.

Es interesante destacar que, a menudo, la actividad en la red y de la itinerancia en México ocurren a escasa distancia entre sí, incluso en las grandes ciudades. Por ejemplo, el mapa muestra que la mayoría de la actividad en la red de AT&T se concentró en el centro de México. Lo mismo sucedió con la actividad de la itinerancia de AT&T. A lo largo de este artículo consideramos de manera conjunta a AT&T y el subsidiario Unefon.

Puede verse una superposición similar en la actividad en la red y de la itinerancia de Movistar, aunque esta empresa exhibió itinerancia en menos lugares que AT&T. Telcel prácticamente no mostró itinerancia.

Será interesante ver de qué manera el acuerdo reciente entre AT&T y Telefónica (sociedad matriz de Movistar) afecta la experiencia de itinerancia para los suscriptores.

AT&T exhibe la mayor cantidad de tiempo de permanencia en 4G dentro y fuera de la red

Time-Spent-On-Net-Roam-in-Mexico-SP-2

Empleamos datos de Speedtest sobre el tiempo de permanencia para mostrar con qué frecuencia, en promedio, los consumidores pudieron conectarse a 4G LTE en su propia red y durante la itinerancia en México en el tercer trimestre de 2019. Los datos de itinerancia incluyen reportes en modo de emergencia, los cuales no implican servicio de voz y datos.

Los usuarios móviles en México pudieron conectarse a la red en 4G LTE el 69,9 % del tiempo en promedio en todo el país. Si separamos los resultados por proveedor, AT&T tuvo la mayor cantidad de tiempo de permanencia en 4G, con el 74,4 %. Le siguió Telcel, con el 70,4 %, y Movistar quedó tercero con el 58,1 %.

El tiempo de permanencia en itinerancia en 4G en México fue una fracción de lo que se vio de tiempo de permanencia en 4G en la red, ya que los usuarios pasaron, en promedio, el 0,7 % del tiempo con itinerancia en 4G durante el tercer trimestre de 2019. En el caso de los operadores, los suscriptores de AT&T exhibieron la mayor cantidad de tiempo de permanencia en 4G, con el 1,4 %. Le siguió Movistar, con el 0,7 % del tiempo de permanencia en itinerancia en 4G de parte de los suscriptores. Los suscriptores de Telcel tuvieron la menor cantidad de tiempo de permanencia en itinerancia en 4G, con el 0,3 %.

Telcel fue el operador más rápido

On-net-Roaming-Mobile-Speeds-in-Mexico-SP-1

Con 24,18 Mbps, la velocidad media de descarga en la red en México fue un 173,5 % más rápida que la velocidad media de descarga en itinerancia durante el tercer trimestre de 2019. Utilizamos Speed Score™ para medir el rendimiento de los operadores, ya que incluye mediciones de velocidades de carga y descarga. Telcel fue el operador más rápido, con una puntuación de velocidad en la red de Speed Score de 28,80. Movistar tuvo el segundo mejor rendimiento de velocidad en la red, con un resultado de 17,15. AT&T se quedó con el tercer lugar, con 9,87, un 65,7 % más lento que Telcel.

AT&T demostró tener la puntuación más alta de Speed Score, con una velocidad de itinerancia de 5,18, un 47,5% más lento que su medición de la red. Movistar quedó segundo con un resultado de Speed Score de 3,02 en cuanto a la velocidad de la itinerancia, 82,4% inferior al rendimiento de la red de la empresa. Telcel prácticamente no mostró itinerancia.

Continuaremos analizando las tendencias emergentes y sus efectos sobre el rendimiento de Internet a medida que las telecomunicaciones en México continúen evolucionando. Para obtener más información sobre las velocidades de Internet en México y otras partes del mundo, comuníquese con nosotros.

Nota del editor: Los mapas y datos de velocidades en este artículo fueron ajustados para reflejar más a fondo la experiencia de los consumidores usando una sola tarjeta SIM. .

About the Author

Maria Gómez

| More Articles

Ookla Research™ Articles

Maria Gómez | January 22, 2020

What Went Down? The Most Significant Online Service Outages in Q4 2019

“Unable to connect” — three words that bring people together in frustration and annoyance around the world. In the last three months of 2019, outages affected users across the globe as they tried to access all kinds of different industries, from gaming to streaming services. Downdetector® data from Q4 2019 provides unique insight into outages including the dates, duration and location and number of reports submitted by users. We’ve divided the outages into five categories: Gaming, Mobile Operators, Social Media Sites, Streaming Services and Financial Institutions.

Gaming

Discord (December 7, 2019): 15,976 outage reports at peak

The largest gaming outage we saw during Q4 2019, Discord, was apparently “due to an issue with Google compute platform.” Starting on December 7, 2019, outage reports were still trickling into Downdetector on December 8. The highest volume of reports happened over 5.5 hours and with a peak of 15,976 reports in the United States.

Downdetector_Outage_Discord

Fortnite (October 13, 2019): 11,326 outage reports at peak

“The End”, an event where Fortnite was preparing for a big release, coincided with the largest Fortnite outage we saw during Q4 2019. On October 13, 2019, Fortnite users flooded to Downdetector to report an outage that started around 10:30 a.m. PST and lasted for six and a half hours. At the peak of the outage there were 11,326 reports, predominantly from users in the United States.

Mobile Operators

Vodafone Germany (October 23, 2019): 21,065 outage reports at peak

The multinational mobile service provider was down for about four hours for thousands of users on October 23, 2019. At the peak of the outage, 21,065 users, primarily in Germany, reported having problems with their service.

Downdetector_Outage_Vodafone-DE

Social Media Sites

Facebook (November 28, 2019): 12,726 outage reports at peak

Instagram (November 28, 2019): 21,682 outage reports at peak

The Facebook family of social media sites outage on November 28, 2019 was one of many social outages in Q4 2019. The outage lasted about five hours and affected users in the U.S., Germany, Italy and Spain. A combined 34,408 Facebook and Instagram users reported outages at the peak of the outage. Facebook’s Messenger app experienced a smaller outage on November 18, 2019 with 8,952 users reporting outages at the peak.

Downdetector_Outage_Facebook_Instagram

Snapchat (October 14, 2019): 18,252 outage reports at peak

This popular multimedia messaging app was down for five and a half hours on October 14, 2019, leaving users unable to chat, send or receive photos from their friends. At the peak of the outage, 18,252 users from the U.S. reported problems. The outage was so significant, the hashtag #SnapchatDown was trending on Twitter at the time of the outage.

Twitter (October 22, 2019): 15,952 outage reports at peak

Where do Twitter users go to complain when Twitter is down? Our data shows Twitter users in the east coast of the U.S. rushed to Downdetector to report problems with the social media platform on October 22, 2019. Even though the outage lasted only about a half an hour, 15,952 users reported that they were unable to tweet, retweet, like tweets or access their account at the peak of the outage. Problems were also reported in parts of Europe and Latin America.

Streaming Services

Hulu (December 19, 2019): 25,777 outage reports at peak

Hulu users reported problems with the streaming service on the morning of December 19, 2019. At the peak, 25,777 users in the United States were reporting that the streaming service was down. Seven hours after the outage began, most users had regained access to their accounts and were able to watch their favorite shows and movies. In contrast, Netflix, Hulu’s main competitor, had a small outage in Germany on November 13, 2019. At its peak, only 3,197 people were reported problems with the service.

Downdetector_Outage_Hulu

Disney+ (November 12, 2019): 8,441 outage reports at peak

After months of hyped advertising and special offers, Disney+ users were eager to begin streaming their favorite movies and shows on the new service on November 12, 2019. However, problems began early in the morning on launch day in the U.S. and continued throughout the day. At the peak of the outage, 8,441 users reported that they couldn’t access the platform through the app or stream any of the content Disney+ was offering.

Financial Institutions

ING (November 17,2019): 1,342 outage reports at peak

Banks take the security and uptimes of their websites very seriously, so we expect to see fewer outages and for the outages that do occur to be small and short. This held true when Dutch users of banking giant ING reported problems with that website on November 17, 2019. The outage lasted about an hour and a half and there were 1,342 reports at the peak. No other countries reported problems with the bank’s mobile banking that day.

Downdetector_Outage_ING

These were the most significant outages we saw in Q4 2019 among the more than 5,500 sites that Downdetector monitors across the internet. Do outages impact your customers’ experience with your services? To learn more about how data from Downdetector can help your network operations center and customer care team detect and resolve issues faster, contact us here.

About the Author

Maria Gómez

| More Articles

Ookla Research™ Articles

Maria Gómez | February 12, 2020

How Roaming Affects Mobile Speeds in Europe

Last year, we conducted an expansive analysis of mobile roaming in the European Union. We’re updating our data to see which countries have the best roaming speeds and how roaming affects time spent on Wi-Fi. This analysis is based on Speedtest data from Android devices in Europe during Q3-Q4 2019. Despite Brexit, we’ve included data from the United Kingdom to keep parity year-over-year, as the roaming agreement remains in place for 2020 and this data involves travel within a specific set of countries.

Roaming in Europe typically means slower download speeds

Mobile roaming speeds are affected by the deals struck between individual mobile operators on a country-by-country basis. This means roaming speeds can depend on the plan a subscriber has selected in their home country and on what is offered in the country of destination.

Out of the 28 countries we examined, residents of 22 countries experienced slower speeds when roaming elsewhere in Europe than they do in their home countries. Travelers from the Netherlands experience the most significant loss in speed, with a roaming download speed that is 54.7% slower than at home. However, users in Slovakia barely notice a change in download speed with a roaming download speed only 2.6% slower than the mean download speed in Slovakia during this period.

Roaming vs. Local Mobile Download Speeds in Europe

Speedtest^® Data | Q3-Q4 2019

Country	Local Speeds (Mbps)	Roaming Speed Upload (Mbps)	% Change
Austria	46.30	35.93	-22.4%
Belgium	50.95	33.38	-34.5%
Bulgaria	55.32	30.95	-44.1%
Croatia	52.40	43.73	-16.6%
Cyprus	43.50	21.04	-51.6%
Czech Republic	47.00	36.37	-22.6%
Denmark	48.38	32.28	-33.3%
Estonia	42.61	43.90	3.0%
Finland	43.38	41.46	-4.4%
France	44.34	34.00	-23.3%
Germany	34.07	30.48	-10.5%
Greece	39.90	45.03	12.9%
Hungary	45.53	29.55	-35.1%
Ireland	26.73	32.05	19.9%
Italy	33.64	40.98	21.8%
Latvia	33.20	38.54	16.1%
Lithuania	46.72	39.73	-15.0%
Luxembourg	52.55	34.61	-34.1%
Malta	48.88	32.30	-33.9%
Netherlands	62.52	28.31	-54.7%
Poland	33.07	25.57	-22.7%
Portugal	34.49	36.34	5.4%
Romania	37.84	33.69	-11.0%
Slovakia	34.74	33.83	-2.6%
Slovenia	39.34	32.67	-16.9%
Spain	34.28	27.56	-19.6%
Sweden	48.52	36.55	-24.7%
United Kingdom	36.36	34.96	-3.8%

Upload speeds locally are also mostly faster than those experienced while roaming. Cyprus, Denmark, and the Netherlands all had relatively high mean upload speeds locally, but they also had the largest loss in roaming upload speeds ranging from 31.3% to 36.7% slower upload speeds on mobile roaming.

