THE 2018 MOBILE REVIEW FOR SINGAPORE
←
→
Page content transcription
If your browser does not render page correctly, please read the page content below
MOBILE MOBILE REVIEW BENCHMARK The XXXX The Mobile 2018 Network Mobile Test in REVIEW the FOR SINGAPORE XXXXXXXX For the first time, the network bench- marking expert P3 and connect magazine have tested the mobile Thanks to strict regulatory require- networks in the city-state of Singapore. ments, the network quality of all We have conducted a drivetest as Singapore operators is very high. well as a crowd based analysis over This makes it very intersting to compare six months resulting in the following their performance with that of other comprehensive Mobile Review. leading world capitals. Also, we have taken a detailed look into the perfor- mance of each Singapore operator.
SINGAPORE All three Singapore opera- CROWD tors d eliver a high level of data performance while ranking very closely. However, in the drivetest DRIVETEST results, Singtel is s lightly ahead, The crowd results are good o verall, but show a slight lead of M1 in some of the KPIs. Q2 Q3 Q2 Q3 Q2 Q3 94,2 95,3 95,3 94,0 93,7 95,1 93,2 96,6 95,3 Operator Singtel Starhub M1 Shown scores in % Results in a nutshell DRIVE- CROWD- The Singapore market is known for its high network quality. Thus, it is TEST SOURCING not surprising that the three operators considered in this review rank FACTS FACTS closely. And still, our holistic network testing approach, comprised of a data-centric drivetest and sophisticated crowdsourcing analyses, reveals specific strenghts of each operator. As this Mobile Review for the net- the fastest file downloads, while 99.5% 19,257 works in Singapore does not use StarHub scores best in the of population users the full methodology of our regular YouTube measurements. covered Network Benchmarks, we abstai- The extensive crowdsourcing ned from ranking or g rading the analysis, based on 344 million operators. However, the r esults samples collected in two con- of our data-centric drivetest and secutive 3-month-periods from crowdsourcing analysis provides April to June and July to Sep- a good indication of their achieve tember 2018, overall validates ments. The drivetest result shows these findings. The results show the maximum performance of the networks, whereas the crowd the high performance level of all three networks. The quality of 4G 31,003 344 evaluation reflects the actual user coverage was a little lower in the samples million samples experience. first evaluation period. Singtel The drivetest results confirm and StarHub improved regarding that all three Singapore opera- this KPI in the s econd period, tors deliver a high level of data only M1 fell a little behind. On performance and rank closely. the other hand, the average data However, in our drivetest, Singtel rates dropped in all networks in shows a slight advantage over the second observation period. its competitors. In the detailed analysis, each operator demons- In some of the KPIs, M1 showed a slight advantage. The e xamination 2,100 97.8% trates specific strenghts: M1 is of the Data Service Availability km drivetest of the built-up ahead in the web browsing tests showed minor degradations in the area and also takes a narrow lead in StarHub network in March and in the file uploads. Singtel provides the M1 network in September. MOBILE REVIEW 2
operators in SingaPORE According to the Singapore Info-Communications Media Development Authority’s (IMDA) statistics, S ingapore’s mobile market has a penetration rate of almost 150 per cent. A total of 5.6 million inhabitants hold about 8.5 million mobile subscriptions – comprising of roughly 60 per cent postpaid and 40 per cent prepaid subscriptions. Singapore Telecommunica- In the course of the libera- M1 Limited (previously tions Limited (Singtel) was lisation of the Singapore known as MobileOne) was founded in 1992 and today telecommunications mar- formed in 1994, launched employs more than 23,000 ket, StarHub was awarded its mobile service in 1997 staff worldwide. Headquar- the licence to provide fixed and became a p ublicly tered in Singapore, Singtel network and mobile servi- listed company at the end is also the parent company ces in 1998. The c ompany of 2002. M1 now employs of the renowned Australian was officially launched in over 1,500 people. The operator Optus. April 2000 and has been company prides itself to Additionally, Singtel is listed on the Singapore have launched South East invested in leading com- stock exchange since 2004. Asia’s first commercial LTE panies in Asia and A frica, StarHub currently counts network in June 2011 and including Bharti Airtel about 2,500 employees. to have become the first (India, South Asia and For its fiscal year 2017, operator in South East Asia Africa), Telkomsel (Indo- the Group published a to offer nationwide 4G ser- nesia), Globe Telecom (the mobile revenue of 1.2 vices from September 2012. Philippines) and Advanced billion Singapore Dollars For its fiscal year 2017, Info Service (Thailand). (about 0.9 billion US Dol- M1 published an opera- With an operating revenue lars). When also taking ting revenue of 1.1 billion of 17.5 billion Singapore Pay TV, broadband, fixed Singapore Dollars (about Dollars (about 12.5 billion line and other revenues 0.8 b illion US Dollars). US Dollars) in the fiscal into a ccount, the company Additionally, the company year 2018, Singtel is one of achieved a total revenue specified a fixed services the largest telecommuni of 2.4 billion Singapore revenue of 130 million cations companies in Asia. Dollars (about 1.7 billion Singapore Dollars (about The group specifies over US Dollars) in 2017. 94 m illion US Dollars). 650 million customers in For Singapore, StarHub At the end of 2017, M1 21 countries. In Singapore, specifies 2.3 million mobile had a total of 2.2 m illion Singtel has 4.1 million subscribers which equals mobile customers and mobile subscribers, which a mobile market share of 189,000 fi bre subscribers. equals a share in the about 27 per cent. Its mobile customer base mobile market of about 48 equals a market share of per cent. about 26 per cent. MOBILE REVIEW 3