Residents of some European countries do benefit from faster speeds when roaming on mobile. Users from Estonia, Portugal, Greece, Latvia, Ireland, and Italy all experienced faster download speeds when roaming elsewhere in Europe than they did in their home countries during Q3-Q4 2019. Italian and Irish travelers gained the most when roaming with download speeds that were 21.8% and 19.9% faster than their local speeds, respectively.

A total of 12 countries had higher mean upload speeds while roaming than they experienced locally during this period. France, Latvia, and Ireland benefitted the most with an increase in upload speeds that ranged from 26.5% to 21.4% while roaming.

Latency increases dramatically while roaming in Europe

Unlike mean download and upload speeds, latency is almost always dramatically different while roaming. This is because roaming signals are routed through a user’s home network, making latency a significant issue for Europeans when traveling.

Local vs. Roaming Latency in Europe

Speedtest^® Data | Q3-Q4 2019

Country	Local Latency (ms)	Roaming Latency (ms)	% Change
Austria	26	84	220.4%
Belgium	27	83	205.7%
Bulgaria	27	127	368.3%
Croatia	33	89	166.0%
Cyprus	23	213	825.5%
Czech Republic	26	74	187.6%
Denmark	26	100	287.9%
Estonia	25	76	201.5%
Finland	27	97	254.6%
France	41	82	98.8%
Germany	38	87	128.1%
Greece	29	129	349.8%
Hungary	25	93	272.6%
Ireland	35	100	185.0%
Italy	50	91	81.2%
Latvia	26	95	268.2%
Lithuania	27	107	302.2%
Luxembourg	23	83	262.6%
Malta	19	136	611.6%
Netherlands	28	87	208.3%
Poland	35	109	209.2%
Portugal	30	109	258.8%
Romania	30	113	275.3%
Slovakia	31	76	105.5%
Slovenia	24	74	208.2%
Spain	45	107	137.4%
Sweden	29	118	304.3%
United Kingdom	38	103	173.1%

In their home country, residents of Malta (19 ms), Luxembourg (22 ms) and Cyprus (23 ms) enjoyed the lowest latencies in Europe. Residents of France, Spain and Italy experienced the highest latencies in their home countries at 41 ms, 45 ms and 50 ms, respectively.

European residents experienced an increase of at least 81.2% in their latency when visiting other countries within Europe. Residents from Cyprus, Malta and Bulgaria experience the biggest increase in latency when visiting other countries in Europe, ranging from Cyprus’ 825.5% to Bulgaria’s 368.3%. The countries that experienced the smallest increase in latency were Italy (81.2%), France (98.8%) and Germany (128.1%).

Locals spend more time on Wi-Fi than visitors in Europe

While there are no roaming changes within the E.U., users often opt to connect to Wi-Fi to avoid additional data overages at home and abroad. The following table compares the percentage of time spent on Wi-Fi by a resident of a country with that of a visitor to the same country during Q3-Q4 2019.

Percentage of Time Spent on Wi-Fi in Europe

Speedtest^® Data | Q3-Q4 2019

Country	Local Customers	Visitors	% Change
Austria	75.5%	40.5%	-46.3%
Belgium	77.9%	35.8%	-54.1%
Bulgaria	74.0%	60.3%	-18.5%
Croatia	74.3%	53.8%	-27.5%
Cyprus	79.3%	70.2%	-11.5%
Czech Republic	80.1%	46.6%	-41.8%
Denmark	78.7%	60.3%	-23.4%
Estonia	70.3%	51.9%	-26.2%
Finland	63.0%	60.3%	-4.4%
France	69.3%	48.7%	-29.7%
Germany	80.0%	47.1%	-41.1%
Greece	79.3%	64.3%	-19.0%
Hungary	76.8%	47.8%	-37.8%
Ireland	75.6%	58.9%	-22.0%
Italy	70.9%	55.1%	-22.2%
Latvia	68.8%	49.2%	-28.4%
Lithuania	73.8%	52.2%	-29.2%
Luxembourg	72.4%	30.8%	-57.4%
Malta	79.4%	65.9%	-17.1%
Netherlands	81.8%	49.1%	-40.1%
Poland	70.2%	58.2%	-17.1%
Portugal	76.5%	61.9%	-19.0%
Romania	69.6%	64.1%	-7.9%
Slovakia	76.2%	40.1%	-47.4%
Slovenia	70.7%	24.7%	-65.2%
Spain	78.3%	62.8%	-19.7%
Sweden	81.8%	53.8%	-34.3%
United Kingdom	79.3%	60.9%	-23.1%

As we saw last year, Finnish residents spent the least amount of time on Wi-Fi at 63.0%. Latvia showed the second lowest time spent on Wi-Fi (68.8%), followed by France (69.3%). Residents from the Netherlands, Sweden and the Czech Republic spent the most time on Wi-Fi at 81.8%, 81.8%, and 80.1%, respectively.

Visitors to Cyprus, Malta and Greece spent the most amount of time on Wi-Fi during this period. Time spent on Wi-Fi by visitors ranged from 64.3% in Greece to 70.2% in Cyprus. Visitors to Slovenia, Luxembourg and Belgium as a destination country spent the least amount of time on Wi-Fi at 25.7%, 30.8% and 35.8%, respectively.

Are you interested in more data on roaming performance? Read more here!

About the Author

Maria Gómez

| More Articles

Ookla Research™ Articles

Maria Gómez | March 2, 2020

Mobile and Fixed Broadband Speeds and 4G Availability in Spain

Versión en español

We investigated Spain’s fixed broadband and mobile network performance, 4G Availability, how Spain’s speeds and coverage compare with neighboring European countries and the performance and 4G Availability of top providers in Spain’s ten largest cities during Q3-Q4 2019. In addition to ranking Spain’s providers by top speeds and coverage, we also analyzed the effects of mobile plans with speed caps on Speedtest® results and ranked the providers on consistency of their network performance.

Spain and France were the fastest on our list for fixed broadband

We compared Spain’s fixed broadband performance at the country level to several other western European countries during Q3-Q4 2019. Spain was well positioned for mean download speed on fixed broadband at 113.72 Mbps, second only to France’s 119.37 Mbps. Spain had the fastest mean upload speed on this list at 103.72 Mbps. Italy had the slowest mean download speed on this list at 56.72 Mbps while the United Kingdom had the slowest mean upload speed at 18.28 Mbps.

fixed-broadband-speeds-eng-1

France showed the highest increase in mean download speed during Q3-Q4 2019 at 21.5%, followed by Portugal and Italy at 15.4% and Spain at 14.5%. Speeds remained relatively flat for this period for Italy, the U.K. and Germany.

An examination of latency over fixed broadband during Q3-Q4 2019 revealed a very different performance order. Portugal had the lowest latency during this period at 16 ms. Germany and the U.K. were tied for second at 23 ms. Spain was third (25 ms), Italy fourth (30 ms) and France fifth (32 ms).

Spain ranks fourth for mobile download speed

mobile-speeds-eng-1

Spain did not rank as well for mean download speed over mobile as they did on fixed broadband, coming in fourth with 33.97 Mbps during Q3-Q4 2019. This was 22.8% slower than first-place France. However, Spain showed the fastest mean upload speeds on mobile at 13.12 Mbps during Q3-Q4 2019.

Portugal saw the largest increase in mobile download speed during Q3-Q4 2019 at 18.4%. Spain’s mean download speed over mobile increased only 6.7% during this period. France showed the smallest increase in mobile download speed at 2.8%.

As we saw on fixed broadband, Portugal had the lowest mobile latency at 32 ms. Germany was second at 41 ms, France third at 43 ms, the U.K. fourth at 44 ms, Spain fifth at 47 ms and Italy last at 52 ms.

United Kingdom and Spain lead in 4G Availability

We used data from Speedtest coverage scans on Android to compare 4G Availability in each market during Q3-Q4 2019. While the U.K. showed the highest 4G Availability at 88.8%, Spain tied for a very close second, with 4G available in 87.7% of tested locations. Germany had the lowest 4G Availability at 72.1% during this period.

4G Availability in Major European Markets

Speedtest Intelligence^® | Q3-Q4 2019

Country	4G Availability
United Kingdom	88.8%
Spain	87.7%
France	87.7%
Italy	86.7%
Portugal	85.0%
Germany	72.1%

Looking specifically at Spain’s top mobile providers, Orange had the highest 4G Availability at 88.1%. Movistar was second, Yoigo third, and Vodafone fourth.

4G Availability by Provider in Spain

Speedtest Intelligence^® | Q3-Q4 2019

Provider	4G Availability
Orange	88.1%
Movistar	87.7%
Yoigo	87.4%
Vodafone	86.1%

Spain’s 5G deployments are expanding

5G-Deployments-in-Spain-0220-1
The excitement for 5G is evident in Spain. Vodafone had commercially available 5G deployments in a total of 18 Spanish cities as of February 13, 2020, including Barcelona, Madrid, Valencia, Málaga and Bilbao. Speedtest data reveals the mean download speed on 5G for Spain in Q3-Q4 2019 was 350.68 Mbps — an order of magnitude faster than the national average for mobile download speed — and the mean upload speed over 5G was 31.82 Mbps. Mean latency over 5G in Spain was 27 ms during Q3-Q4 2019.

MÁSMÓVIL was Spain’s fastest fixed broadband provider

Using Speed Score^™, a metric that combines measures of download and upload speed, to compare top ISPs in Spain during Q3-Q4 2019, we found that MÁSMÓVIL led with a score of 133.75 on fixed broadband. Orange was in second place with 115.25, followed by Vodafone (105.07) and Movistar (103.01).

Fixed Broadband Performance by Provider in Spain

Speedtest Intelligence^® | Q3-Q4 2019

Provider	Speed Score^TM
MÁSMÓVIL	133.75
Orange	115.25
Vodafone	105.07
Movistar	103.01

Movistar was Spain’s fastest mobile operator

Movistar had the highest Speed Score among Spain’s mobile operators during Q3-Q4 2019 at 37.76. Orange was second at 33.02, Vodafone third at 26.34 and Yoigo fourth at 22.43. As we will discuss below, overall speed performance can be impacted by the speed caps and tariff plans that are in place in Spain.

Mobile Performance by Operator in Spain

Speedtest Intelligence^® | Q3-Q4 2019

Provider	Speed Score^TM
Movistar	37.76
Orange	33.02
Vodafone	26.34
Yoigo	22.43

Speed capping affects overall performance

Speed capping, a limit of service imposed onto an internet connection by an operator, allows operators to also offer plans that focus on affordability rather than maximum performance.
Ookla_Distribution-Speedtest-Results_Vodafone_Spain_0220-3_en
The chart above shows how Vodafone’s introduction of subscription plans that use speed capping affected the distribution of download speeds. While those plans (with caps at 2 Mbps and 10 Mbps, respectively) were available starting in April, we don’t see the true effect on average speeds until May and June when the 0-5 Mbps and 10-15 Mbps buckets start to spike. We’ve only showed bins up to 60 Mbps to make this and the following graphs more legible.

In contrast, we see more even distributions of download speeds among all other mobile operators in Spain.
Ookla_Distribution-Speedtest-Results_Movistar_Spain_0220-3_en

MÁSMÓVIL and Movistar (mobile) offer the most consistent speeds

Speed is important, but if those speeds are inconsistent, it becomes difficult for users to reliably use their devices to stream HD video, browse the web or use online gaming. We used Speedtest data to calculate each top provider’s APS (Acceptable Performance Score) in Spain during Q3-Q4 2019.

APS for fixed broadband measures the percentage of samples that equal or exceed a download speed of 25 Mbps and also equal or exceed an upload speed of 3 Mbps. MÁSMÓVIL provided the most consistent experience in Spain on fixed broadband in Q3-Q4 2019 with an APS of 81.6%. Movistar was in last place with an APS of 61.9%.

Speed Consistency by Provider on Fixed Broadband

Speedtest Intelligence^® | Q3-Q4 2019

Provider	APS
MÁSMÓVIL	81.6%
Orange	75.1%
Vodafone	73.5%
Movistar	61.9%

APS on mobile measures the percentage of samples that equal or exceed a download speed of 5 Mbps and also equal or exceed an upload speed of 1 Mbps. Movistar took the top spot for consistency on mobile broadband with an APS of 88.9%. Vodafone had the lowest consistency score with an APS of 79.8%.

Speed Consistency by Operator on Mobile

Speedtest Intelligence^® | Q3-Q4 2019

Provider	APS
Movistar	88.9%
Orange	84.3%
Yoigo	82.6%
Vodafone	79.8%

Madrid led fixed and mobile broadband speeds

Ookla_Fixed-Broadband-Speeds_Spain_0220-1
We used Speedtest data from 2,183,336 user-initiated tests to investigate internet speeds in Spain’s ten most populous cities during Q3-Q4 2019. Madrid had the fastest mean download speeds on both fixed broadband and mobile.

Mean download speeds on fixed broadband ranged from 141.87 Mbps in Madrid to 91.66 Mbps in Las Palmas de la Gran Canaria, a 35.4% difference. Madrid also had the fastest mean upload speed over fixed broadband at 145.08 Mbps. Oviedo took the last place in upload speeds on fixed broadband with a 61.4% slower upload speed than Madrid.

Oviedo had the lowest latency over fixed broadband of the cities on our list during Q3-Q4 2019 at 15 ms. Málaga had the highest latency at 28 ms.
Ookla_Mobile-Broadband-Speeds_Spain_0220-1
The gap between fastest and slowest speeds between Spanish cities on mobile broadband was similar to what we saw fixed broadband during Q3-Q4 2019. Madrid once again led in performance on mobile broadband with a mean download speed of 44.35 Mbps and a mean upload speed of 15.89 Mbps. Las Palmas was in last place on download speeds with a 37.9% slower mean download speed than Madrid. Valencia had the slowest mean upload speed over mobile at 13.86 Mbps. Seville had the lowest latency over mobile at 36 ms during Q3-Q4 2019 and Las Palmas had the highest mobile latency at 73 ms.

Zaragoza led 4G Availability in Spain’s most populous cities

4G Availability in Spain’s Largest Cities

Speedtest Intelligence^® | Q3-Q4 2019

City	4G Availability
Zaragoza	97.2%
Valencia	97.1%
Seville	96.8%
Málaga	96.7%
Oviedo	95.4%
Madrid	95.1%
Las Palmas de la Gran Canaria	95.0%
Alicante	94.6%
Bilbao	94.3%
Barcelona	87.7%

4G Availability was higher in Spain’s largest cities than the country’s average, with Zaragoza having the highest 4G Availability at 97.2% of tested locations. Barcelona had the lowest 4G Availability on our list at 87.7%.

MÁSMÓVIL was fastest fixed broadband provider in 6 cities

Looking specifically at Speed Score in individual Spanish cities during Q3-Q4 2019, we found MÁSMÓVIL was the fastest provider in six cities: Málaga, Seville, Zaragoza, Madrid, Valencia and Alicante. Adamo, the fastest fixed provider in Barcelona, had the highest Speed Score on the list at 185.29. Vodafone was the fastest provider in Bilbao and Oviedo.

Fixed Broadband Performance by Operator in Spain’s Largest Cities

Speedtest Intelligence^® | Q3-Q4 2019

City	Provider	Speed Score^TM
Barcelona	Adamo	185.29
Málaga	MÁSMÓVIL	156.24
Seville	MÁSMÓVIL	148.78
Madrid	MÁSMÓVIL	145.83
Bilbao	Vodafone	144.65
Zaragoza	MÁSMÓVIL	144.40
Oviedo	Vodafone	143.85
Valencia	MÁSMÓVIL	143.99
Alicante	MÁSMÓVIL	143.72
Las Palmas	Orange	114.46

Looking at the fastest Spanish providers on mobile broadband for each city during Q3-Q4 2019, Movistar dominated the list with the fastest Speed Score in 7 cities. Movistar also had the highest Speed Score overall at 52.04 in Seville. Vodafone was the fastest mobile provider in Madrid (35.67). Orange was the fastest provider in Barcelona (35.37) and Oviedo (45.10).

Mobile Performance by Operator in Spain’s Largest Cities

Speedtest Intelligence^® | Q3-Q4 2019

City	Provider	Speed Score^TM
Seville	Movistar	52.04
Valencia	Movistar	50.80
Alicante	Movistar	50.36
Málaga	Movistar	48.97
Bilbao	Movistar	48.86
Oviedo	Orange	45.10
Zaragoza	Movistar	40.79
Madrid	Vodafone	35.67
Las Palmas	Movistar	35.37
Barcelona	Vodafone	31.37

We look forward to following these markets and investigating the changing landscape of internet performance and mobile coverage in Europe. Want to learn more about fixed and mobile speeds in different markets? Click here to read more Ookla Research.

Editor’s note: This article was updated on March 4, 2020 to correct an editing error that incorrectly reported the figures for mobile APS.

Velocidades de banda ancha fija y móvil, y disponibilidad de 4G en España

Hemos investigado el rendimiento de la banda ancha fija y móvil, así como la disponibilidad de 4G en España. También hemos comparado las velocidades y la cobertura españolas con las de los países europeos vecinos, así como el rendimiento y la disponibilidad de 4G de los principales proveedores de las diez ciudades más grandes de España en el segundo semestre de 2019. Además de clasificar a los proveedores españoles por su velocidad y cobertura máximas, también hemos analizado los efectos de los planes móviles con limitaciones de velocidad basándonos en los resultados de Speedtest® y hemos clasificado a los proveedores según la constancia del rendimiento de sus redes.

España y Francia fueron los países más rápidos de nuestra lista de banda ancha fija

Hemos comparado el rendimiento de la banda ancha fija de España a nivel de país con varios otros países de la Europa occidental en el segundo semestre de 2019. España obtuvo una buena posición en cuanto a velocidad de descarga media en banda ancha fija, con 113,72 Mbps, en segundo lugar solo después de los 119,37 Mbps de Francia. España registró la velocidad de subida media más alta de esta lista: 103,72 Mbps. Italia obtuvo la velocidad de descarga media más baja de esta lista, 56,72 Mbps, mientras que el Reino Unido registró la velocidad de subida media más baja, 18,28 Mbps.

fixed-broadband-speeds-sp-2

Francia registró el mayor aumento en la velocidad de descarga media del segundo semestre de 2019, con el 21,5 %, seguida de Portugal y Italia, con el 15,4 %, y España, con el 14,5 %. Las velocidades se mantuvieron relativamente invariables durante este periodo en Italia, Reino Unido y Alemania.

Un análisis de la latencia en banda ancha fija en el segundo semestre de 2019 reveló una clasificación por rendimiento muy distinta. Portugal se anotó la latencia más baja de este periodo: 16 ms. Alemania y Reino Unido empataron en el segundo puesto con 23 ms. España resultó tercera (25 ms), Italia cuarta (30 ms) y Francia quinta (32 ms).

España, cuarto país en velocidad de descarga en Internet móvil

mobile-speeds-sp-3

La clasificación de España en velocidad de descarga media por móvil no fue tan buena como la de banda ancha fija, quedándose con un cuarto puesto con 33,97 Mbps en el segundo semestre de 2019. Esto supuso una velocidad un 22,8 % más lenta que la del primer país clasificado, Francia. Sin embargo, España registró las velocidades de subida medias más altas en Internet móvil, con 13,12 Mbps en el segundo semestre de 2019.

Portugal experimentó el mayor aumento en velocidad de descarga de Internet móvil en el segundo semestre de 2019, el 18,4 %. La velocidad de descarga media de España en Internet móvil aumentó solo un 6,7 % en este periodo. Francia registró el menor aumento de velocidad de descarga de Internet móvil, un 2,8 %.

Como vimos en la banda ancha fija, la latencia móvil más baja fue la de Portugal, con 32 ms. Alemania fue segunda con 41 ms, Francia tercera con 43 ms, Reino Unido cuarto con 44 ms, España quinta con 47 ms e Italia última con 52 ms.

Reino Unido y España, líderes en disponibilidad de 4G

Utilizamos datos de exploraciones de cobertura de Speedtest en Android para comparar la disponibilidad de 4G en cada mercado en el segundo semestre de 2019. Mientras que el Reino Unido mostró la mayor disponibilidad de 4G con el 88,8 %, España y Francia ocuparon el segundo lugar, con 4G disponible en el 87,7 % de las ubicaciones probadas. Durante ese período, Alemania tuvo la menor disponibilidad de 4G con un 72,1 %.

Disponibilidad de 4G en los principales mercados europeos

Speedtest Intelligence^® | Segundo semestre de 2019

País	Disponibilidad de 4G
Reino Unido	88,8 %
España	87,7 %
Francia	87,7 %
Italia	86,7 %
Portugal	85,0 %
Alemania	72,1 %

Si nos fijamos específicamente en los principales proveedores móviles de España, Orange ofreció la mayor disponibilidad de 4G con un 88,1 %. Movistar fue segundo, Yoigo tercero, y Vodafone cuarto.

Disponibilidad de 4G por proveedor en España

Speedtest Intelligence^® | Segundo semestre de 2019

Proveedor	Disponibilidad de 4G
Orange	88,1 %
Movistar	87,7 %
Yoigo	87,4 %
Vodafone	86,1 %

Los despliegues de 5G en España se están expandiendo

5G-Deployments-in-Spain-0220_es
El entusiasmo por el 5G es evidente en España. Vodafone desplegó 5G comercialmente disponible en un total de 18 ciudades españolas a fecha del 13 de febrero de 2020, incluidas Barcelona, Madrid, Valencia, Málaga y Bilbao. Los datos de Speedtest revelan que en España la velocidad media de descarga en 5G en el segundo semestre de 2019 fue de 350,68 Mbps (velocidad superior a la media nacional para Internet móvil) y la velocidad de subida media por 5G fue de 31,82 Mbps. La latencia media por 5G en España fue de 27 ms en el segundo semestre de 2019.