This year, P3 and connect have conducted a Mobile Review in Singapore for the first time. This review just marks the start – further, even more complex investigations of the network quality applying more KPIs are planned in the future. A STUDY INTO the Singapore mobile market P3, headquartered in Aachen, Germany, is a world leader in mobile network testing. The company has over 3,500 employees worldwide and a t urnover of more than 350 million Euros. P3 is partnering with the international telecommu- nications magazine connect, which has more than 20 years of e ditorial ex- pertise and is one of the leading test authorities in Europe for telecommuni- cations products and services. T ogether, P3 and connect h ave been con- ducting the most important network b enchmark test in Germany for more than 15 years, extending it to other E uropean countries since 2009. Above that, P3 has been conducting its authoritative Mobile Benchmarking in A ustralia Singapore‘s mobile operators since 2014 where it is well known and respected. P3 and connect also exa are among the best in a global mined many other mobile networks all over the world including those the USA. comparison with very smart re- In 2017 alone, P3 compiled more than 60,000 measurement hours sults. Our review confirms that in more than 80 countries across five continents, with its test vehicles all three Singapore operators de- covering almost one million kilometres. Recently, P3 has expanded its liver high quality while Singtel is testing scope by amending the drivetests and walktests with a sophi- showing a slight advantage over its competitors. Our crowd-based sticated and statistically highly reliable crowdsourcing approach. As the benchmarking also reveals that de-facto industry standard, the P3 benchmarking methodology focuses particularly Sydney, Australia, on customer-perceived network quality. P3‘s network benchmarks are achieves comparably high data widely accepted as a completely objective authority. throughputs. Therefore, we are looking forward to the certainly very interesting results of our drivetest and crowdsourcing reflect high upcoming Mobile Benchmark in Australia.“ Performance of THE Singapore networks Hakan Ekmen, In 2018, P3 and connect have extended the scope of their network CEO of P3 communications assessments to the city-state of Singapore. This market is p articularly interesting with its very large market penetration of almost 150 per cent (see the previous page) as well as its very high standard of network quality which is also e nsured by strict regulatory requirements. This year, P3 and connect have started their qualified look into the Singapore mobile market by mainly concentrating on the data quality of the networks – t aking into account the typical usage patterns of m obile subscribers as well as the importance of mobile data availability and performance for future applications such as IoT or automotive c onnectivity. To achieve this, P3 have conducted a data-centric drivetest in May 2018. In order to validate and supplement its results and analyse trends in the development of the networks‘ performance, we have also p erformed an extensive crowdsourcing analysis covering two periods of three months each in order to be able to also. The results of both approaches clearly reflected the high performance level of the mobile networks in Singapore. MOBILE REVIEW 4
SINGTEL IS Data drivetests CLOSELY AHEAD OF A FOR THIS MOBILE REVIEW, WE HAVE CONDUCTED A DATA-CENTRIC VERY STRONG DRIVETEST, taking into account the typical usage patterns of mobile subscribers as well as the importance of mobile COMPETITION data availability and performance for future applications such as IoT or automotive connectivity. In order to assess the performance and reliability of data con- nections in the Singapore networks, two d rivetest cars drove a total of 2,100 k ilometres, c overing 99.5 per cent of Singapore‘s population. Each of P3‘s drivetest cars was equipped with arrays of Galaxy S8 Smartphones capable of supporting 300 Mbit/s download and 50 Mbit/s upload speeds. All data measure- ments were done in 4G p referred mode. Each car carried one smartphone per operator. P3‘s drivetesting considers fast throughputs as well as the networks‘ availability and stability. Web page and file down loads or file uploads reward fast speeds, while recording success ratios and assessing YouTube playouts concentrate on reliability a spects. As YouTube streams videos at adaptive bitrates, the average value of the received video resolution is another important performance indicator. Overall high data performance