MÁSMÓVIL fue el proveedor de banda ancha fija más rápido de España

Al utilizar Speed Score™, sistema que combina mediciones de velocidades de descarga y de subida, para comparar los principales proveedores de servicios de Internet de España en el segundo semestre de 2019, descubrimos que MÁSMÓVIL obtuvo los mejores resultados, con una puntuación de 133,75 en banda ancha fija. Orange ocupó el segundo lugar con 115,25, seguido de Vodafone (105,07) y Movistar (103,01).

Rendimiento de la banda ancha fija en España por proveedor

Speedtest Intelligence^® | Segundo semestre de 2019

Proveedor	Speed Score^TM
MÁSMÓVIL	133,75
Orange	115,25
Vodafone	105,07
Movistar	103,01

Movistar fue el operador móvil más rápido de España

Movistar obtuvo la máxima puntuación de velocidad de entre los operadores móviles de España en el segundo semestre de 2019, con 37,76. Orange fue segundo con 33,02, Vodafone la tercera con 26,34 y Yoigo cuarta con 22,43. Como veremos más adelante, el rendimiento general de la velocidad puede verse afectado por los límites de velocidad y los planes de tarifas vigentes en España.

Rendimiento de Internet móvil en España por operador

Speedtest Intelligence^® | Segundo semestre de 2019

Proveedor	Speed Score^TM
Movistar	37,76
Orange	33,02
Vodafone	26,34
Yoigo	22,43

La limitación de velocidad afecta al rendimiento general

La limitación de la velocidad, una restricción de servicio impuesta a una conexión de Internet por los operadores, permite a estos ofrecer también planes que se centren en la asequibilidad más que en el rendimiento máximo.
Ookla_Distribution-Speedtest-Results_Vodafone_Spain_0220-3_es
En el gráfico anterior se muestra cómo los planes de suscripción con limitación de velocidad introducidos por Vodafone afectaron a la distribución de las velocidades de descarga. Aunque esos planes (con limitaciones de 2 Mbps y 10 Mbps, respectivamente) estaban disponibles desde abril, no vemos un efecto verdadero en las velocidades medias hasta mayo y junio cuando las medidas de 0-5 Mbps y 10-15 Mbps empiezan a sobresalir.

En cambio, en el resto de operadores móviles de España vemos distribuciones más uniformes de las velocidades de descarga.
Ookla_Distribution-Speedtest-Results_Movistar_Spain_0220-3_es

MÁSMÓVIL y Movistar (móvil) ofrecen las velocidades más constantes

La velocidad es importante, pero si no es constante, resulta difícil que los usuarios utilicen sus dispositivos con fiabilidad para transmitir vídeo de alta definición, navegar por la web o jugar en línea. Utilizamos los datos de Speedtest para calcular la PRA (puntuación de rendimiento aceptable) de cada uno de los principales proveedores de España durante el segundo semestre de 2019.

La PRA de banda ancha fija mide el porcentaje de muestras que igualan o superan una velocidad de descarga de 25 Mbps y que también igualan o superan una velocidad de subida de 3 Mbps. MÁSMÓVIL proporcionó el mayor nivel de constancia de España en banda ancha fija en el segundo semestre de 2019 con una PRA de 81,6 %. Movistar quedó en último lugar con una PRA del 61,9 %.

Constancia de velocidad por proveedor en banda ancha fija

Speedtest Intelligence^® | Segundo semestre de 2019

Proveedor	PRA
MÁSMÓVIL	81,6 %
Orange	75,1 %
Vodafone	73,5 %
Movistar	61,9 %

La PRA de Internet móvil mide el porcentaje de muestras que igualan o superan una velocidad de descarga de 5 Mbps y que también igualan o superan una velocidad de subida de 1 Mbps. Movistar ocupó el primer lugar en constancia de banda ancha móvil con una PRA del 88,9 %, Vodafone obtuvo la puntuación más baja en constancia con una PRA del 79,8 %.

Constancia de velocidad por operador en Internet móvil

Speedtest Intelligence^® | Segundo semestre de 2019

Proveedor	PRA
Movistar	88,9 %
Orange	84,3 %
Yoigo	82,6 %
Vodafone	79,8 %

Madrid lideró las velocidades de banda ancha fija y móvil

Ookla_Fixed-Broadband-Speeds_Spain_0220_es-2
Para estudiar las velocidades de Internet de las diez ciudades más pobladas de España en el segundo semestre de 2019 utilizamos los datos de Speedtest de 2.183.336 pruebas realizadas por los usuarios. Madrid registró las velocidades medias de descarga más altas tanto en banda ancha fija como en Internet móvil.

Las velocidades medias de descarga por banda ancha fija oscilaron entre los 141,87 Mbps de Madrid y los 91,66 Mbps de Las Palmas de la Gran Canaria, una diferencia del 35,4 %. Madrid registró también la velocidad de subida media más alta por banda ancha fija, con 145,08 Mbps. Oviedo ocupó el último lugar en velocidad de subida por banda ancha fija con un 61,4 % menos de velocidad de subida que Madrid.

Oviedo tuvo la menor latencia en banda ancha fija de las ciudades de nuestra lista durante el segundo semestre de 2019, con 15 ms. Málaga registró la latencia más alta, con 28 ms.
Ookla_Mobile-Broadband-Speeds_Spain_0220_es-1
La brecha entre las velocidades más rápidas y más lentas de las ciudades españolas por banda ancha móvil fue similar a la que apreciamos en la banda ancha fija en el segundo semestre de 2019. Madrid volvió a liderar el rendimiento de la banda ancha móvil con una velocidad media de descarga de 44,35 Mbps y una velocidad media de subida de 15,89 Mbps. Las Palmas quedó en último lugar en velocidades de descarga con un 37,9 % menos de velocidad media de descarga que Madrid. Valencia registró la velocidad de subida media más baja por Internet móvil: 13,86 Mbps. Sevilla obtuvo la menor latencia por Internet móvil con 36 ms en el segundo semestre de 2019 y Las Palmas registró la mayor latencia por Internet móvil con 73 ms.

Zaragoza lideró la disponibilidad de 4G en las ciudades españolas más pobladas

Disponibilidad de 4G en las ciudades más grandes de España

Speedtest Intelligence^® | Segundo semestre de 2019

Ciudad	Disponibilidad de 4G
Zaragoza	97,2 %
Valencia	97,1 %
Sevilla	96,8 %
Málaga	96,7 %
Oviedo	95,4 %
Madrid	95,1 %
Las Palmas de la Gran Canaria	95,0 %
Alicante	94,6 %
Bilbao	94,3 %
Barcelona	87,7 %

La disponibilidad de 4G en las ciudades más grandes de España superó el promedio del país, siendo Zaragoza la de mayor disponibilidad de 4G con el 97,2 % de las ubicaciones analizadas. Barcelona tuvo la menor disponibilidad de 4G de nuestra lista, con el 87,7 %.

MÁSMÓVIL fue el proveedor de banda ancha fija más rápido de 6 ciudades

Si nos fijamos específicamente en la puntuación de velocidad en cada una de las ciudades españolas en el segundo semestre de 2019, vemos que MÁSMÓVIL fue el proveedor más rápido en seis ciudades: Málaga, Sevilla, Zaragoza, Madrid, Valencia y Alicante. Adamo, el proveedor fijo más rápido de Barcelona, consiguió la puntuación de velocidad más alta de la lista con 185,29. Vodafone fue el proveedor más rápido en Bilbao y Oviedo.

Rendimiento de la banda ancha fija por operador en las ciudades más grandes de España

Speedtest Intelligence^® | Segundo semestre de 2019

Ciudad	Proveedor	Speed Score^TM
Barcelona	Adamo	185,29
Málaga	MÁSMÓVIL	156,24
Sevilla	MÁSMÓVIL	148,78
Madrid	MÁSMÓVIL	145,83
Bilbao	Vodafone	144,65
Zaragoza	MÁSMÓVIL	144,40
Valencia	MÁSMÓVIL	143,99
Oviedo	Vodafone	143,85
Alicante	MÁSMÓVIL	143,72
Las Palmas	Orange	114,46

Si nos fijamos en los proveedores españoles más rápidos en banda ancha móvil de cada ciudad en el segundo semestre de 2019, Movistar dominó la lista con la puntuación de velocidad más alta en 7 ciudades. Movistar consiguió también la puntuación de velocidad máxima, con 52,04 en Sevilla. Vodafone fue el proveedor de Internet móvil más rápido en Madrid (35,67). Orange fue el proveedor más rápido de Barcelona (35,37) y Oviedo (45,10).

Rendimiento de Internet móvil por operador en las ciudades más grandes de España

Speedtest Intelligence^® | Segundo semestre de 2019

Ciudad	Proveedor	Speed Score^TM
Sevilla	Movistar	52,04
Valencia	Movistar	50,80
Alicante	Movistar	50,36
Málaga	Movistar	48,97
Bilbao	Movistar	48,86
Oviedo	Orange	45,10
Zaragoza	Movistar	40,79
Madrid	Vodafone	35,67
Barcelona	Orange	35,37
Las Palmas	Movistar	31,37

Nos encantará seguir la evolución de estos mercados y estudiar el cambiante panorama del rendimiento de Internet y de la cobertura móvil en Europa. Si desea obtener más información sobre las velocidades fijas y móviles en diferentes mercados, haga clic aquí para leer más información de Ookla.

Nota del editor: Este artículo fue actualizado el 4 de marzo del 2020 que para corregir un error de edición que reportaba cifras incorrectas del PRA móvil.

About the Author

Maria Gómez

| More Articles

Ookla Research™ Articles

Maria Gómez | March 31, 2020

Exploring Internet Performance in Malaysia

Internet performance in Malaysia as a whole is similar to performance in other major Southeast Asian markets, but performance within Malaysia varies greatly at the state and city level. This article explores the current state of Malaysia’s fixed broadband and mobile network performance, including data on: internet speeds, latency and 4G Availability at the country level. We include information on how Malaysia compares to major Southeast Asian countries and examines performance variations across Malaysian states and cities during Q3-Q4 2019.

Malaysia’s fixed broadband ranks third in Southeast Asia

Ookla_Fixed_Speeds_Chart_Malaysia_0320
We compared Malaysia’s fixed broadband performance at the country level to other major markets in Southeast Asia during Q3-Q4 2019. Singapore ranked first on our list for both download and upload speeds over fixed broadband with a mean download speed of 191.89 Mbps and a mean upload speed of 199.32 Mbps. Malaysia ranked third for download speed with a mean speed of 76.69 Mbps. Indonesia had the slowest mean download and upload speeds over fixed broadband during this period. At 20.49 Mbps, Indonesia’s mean download speed was 836.6% slower than that of Singapore.

Thailand showed the highest increase in mean download speed over fixed broadband during Q3-Q4 2019 at 59.4%. Vietnam, the Philippines and Malaysia had more modest increases in mean download speed at 11.8%, 11.3% and 9.1%, respectively. Indonesia followed Malaysia with a 5.4% increase in mean download speed over fixed broadband. Singapore’s mean download speed increased only by 4.2% during this period.

Speedtest data for latency in major Southeast Asian markets during Q3-Q4 2019 revealed Vietnam had the lowest latency on the list at 9 ms. Singapore was second at 11 ms, Malaysia was in a less favourable fifth place with a latency of 24 ms, followed only by the Philippines with the highest latency on the list at 34 ms.

Malaysia ranks fourth in Southeast Asia for mobile broadband download speed

Ookla_Mobile_Speeds_Chart_Malaysia_0320
During Q3-Q4 2019, Singapore had the fastest mean download speed on mobile in major Southeast Asian markets at 52.28 Mbps, followed by Vietnam at 26.28 Mbps. Malaysia ranked fourth for mean download speed over mobile with 22.12 Mbps, while Indonesia was last with 12.65 Mbps.

Rankings for mean upload speed over mobile during Q3-Q4 2019 followed almost the same rankings as we saw for download speed with Singapore at the top of the list at 19.62 Mbps. Malaysia ranked fourth with a mean upload speed of 11.40 Mbps over mobile. In this category, the Philippines ranked last with a mean upload speed of 7.12 Mbps on mobile.

Thailand experienced the largest increase in mobile download speed during Q3-Q4 2019 at 28.7%. Vietnam followed closely with an increase of 23.5% in mean download speed. Malaysia was third with an increase of 11.8% in mean download speed on mobile broadband, only slightly better than the 6.4% increase in mean download speed in the Philippines.

Malaysia was second only to the Philippines for mobile latency during Q3-Q4 2019 at 29 ms. Thailand had the highest latency on this list at 52 ms.

How Malaysian mobile operators are preparing for 5G

Malaysian mobile operators spent the better half of 2019 conducting 5G field trials and signing multiple memorandums of understanding (MoUs) with infrastructure vendors in preparation for commercial deployments.

In February 2019, Maxis and Huawei signed an MoU under which the two companies agreed to collaborate on 5G field trials. By October, the two companies inked a deal ensuring a full-fledged deployment of 5G equipment and services, which also involves modernizing the existing LTE infrastructure supplied by Huawei.

Similarly, U Mobile and ZTE followed up with an MoU in March, but the operator signed a three-year contract with Nokia in which the vendor will supply Single RAN and transport infrastructure, paving the way towards commercial 5G rollouts.

According to the Malaysian Communications and Multimedia Commission (MCMC), 5G commercial rollouts are projected for Q3 2020 using spectrum allocations in the 700MHz, 3.5GHz and 26GHz-28GHz millimeter wave frequency bands.

In January 2020, numerous 5G related announcements were made by Malaysian operators. Telekom Malaysia and Digi Telecommunications announced a 5G demonstration project during which the two operators will be exploring both mobile and fixed 5G use cases. Under this project, Digi will operate Radio Access and Core network while leveraging fiber backhaul provided by Telekom Malaysia. The two operators also explored 5G network sharing possibilities. In parallel, Telekom Malaysia and U Mobile joined forces to explore network sharing opportunities both with shared and dedicated spectrum licenses. This will help the two operators better understand the economic and technological efficiencies associated with 5G network sharing.

We have been tracking the progress of 5G testing across Malaysia using Speedtest data and have seen multiple operators achieve download speeds over 1 Gbps, upload speeds over 100 Mbps and single-digit latency. Celcom and Maxis have successfully conducted 5G MOCN (Multi Operator Core Network) limited trials, reaching peak download speeds of over 1.1 Gbps. MOCN functionality allows two or more operators to use the same radio access network while maintaining individual network cores. During the same month, Telekom Malaysia successfully tested standalone 5G using aggregated 700MHz and 3.5GHz achieving a downlink throughput of 1.5 Gbps.

Selangor had the fastest fixed broadband in Malaysia

We explored internet speeds and 4G Availability in Malaysia’s 13 states and 3 federal territories using Speedtest data during Q3-Q4 2019.
Ookla_Malaysia_States_Fixed-Broadband-Speeds_0320
Selangor had the fastest mean download and upload speeds on fixed broadband during Q3-Q4 2019, as well as having the lowest latency (14 ms). Mean download speeds on fixed broadband ranged from a high of 91.83 Mbps in Selangor to a low of 36.62 Mbps in Kedah, a 60.1% difference. Selangor’s mean upload speed over fixed broadband was 57.89 Mbps while Kelantan had the slowest mean upload speed over fixed broadband at 24.93 Mbps. Labuan had the highest latency at 69 ms over fixed broadband during this period.
Ookla_Malaysia_States_Mobile-Speeds_0320
The difference between fastest and slowest speeds on mobile broadband was smaller than that of fixed broadband. Sarawak led the group with the fastest mean download and upload speeds on mobile broadband at 25.71 Mbps and 12.31 Mbps, respectively. Perlis was slowest with a mean download speed of 16.49 Mbps, a 35.9% difference. Kelantan had the slowest mean upload speed on mobile at 9.77 Mbps. Latency rankings over mobile were very different among the states and territories than those for speeds. The Federal Territory of Putrajaya had the lowest mobile latency during Q3-Q4 2019 at 33 ms. Sarawak had the highest mobile latency at 59 ms during this period.

4G dominates throughout Malaysia

Ookla_Malaysia_Best-Available-Signal_0320
The map above illustrates the best available mobile technologies throughout the country of Malaysia as represented in Speedtest data. We saw 4G available in most parts of Malaysia that were tested, especially in the western coast of the Malaysian peninsula. The second most prevalent signal was 3G, often found at the edges of places where 4G is more readily available. 2G is the least available signal and is found mostly in East Malaysia, near Brunei. 5G is currently not commercially available in the country, but deployments are expected by Q3 2020.
Ookla_Malaysia_States_4G-Availability_0320
The Federal Territory of Kuala Lumpur had the highest 4G Availability with 97.6% of tested locations showing access to 4G during Q3-Q4 2019. Pahang had the lowest 4G Availability at 66.8%.

Kuala Lumpur was fastest city for fixed broadband

Speedtest data on internet speeds and 4G Availability in Malaysia’s 10 largest cities during Q3-Q4 2019 revealed a stark difference in speeds between cities on fixed broadband. The nation’s capital of Kuala Lumpur led the group with the fastest mean download and upload speeds on fixed broadband at 82.38 Mbps and 53.96 Mbps, respectively. George Town had the slowest mean download speed on fixed broadband at 39.47 Mbps, 52.1% slower than Kuala Lumpur. Ipoh was last for mean upload speed over fixed broadband at 30.27 Mbps, 43.9% slower than Kuala Lumpur.
Ookla_Malaysia_Cities_Fixed-Broadband-Speeds_0320

Kuala Lumpur came on top once again with the lowest mean latency over fixed broadband on the list at 17 ms during Q3-Q4 2019. Kota Kinabalu showed the highest mean latency over fixed broadband at 58 ms.

Nusajaya ranks first among cities for mobile broadband

Differences in speed on mobile broadband were not as stark as those on fixed broadband for Malaysia’s ten largest cities during Q3-Q4 2019. First-place Nusajaya showed a mean download speed on mobile of 28.10 Mbps, followed closely by Kuantan at 26.51 Mbps. Alor Setar ranked last with a mean download speed of 19.05 Mbps, a 32.2% difference from Nusajaya.
Ookla_Malaysia_Cities_Mobile-Speeds_0320
Upload speeds on mobile showed a very different ranking from download speeds with Kota Kinabalu first with a mean upload speed of 13.18 Mbps. Kuala Lumpur was second to last at 11.79 Mbps, followed only by Seremban with a mean upload speed of 11.27 Mbps.

Latency over mobile was higher than latency over fixed for most cities during Q3-Q4 2019. First place Seremban had the lowest latency at 35 ms. Kota Kinabalu was last at 53 ms.

We’ll continue to follow Malaysia’s internet speeds, mobile performance and 5G deployments. If you’d like to learn more about internet speeds in Southeast Asia and other markets around the world, click here to read more insights from Ookla.

Editor’s note: This article was updated on March 31 to correct a typo in the second paragraph that incorrectly identified Malaysia’s mean download speed.

About the Author

Maria Gómez

| More Articles

Ookla Research™ Articles

Maria Gómez | July 28, 2020

Exploring the Online Gaming Experience in Latin America

Español | Português

Online gaming is on the rise in Latin America and all over the world. This is creating a demand for lower-latency network connections so gamers can enjoy a seamless, lag-free gameplay experience. According to data from Speedtest Intelligence® during Q2 2020, latency varied greatly across both countries and cities in the five most robust economies in Latin America, providing some gamers with an unexpected advantage. Read on for details about latency on mobile and fixed broadband in Argentina, Brazil, Chile, Colombia and Mexico to see where users are likely to have the best gaming experience both nationwide and in each country’s most populous cities.

Latency, the reaction time of a connection, is a key metric in gaming. For the best gaming experience, users should look for a latency of 59 ms or less. The data below provides the mean latency for each location. Mobile latency is applicable for games played on mobile devices, while latency on fixed broadband measures latency while playing PC and console games.

Brazil had lowest latency on fixed broadband, Argentina on mobile

Ookla_Latin_America_latency_countries_0720_en

Gamers in Brazil enjoyed the lowest mean latency on fixed broadband during Q2 2020 at 19 ms. Brazil was followed by Chile, Mexico and Argentina. Colombia had the highest latency on fixed broadband during this period at 43 ms.

Recent investments in fiber all across the region resulted in improved latencies when comparing Q2 2019 and Q2 2020. Brazil showed the biggest improvement in fixed broadband latency compared to Q2 2019, decreasing from 23 ms in Q2 2019 to 19 ms in Q2 2020. Latency in Mexico also decreased, dropping from 34 ms in Q2 2019 to 31 ms in Q2 2020. The other three countries on our list experienced little to no change in latency on fixed broadband between Q2 2019 and Q2 2020.

There was less variation between countries in mobile latency during Q2 2020. Argentina had the best latency on mobile at 40 ms, followed closely by Chile at 41 ms. Colombia had the highest mobile latency during this period at 47 ms.

Mexico experienced the largest improvement in their mobile latency when comparing Q2 2019 to Q2 2020. Mobile latency improved from 57 ms in Q2 2019 to 44 ms in Q2 2020. The other countries on our list saw smaller improvements in their mobile latency during this period. In Chile, mobile latency actually increased year over year, from 38 ms in Q2 2019 to 41 ms in Q2 2020.

Deployments of 5G in the region have recently begun, including the launch of 5G DSS technology in Brazil. Though the technology is still fairly new to Latin America, it has the potential to radically improve latency over what 4G-capable devices currently offer.

Internet speeds increased during the pandemic

Having a fast internet connection is also important to gamers. We’ve been using data from Speedtest Intelligence to track internet performance at a global level during the pandemic. While some countries experienced a dip in speeds in March, on the whole, internet speeds on fixed broadband have increased in Argentina, Brazil, Chile, Colombia and Mexico since the baseline week of March 2, 2020. Except for Chile, these countries have also experienced an increase in mobile speeds, ranging from a 2% increase in Colombia to a 19% increase in Mexico.

São Paulo had lowest fixed broadband latency, tied with Buenos Aires on mobile

Ookla_Latin_America_latency_cities_0720_en2
Fixed broadband latency in some of Latin America’s largest cities ranged from Guadalajara’s 17 ms to 45 ms in Cali. Guadalajara, São Paulo and Rio de Janeiro had the lowest fixed broadband latencies on our list during Q2 2020 at 17 ms, 18 ms and 20 ms, respectively. The Colombian cities of Bogotá, Medellín and Cali showed the highest latencies over fixed broadband on our list at 38 ms, 40 ms and 45 ms, respectively. All cities on this list still showed fixed broadband latencies below 59 ms, the recommended measurement for a smooth gaming experience.

When comparing year-over-year results from Q2 2019, we saw increases in fixed broadband latency in almost all cities in Q2 2020. Cali was the only city on the list where latency improved, down from 49 in Q2 2019 to 45 ms in Q2 2020.

São Paulo and Buenos Aires tied for the lowest mobile latency on our list during Q2 2020 at 31 ms. Rio de Janeiro followed at 35 ms, respectively. Brasília, Cali and Medellín took the last three places on the list with mobile latencies of 43 ms, 48 ms and 51 ms, respectively.

Mexico City’s and São Paolo’s mobile latencies improved significantly in Q2 2020 compared to Q2 2019. Mexico City’s mobile latency decreased from 53 ms to 37 ms during the same period. In São Paolo, mobile latency decreased from 40 ms in Q2 2019 to 31 ms in Q2 2020. Only one city on our list did not show an improvement in mobile latency when comparing Q2 2019 to Q2 2020 — Santiago, where mobile latency increased from 36 ms to 40 ms during this period.

Devoted gamers in Latin America will be glad to hear that Internet speeds and mobile latency are improving in many parts of Latin America. We look forward to seeing if fixed broadband latency makes it to the next level in the near future. If you’re interested in learning more about internet performance in other parts of the world, visit Ookla Research^TM.

Exploramos la experiencia de juegos en línea en América Latina

Los juegos en línea están en aumento en América Latina y en todo el mundo. Esto crea una demanda de conexiones de red con latencia más baja para que los jugadores puedan disfrutar una experiencia de juego sin demoras ni problemas. Según los datos de Speedtest Intelligence® durante el segundo trimestre de 2020, la latencia varió mucho entre ambos países y ciudades en las cinco economías más sólidas de América Latina, lo que les da a algunos jugadores una ventaja inesperada. Lea los detalles sobre la latencia en banda ancha móvil y fija en Argentina, Brasil, Chile, Colombia y México para ver qué usuarios es probable que tengan la mejor experiencia de juego en toda la nación y en las ciudades más pobladas del país.

La latencia, el tiempo de reacción de una conexión, es una métrica clave en el juego en línea. Para lograr la mejor experiencia de juego en línea, los usuarios deben buscar una latencia de 59 ms o menos. Los datos siguientes indican la latencia media para cada ubicación. La latencia móvil se aplica a los juegos que se juegan en dispositivos móviles, mientras que la latencia de banda ancha fija se mide cuando se juega en consolas y computadoras.

Brasil tiene la latencia más baja en banda ancha fija y Argentina, en móvil

Ookla_Latin_America_latency_countries_0720_es
Los jugadores en Brasil disfrutaron la latencia media más baja en banda ancha fija durante el segundo trimestre de 2020 con 19 ms. A Brasil, lo siguieron Chile, México y Argentina. Colombia tuvo la latencia más alta en banda ancha fija durante este período con 43 ms.

Las inversiones recientes en cable de fibra óptica en toda la región dieron como resultado mejores latencias cuando comparamos el segundo trimestre de 2019 y el segundo trimestre de 2020. Brasil mostró una gran mejora en la latencia de banda ancha fija en comparación con el segundo trimestre de 2019, que disminuyó de 23 ms en el segundo trimestre de 2019 a 19 ms en el segundo trimestre de 2020. La latencia en México también cayó de 34 ms en el segundo trimestre de 2019 a 31 ms en el segundo trimestre de 2020. Los otros tres países de nuestra lista no experimentaron prácticamente ningún cambio en la latencia de banda ancha fija entre el segundo trimestre de 2019 y el segundo trimestre de 2020.

Hubo una variación entre los países en la latencia móvil durante el segundo trimestre de 2020. Argentina tuvo la mejor latencia móvil en 40 ms, seguido por Chile con 41 ms. Colombia tuvo la latencia móvil más alta durante este período con 47 ms.

México experimentó la mejora más amplia en su latencia móvil cuando se compara el segundo trimestre de 2019 y el segundo trimestre de 2020. La latencia móvil mejoró de 57 ms en el segundo trimestre de 2019 a 44 ms en el segundo trimestre de 2020. Los demás países de nuestra lista tuvieron pequeñas mejoras en su latencia móvil durante este período. En Chile, la latencia móvil aumentó año tras año, de 38 ms en el segundo trimestre 2019 a 41 ms en el segundo trimestre de 2020.

Las implementaciones de 5G en la región han comenzado recientemente, incluyendo el lanzamiento de la tecnología 5G DSS en Brasil. Aunque la tecnología todavía es bastante nueva en América Latina, tiene el potencial de mejorar radicalmente la latencia sobre lo que ofrecen actualmente los dispositivos con capacidad 4G.

Las velocidades de Internet aumentaron durante la pandemia

Tener una conexión a Internet rápida también es importante. Hemos usado datos de Speedtest Intelligence para rastrear el rendimiento de Internet a nivel global durante la pandemia. Si bien algunos países sufrieron una caída en las velocidades en Marzo, en líneas generales, las velocidades de banda ancha fija aumentaron en Argentina, Brasil, Chile, Colombia y México desde la semana de referencia del 2 de marzo de 2020. Salvo por Chile, estos países también experimentaron un aumento en las velocidades móviles, que varió de un aumento de 2 % en Colombia a un aumento de 19 % en México.

São Paulo tuvo la latencia de banda ancha fija más baja junto con Buenos Aires en móvil

Ookla_Latin_America_latency_cities_0720es-01
Latencia de banda ancha fija en algunas de las ciudades más grandes de América Latina, desde 17 ms en Guadalajara a 45 ms en Cali. Guadalajara, São Paulo y Río de Janeiro tuvieron la latencia de banda ancha fija más baja de nuestra lista durante el segundo trimestre de 2020 con 17 ms, 18 ms y 20 ms, respectivamente. Las ciudades colombianas de Bogotá, Medellín y Cali tuvieron las latencias más altas de banda ancha fija de nuestra lista con 38 ms, 40 ms y 45 ms, respectivamente. Todas las ciudades de esta lista tuvieron latencias de banda ancha fija por debajo de 59 ms, la medición recomendada para una experiencia de juego en línea sin problemas.

Al comparar los resultados año tras año del segundo trimestre de 2019, notamos aumentos en la latencia de banda ancha fija en casi todas las ciudades en el segundo trimestre de 2020. Cali fue la única ciudad de la lista en la que la latencia mejoró, desde 49 en el segundo trimestre de 2019 a 45 ms en el segundo trimestre de 2020.

São Paulo y Buenos Aires empataron con la latencia móvil más baja de la lista durante el segundo trimestre de 2020 con 31 ms. Río de Janeiro siguido con 34 ms. Brasilia, Cali y Medellín ocuparon los últimos tres puestos de la lista con latencias móviles de 43 ms, 48 ms y 51 ms, respectivamente.

La latencia móvil de Ciudad de México y São Paulo mejoró notablemente en el segundo trimestre de 2020 en comparación con el segundo trimestre de 2019. La latencia móvil de Ciudad de México cayó de 53 ms a 37 ms durante el mismo período. La latencia móvil en São Paulo cayó de 40 ms en el segundo trimestre de 2019 a 31 ms en el segundo trimestre de 2020. Solo una ciudad de nuestra lista no demostró mejoras en la latencia móvil al comparar el segundo trimestre de 2019 con el segundo trimestre de 2020, Santiago, donde la latencia móvil aumentó de 36 ms a 40 ms durante este período.

Los jugadores experimentados les gustará saber que las velocidades de internet y la latencia móvil están mejorando en muchas partes de América Latina. Deseamos ver cómo la latencia de banda ancha fija pasa al próximo nivel en un futuro cercano. Si desea obtener más información sobre el rendimiento de Internet en otras partes del mundo, visite Ookla Research^TM.

Explorando a experiência de jogos online na América Latina

Os jogos online estão em alta na América Latina e no mundo inteiro. Isso está criando uma demanda por conexões de rede de baixa latência para que os jogadores possam desfrutar de uma experiência de jogo sem problemas nem atrasos. De acordo com dados da Speedtest Intelligence®, durante o segundo trimestre de 2020, a latência variou bastante entre países e cidades nas cinco economias mais fortes da América Latina, dando a alguns jogadores uma vantagem inesperada. Continue lendo para saber mais sobre a latência de banda larga móvel e fixa na Argentina, Brasil, Chile, Colômbia e México e ver em que país e em quais das cidades mais populosas os usuários provavelmente terão a melhor experiência de jogo.

A latência, ou o tempo de reação de uma conexão, é uma métrica fundamental nos jogos. Para ter a melhor experiência de jogo, os usuários devem ver a latência de 59 ms ou menos. Os dados abaixo mostram a latência média de cada local. A latência móvel é aplicável a jogos em dispositivos móveis, enquanto a latência de banda larga fixa mede a latência ao jogar no PC e em consoles.

O Brasil teve a menor latência de banda larga fixa e a Argentina teve a menor latência móvel

Ookla_Latin_America_latency_countries_0720_pt
Os jogadores no Brasil tiveram a menor latência média de banda larga fixa durante o segundo trimestre de 2020, com 19 ms. O Brasil foi seguido por Chile, México e Argentina. A Colômbia teve a maior latência de banda larga fixa durante esse período, com 43 ms.

Investimentos recentes em fibra em toda a região resultaram em latências aprimoradas ao comparar o segundo trimestre de 2019 e o segundo trimestre de 2020. O Brasil mostrou a maior melhoria na latência de banda larga fixa em comparação com o segundo trimestre de 2019, diminuindo de 23 ms para 19 ms no segundo trimestre de 2020. A latência no México também diminuiu, passando de 34 ms no segundo trimestre de 2019 para 31 ms no segundo trimestre de 2020. Os outros três países da nossa lista tiveram pouca ou nenhuma alteração na latência de banda larga fixa entre o segundo trimestre de 2019 e o segundo trimestre de 2020.

Os países tiveram menos variação na latência móvel durante o segundo trimestre de 2020. A Argentina teve a melhor latência em dispositivos móveis, com 40 ms, seguida de perto pelo Chile, com 41 ms. A Colômbia teve a maior latência móvel durante esse período, com 47 ms.

No segundo trimestre de 2020, o México conseguiu a maior melhoria na latência móvel em comparação com o segundo trimestre de 2019. A latência móvel melhorou de 57 ms no segundo trimestre de 2019 para 44 ms no segundo trimestre de 2020. A latência móvel dos outros países da lista melhorou um pouco menos durante esse período. No Chile, a latência móvel inclusive aumentou de um ano para outro, de 38 ms no segundo trimestre de 2019 para 41 ms no segundo trimestre de 2020.

As implementações de 5G na região começaram recentemente, incluindo o lançamento da tecnologia 5G DSS no Brasil. Embora a tecnologia ainda seja relativamente nova na América Latina, ela tem o potencial de melhorar radicalmente a latência sobre o que os dispositivos com capacidade 4G oferecem atualmente.

A velocidade da Internet aumentou durante a pandemia

Ter uma conexão rápida com a Internet também é importante para os jogadores. Usamos dados da Speedtest Intelligence para acompanhar o desempenho da Internet em nível global durante a pandemia. Embora alguns países tenham tido uma queda nas velocidades em março, no geral, as velocidades de Internet de banda larga fixa aumentaram na Argentina, no Brasil, no Chile, na Colômbia e no México desde a semana base de 2 de março de 2020. Exceto no Chile, esses países também tiveram um aumento nas velocidades em dispositivos móveis, variando de um aumento de 2% na Colômbia até um aumento de 19% no México.

São Paulo teve a menor latência de banda larga fixa, enquanto Buenos Aires em dispositivos móveis

Ookla_Latin_America_latency_cities_0720_pt2
A latência de banda larga fixa em algumas das maiores cidades da América Latina variou de 17 ms em Guadalajara a 45 ms em Cali. Guadalajara, São Paulo e Rio de Janeiro tiveram as menores latências de banda larga fixa da nossa lista durante o segundo trimestre de 2020, com 17 ms, 18 ms e 20 ms, respectivamente. As cidades colombianas de Bogotá, Medellín e Cali apresentaram as maiores latências em banda larga fixa em nossa lista, com 38 ms, 40 ms e 45 ms, respectivamente. Todas as cidades desta lista ainda apresentaram latências em banda larga fixa abaixo de 59 ms, a medida recomendada para uma experiência de jogo sem problemas.

Ao comparar os resultados ano a ano para o segundo trimestre de 2019, vimos aumentos na latência de banda larga fixa em quase todas as cidades no segundo trimestre de 2020. Cali foi a única cidade da lista em que a latência melhorou, passando de 49 ms no segundo trimestre de 2019 para 45 ms no segundo trimestre de 2020.

São Paulo e Buenos Aires empataram na menor latência móvel da nossa lista durante o segundo trimestre de 2020, com 31 ms. Rio de Janeiro em seguida, com 35 ms. Brasília, Cali e Medellín ocuparam os três últimos lugares da lista, com latências móveis de 43 ms, 48 ms e 51 ms, respectivamente.

As latências móveis da Cidade do México e de Guadalajara melhoraram significativamente no segundo trimestre de 2020, em comparação com o segundo trimestre de 2019. A latência móvel na Cidade do México diminuiu de 53 ms para 37 ms no mesmo período. A latência móvel em São Paulo caiu de 40 ms no segundo trimestre de 2019 para 31 ms no segundo trimestre de 2020. Apenas uma cidade da nossa lista não mostrou melhoria na latência móvel na comparação entre o segundo trimestre de 2019 e o segundo trimestre de 2020: Santiago. Lá, a latência móvel aumentou de 36 ms para 40 ms durante esse período.

Quem gosta de jogos eletrônicos na América Latina ficará muito feliz em saber que as velocidades da Internet e a latência móvel estão melhorando em muitas partes da região. Estamos ansiosos para ver se a latência de banda larga fixa alcançará outro nível em um futuro próximo. Se você estiver interessado em aprender mais sobre o desempenho da Internet em outras partes do mundo, visite Ookla Research^TM.

About the Author

Maria Gómez

| More Articles

Ookla Research™ Articles

Maria Gómez | August 12, 2020

Mobile Speeds Lead in Southern Africa

In many parts of the world, consumers are used to fixed broadband speeds that vastly outpace those on mobile. Not so in Southern Africa where the infrastructure challenges associated with improving fixed-line broadband have often made mobile a more attractive option for consumers. Today we are exploring mean and median download speeds on both mobile and fixed broadband in Botswana, Eswatini, Lesotho, Namibia and South Africa during Q2 2020 using Speedtest Intelligence®.

This article uses two measures to describe “typical” internet performance: mean and median. Mean, which people are often most familiar with, is calculated by adding all data points together and dividing by the number of data points. Median is the middle value in a series. Approximately half of the observed values are lower than the median, and half are greater. Both mean and median have advantages and disadvantages. Because median is less impacted by rare instances of values that are extremely high or low, it is more “robust” than mean at representing the data that have already been seen. Mean, on the other hand, is much more sensitive to high and low values. Because its value can change more quickly, the mean can be more informative about what values to expect in the future. Importantly, while mean and median are both informative on their own, viewing them in combination can reveal how “skewed” the data are. With Speedtest® data, this can indicate when certain customers are receiving service that is notably fast or slow among other things.

South Africa had fastest speeds on fixed broadband

Ookla_Southern-Africa_fixed_downloads_0820
Mean download speeds over fixed broadband varied greatly between Southern African countries during Q2 2020. South Africa showed the fastest mean download speeds on our list at 34.19 Mbps, followed by Lesotho at 22.50 Mbps. Botswana had the slowest mean download speed at 8.24 Mbps, 75.3% slower than South Africa.

The differences between countries are much smaller when comparing median speeds, indicating that a smaller number of faster connections is bringing up the mean. This effect was least obvious in Botswana. Median speeds on fixed broadband ranged from a high of 17.28 Mbps in South Africa to a low of 3.49 Mbps in Eswatini. Lesotho followed South Africa closely with 13.18 Mbps, while Botswana almost tied with Eswatini at the bottom of the list at 3.60 Mbps. This leaves Eswatini with a very large skew between mean and median download speed, showing that there is a large difference between the speeds consumers experience. The smaller gap between median and mean speeds in South Africa shows that customers are more likely overall to have faster service.

South Africa had fastest mean download speeds on mobile

Ookla_Southern-Africa_mobile_downloads_0820
The disparity in mean download speed on mobile was less severe during Q2 2020 than we saw on fixed broadband, ranging from 31.11 Mbps in South Africa to 14.93 Mbps in Namibia. Namibia’s mean download speed on mobile broadband was 52.0% slower than South Africa’s. South Africa is also the only country on this list with commercially available 5G, according to the Ookla 5G Map™. 5G will have a much larger impact on mean speeds than on median.

Median download speeds on mobile broadband compared much more closely with mean download speeds than we saw when we were looking at fixed broadband. This indicates that more consumers are experiencing similar speeds without the top-tier outliers we saw on fixed broadband. Lesotho had the fastest median download speed on mobile at 22.75 Mbps and was followed closely by South Africa at 19.82 Mbps. Namibia and Botswana had the slowest median download speeds on mobile at 10.52 Mbps and 10.22 Mbps, respectively.

Mean and median mobile download speeds were closest to equal in Lesotho, indicating their mobile service is uniformly strong across the country and consumers experience relatively similar speeds. Lesotho’s smaller footprint is advantageous to mobile operators because there is less area to cover with mobile service, but the mountainous nature of the country often provides an infrastructure challenge that operators appear to have risen to.

Median download speeds on mobile were faster than on fixed broadband throughout Southern Africa

Ookla_Southern-Africa_mobile-vs-fixed_downloads_0820
In Q2 2020, there was a considerable difference in median download speeds between mobile and fixed broadband in all Southern African countries. Most notably, Eswatini’s median download speed on mobile was 352.7% faster than its median download speed on fixed broadband. With the exception of South Africa, all countries in Southern Africa also showed higher mean download speeds on mobile than those on fixed.

Lesotho led 4G Availability in Southern Africa

Ookla_Southern-Africa_4G-availability_0820-1
We used data from Speedtest coverage scans on Android to compare 4G Availability across Southern Africa during Q2 2020. 4G Availability ranged between a high of 93.7% in Lesotho and a low of 57.3% in Namibia.

We look forward to seeing if fixed broadband speeds improve across Southern Africa in the near future or if Southern Africa is showing a path for continued mobile dominance. To read more about internet performance in other markets, visit Ookla Research™.

About the Author

Maria Gómez

| More Articles

Ookla Research™ Articles

Maria Gómez | September 8, 2020

Exploring the Relationship Between Network Performance and NPS in Taiwan

中文

Network performance is a key element of customer experience, but it may not be the only deciding factor in how a customer perceives their mobile operator and their service. Using data from Speedtest Intelligence®, we compared performance for mobile network operators (MNOs) in Taiwan during Q2 2020. We also compared provider performance to Speedtest Consumer Sentiment™ data on five-star ratings and Net Promoter Score (NPS) to understand how network performance impacts customer satisfaction.

Chunghwa Telecom was fastest in Taiwan

Mobile performance in Taiwan has improved at the country level over the last year. Mean download speed on mobile reached 42.81 Mbps in Q2 2019 and increased 9.8% to 46.99 Mbps in Q2 2020. For context, Taiwan ranked 18th in the world for mobile download speed in July 2020 according to the Speedtest Global Index™.
Taiwan_Speed_Score_Q2_2020_en

We compared mobile performance for top providers in Taiwan using Speed Score™ and found Chunghwa Telecom was the fastest provider in Taiwan in Q2 2020 with a Speed Score of 53.98 on modern chipsets. Chunghwa Telecom was followed by FarEasTone and Taiwan Mobile. The three fastest providers on the list have also deployed 5G in different areas of Taiwan. TSTAR was fourth on our list with a Speed Score of 31.68, while GT was fifth with 19.85.

Chunghwa Telecom was highest-rated mobile provider in Q2 2020

Speedtest Consumer Sentiment data is gathered from single-question surveys presented to users at the end of a Speedtest®. This data set provides rich insights into customer satisfaction over time, as well as competitive benchmarking, by providing data on both customers’ overall satisfaction with their network providers (based on a five-star scale) and Net Promoter Score (NPS). By comparing Q2 2020 ratings data from Taiwan’s top providers with their Speed Score during the same period, we can explore the relationship between customer satisfaction and network performance. This article reflects the ratings and NPS of Speedtest users on modern chipsets and does not imply that the results would be the same when looking at Consumer Sentiment data from users across all technologies.
Taiwan_Speed_Score_Rating_Q2_2020_en

Taiwan’s fastest providers during Q2 2020 also had the highest Consumer Sentiment ratings. Chunghwa Telecom was the highest-rated provider (in addition to being the fastest on our list), while FarEasTone followed with a rating of 3.1 stars. Taiwan Mobile, TSTAR and GT all had an average rating of 2.7 stars, despite a wide difference in Speed Score.

Chunghwa Telecom had highest NPS

Speedtest customers are also asked how likely they are to recommend their provider to friends or family on a 0 to 10 scale. We compared the resulting Net Promoter Score (NPS) with Speed Score to see how performance relates to a customer’s likelihood of recommending a provider. NPS ratings are categorized into Detractors (score 0-6), Passives (score 7-8), and Promoters (score 9-10). NPS is calculated as (% Promoters – % Detractors) x 100. Any NPS score above 0 indicates that a provider’s audience is more loyal than not.
Taiwan_Speed_Score_NPS_Q2_2020_en-1

The ranking of top providers by NPS revealed a different order than we saw with performance and ratings. All of Taiwan’s top providers showed a negative NPS, indicating that fewer customers were likely to recommend their operator’s service to friends and family. Chunghwa Telecom and FarEasTone were ranked first and second for NPS, respectively, which aligned with the rankings for network performance. However, GT was third in NPS, but was ranked fifth for performance, and Taiwan Mobile ranked fifth for NPS and third in performance.

A provider’s performance, network availability and quality can all impact a customer’s overall satisfaction. In Taiwan, top-performing operators had the highest ratings and NPS, but ratings and NPS for operators with lower performance indicated that other factors may also shape customers’ perspectives. We look forward to seeing how network speeds improve as Taiwanese providers continue rolling out 5G, and we will continue to monitor Consumer Sentiment data for the region. To read more about Consumer Sentiment and how to track and benchmark customer satisfaction over time, click here.

探索在台灣網路效能與 NPS 之間的關係

網路效能是客戶經驗的關鍵要素，但是就客戶對於行動電信業者及其服務水準的看法而言，這並不是唯一的決定因素。藉由 Speedtest Intelligence® 提供的資料，我們針對台灣 2020 年第 2 季的行動網路業者 (MNO) 的效能進行了比較。同時，我們還在五星級評等和淨推薦分數 (NPS) 方面，將服務供應商效能與 Speedtest Consumer Sentiment™ 相關資料進行了比較，以瞭解網路效能對於客戶滿意度所造成的影響。

中華電信是台灣速度最快的電信公司

在過去的一年裡，台灣的行動效能已經獲得國家級的改善。在 2019 年第 2 季，行動平均下載速度達到 42.81 Mbps，而在 2020 年第 2 季則增加 9.8%，達到 46.99 Mbps。另外，根據全球網速排行榜™ 的資料，在 2020 年 7 月，台灣的行動下載速度排名為全球第 18 名。
Taiwan_Speed_Score_Q2_2020_tw-1

我們使用 Speed Score™ 針對台灣地區頂尖服務供應商的行動效能進行比較，發現中華電信是 2020 年第 2 季台灣地區速度最快的供應商，其現代晶片組的速度評分為 53.98。在中華電信之後，遠傳電信和台灣大哥大分居二、三名。名單上速度排名前三的供應商也已經在台灣各區域部署 5G。台灣之星在名單上位列第四，速度評分為 31.68，而亞太電信速度評分為 19.85，排名第五。

中華電信是2020 年第 2 季評分最高的行動服務供應商

我們在 Speedtest® 結束時向使用者呈現單一問題調查問卷，並透過該問卷收集 Speedtest 消費者情緒資料。本資料集藉由提供有關客戶對於其網路服務供應商的整體滿意度（以五星級評等）和淨推薦分數 (NPS) 的資料，提供關於客戶滿意度隨著時間變化而不斷變化的豐富見解以及競爭性基準。透過在 2020 年第 2 季比較台灣頂尖服務供應商在同一時期的速度評等，我們能夠探究客戶滿意度於網路效能之間的關係。本文章反映的是 Speedtest 使用者對於現代晶片組的評等與 NPS，並未暗示倘若查看所有技術領域使用者的消費者信心資料時，其結果仍會相同。
Taiwan_Speed_Score_Rating_Q2_2020_tw-1

在 2020 年第 2 季，台灣速度最快的服務供應商同時也獲得最高的消費者信心評等。中華電信是評等最高的供應商（同時也是名單上速度最快的供應商），而遠傳電信緊跟在後，獲得 3.1 顆星的評等。台灣大哥大、台灣之星以及亞太電信三者儘管在速度評分數差異極大，但是皆獲得 2.7 顆星的評分。

中華電信 NPS 最高

我們詢問 Speedtest 客戶是否願意向親友推薦他們的供應商，並要求提供 0 到 10 的分數。我們將淨推薦分數 (NPS) 結果與速度評分進行比較，以瞭解效能與客戶推薦供應商之意願度之間的關聯。NPS 評分歸類為貶低者（0-6 分）、消極者（7-8 分）以及推薦者（9-10 分）。NPS 的計算方式為 (% 推薦者 – % 貶低者) x 100。NPS 分數超過 0 即表示在服務供應商的受訪對象中，忠誠者超過不忠誠者。
Taiwan_Speed_Score_NPS_Q2_2020_tw-2

按 NPS 所得出的頂尖服務供應商排名順序與按效能及評等所得出的排名順序不同。台灣所有頂尖服務供應商都出現負值的 NPS，表示有較少的客戶願意向其親友推薦自己使用的電信業者服務。中華電信和遠傳電信在 NPS 的排名分居一二，此排名也與網路效能的排名一致。但是亞太電信在 NPS 的排名為第三，而在效能的排名卻是第五；台灣大哥大的 NPS 排名為第五，效能排名則是第三。

服務供應商的效能、網路可用性與品質都會影響客戶整體的滿意度。在台灣，效能最佳的電信業者也擁有最高的評等和 NPS，但效能較低的電信業者的評等與 NPS 卻顯示還有其他因素也可能會影響客戶的看法。我們期望台灣的服務供應商在推廣 5G 服務的過程中也能夠提升網路速度，我們也會持續監測此地區的消費者信心資料。若要閱讀更多有關消費者信心以及如何長期追蹤客戶滿意度並設立基準，請按此處。

About the Author

Maria Gómez

| More Articles

How Boingo Leverages Ookla's Comprehensive Network Intelligence

Ookla Joins Ericsson's Leading Telecommunications Open RApp System

How T-Mobile’s Merger with Sprint is Changing the Game for 5G

Get the latest Ookla Research™

Share

The Philadelphia experiment

T-Mobile’s layer cake in Manhattan

Field testing the layers of 5G

LAA boosts speeds

What each NR layer contributes to download speeds

Millimeter wave adds capacity

Device limitations (and opportunities)

How unique device features boost performance

What the future holds

Standalone 5G should improve performance and coverage

About the Author

Milan Milanović

Rakuten is Leveraging O-RAN to Reimagine Mobile and it Could Change the Way We Use Our Phones Forever

Get the latest Ookla Research™

Share

Rakuten disrupts a well-established market

Rakuten leverages the cloud to break the chains of infrastructure

The innovations behind O-RAN

O-RAN opens the door to new solutions

Improving network security and agility

Allowing applications to program the network

How Rakuten made the world’s first Open Radio Access Network

Rakuten Mobile is performing well so far

Median Speeds for Rakuten Mobile in Tokyo

Speedtest Intelligence® | Q2-Q3 2020

What operators can learn from the Rakuten model

楽天がO-RANの活用でモバイルを一新、携帯電話の今後のありかたをまったく変える可能性を示す

楽天が確立された市場を打破

楽天はクラウドを活用してインフラストラクチャの鎖を解く

O-RANの背景にあるイノベーション

新たなソリューションへの扉を開くO-RAN

ネットワークのセキュリティと俊敏性を向上

アプリケーションによるネットワークのプログラミングを実現

楽天はいかにして世界初のオープン無線アクセスネットワークを作り上げたか

楽天モバイルのパフォーマンスは順調

東京における楽天モバイルの中央速度

Speedtest Intelligence® | 2020年第2四半期から第3四半期

事業者が楽天モデルから学べること

About the Author

Milan Milanović

Borrowed Networks: National Roaming in Mexico

Get the latest Ookla Research™

Share

Roaming is common in Mexico, but differs widely by operator and location

AT&T shows the most Time Spent on 4G on and off network

Telcel was the fastest operator for on network use

Redes prestadas: Itinerancia nacional en México

La itinerancia es común en México, pero difiere ampliamente por operador y ubicación

AT&T exhibe la mayor cantidad de tiempo de permanencia en 4G dentro y fuera de la red

Telcel fue el operador más rápido

About the Author

Maria Gómez

What Went Down? The Most Significant Online Service Outages in Q4 2019

Get the latest Ookla Research™

Share

Gaming

Discord (December 7, 2019): 15,976 outage reports at peak

Fortnite (October 13, 2019): 11,326 outage reports at peak

Mobile Operators

Vodafone Germany (October 23, 2019): 21,065 outage reports at peak

Social Media Sites

Facebook (November 28, 2019): 12,726 outage reports at peak

Instagram (November 28, 2019): 21,682 outage reports at peak

Snapchat (October 14, 2019): 18,252 outage reports at peak

Twitter (October 22, 2019): 15,952 outage reports at peak

Streaming Services

Hulu (December 19, 2019): 25,777 outage reports at peak

Disney+ (November 12, 2019): 8,441 outage reports at peak

Financial Institutions

ING (November 17,2019): 1,342 outage reports at peak

About the Author

Maria Gómez

How Roaming Affects Mobile Speeds in Europe

Get the latest Ookla Research™

Share

Speedtest Intelligence^® | Q2-Q3 2020

Speedtest Intelligence^® | 2020年第2四半期から第3四半期

Speedtest^® Data | Q3-Q4 2019

Speedtest^® Data | Q3-Q4 2019

Speedtest^® Data | Q3-Q4 2019

Speedtest Intelligence^® | Q3-Q4 2019

Speedtest Intelligence^® | Q3-Q4 2019

Speedtest Intelligence^® | Q3-Q4 2019

Speedtest Intelligence^® | Q3-Q4 2019

Speedtest Intelligence^® | Q3-Q4 2019

Speedtest Intelligence^® | Q3-Q4 2019

Speedtest Intelligence^® | Q3-Q4 2019

Speedtest Intelligence^® | Q3-Q4 2019

Speedtest Intelligence^® | Q3-Q4 2019

Speedtest Intelligence^® | Segundo semestre de 2019

Speedtest Intelligence^® | Segundo semestre de 2019

Speedtest Intelligence^® | Segundo semestre de 2019

Speedtest Intelligence^® | Segundo semestre de 2019

Speedtest Intelligence^® | Segundo semestre de 2019

Speedtest Intelligence^® | Segundo semestre de 2019

Speedtest Intelligence^® | Segundo semestre de 2019