Data in Cies - Drivetest Singtel Starhub M1 The drivetests show the high Web-Page Download (Live/Sta c) performance level of all three mobile Success Ra o (%) 99,6 99,8 99,9 operators in Singapore. Determining Time to First Byte (ms) 506 425 419 their overall scores for the data drive- tests, all competitors are close KBytes First Second (kB) 566 561 538 together: Singtel achieves 94.2 per HTTP Sta c Page DL Qualifier (%) 99,8 97,7 98,8 cent of the total possible points, HTTP Sta c Page DL Overall Session Time (s) 1,2 1,1 1,0 StarHub follows with 94.0 per cent File Download (3 MB) and M1 with 93.2 per cent. Although Success Ra o/Avg. Session Time (%/s) 99,8/1,0 99,8/1,2 100,0/2,2 we see some differences in their 90%/10% faster than (kbit/s) 15938/72948 15669/45810 9073/46083 detailed performances, this is a very File Upload (1 MB) good result for all three of them. Success Ra o/Avg. Session Time (%/s) 99,5/0,7 99,7/0,6 99,7/0,7 90%/10% faster than (kbit/s) 10012/ 28369 11512/ 27397 10599/ 31873 File Download (7 Seconds) Success Ra o (%) 99,7 99,8 100,0 Avg. Throughput (kbit/s) 66526 60963 39350 90%/10% faster than (kbit/s) 19927/128676 20433/113872 14370/70779 File Upload (7 Seconds) Success Ra o (%) 100,0 100,0 99,5 Avg. Throughput (kbit/s) 25708 27293 36076 90%/10% faster than (kbit/s) 13432/47602 15931/36323 13923/57709 Youtube Video Success Ra o/Start Time (%/s) 99,0/1,8 99,4/1,8 98,9/1,9 Playouts without Interrup ons (%) 99,8% 100,0% 99,7% Average Video Resolu on (p) 1080 1080 1058 MOBILE REVIEW 5
M1 ahead in Web BROWSING IN A CLOSE When it comes to accessing live web pages from the widely RACE, EACH acknowledged Alexa ranking and Web the static Kepler test page, M1 OPERATOR BrowsinG is closely ahead of the other two operators. M1‘s success rates SHOWS are slightly higher and the downlo- ad times slightly shorter. This SPECIFIC results in a fulfillment rate of 96 per cent in this category. STRENgTHS However, Singtel with 94 per cent in THE DATA and StarHub with 93 per cent follow at close distance. CATEGORY SINGTEL PROVIDES THE FASTEST DOWNLOADS, ALL THREE almost FILE ON PAR REGARDING UPLOADS DOWNLOADS In the download tests, Singtel & UPLOADS is one step ahead due to faster average data rates and a higher top speed (P90). StarHub ranks second and M1 third in the down- load evaluation. Examining their upload performance, all three operators are almost on par with a slight advantage for M1 due to slightly higher average upload data rates. Data Results STARHUB SCORES BEST IN At a Glance THE YOUTUBE MEASUREMENTS In our tests of YouTube play- Although all three Singapore ope- backs, StarHub showed a slight rators deliver a high level of data youtube advantage over its competitors. performance and rank closely This operator achieves a 100 per together, in the overall assess cent playout without interruptions ment, Singtel shows a slight (Singtel: 99.8 per cent, M1: advantage over its competitors. 99.7 per cent). Also, the overall In the detailed analysis, each success ratio is slightly higher operator shows specific strenghts: in the StarHub network. Still, M1 is ahead in the web browsing the mobile reception of YouTube tests and is closely ahead of the videos provides a very good competition when it comes to user experience in all three file uploads. Singtel provides Singapore mobile networks. the fastest file downloads, and StarHub scores best in the YouTube measurements. MOBILE REVIEW 6
THE CROWD Crowd SOURCED RESULTS CONFIRM 19,257 users have contributed 344 million samples to our crowd- THE HIGH sourcing analysis of the Singapore networks over two consecutive PERFORMANCE periods spanning three months each. The evaluation area of our crowd- OF THE sourcing represents 97.8 per cent of the built-up area of Singapore. SINGAPORE NETWORKS In order to validate and supplement the results of the drivetest, we have also performed an extensive crowd- sourcing analysis of the mobile market in S ingapore. This analysis is based on crowd data that has been gathered in two periods of three months each. P3 decided to follow this approach because our crowdsourcing metrics are based on three-month periods. As the drivetests Overview Score (April - June) were conducted in May 2018, we wanted to factor in Coverage crowdsourcing data from this period as well — resulting User DL in including the measurement period April to June 2018. Speed On the other hand, we also wanted to present data that Data Service Availability is as current as possible. This is why we also included a second period ranging from July to September 2018. Considering two consecutive three-month crowdsour cing periods allows us to also have a look at trends and developments between the two c ompared periods. The results are based on usage data that have been collected from smartphone users who are utilising one or more of 800+ apps in which P3 has integrated back ground diagnosis processes determining relevant usa- ge data 24/7, 365 days a year. Quarterhourly reports are 93,7 92,0 100,0 91,1 94,7 97,6 95,2 95,0 100,0 generated daily and sent to P3‘s s ervers for a thorough analysis (see detailed description of the methodology Operator Singtel Starhub M1 on p ages 9 and 10). This way, more 19,257 users Shown scores in % have contributed a total of 344 million samples in the described six-month period. Based on the total popu- lation count of 4.6 million people, one of 308 inhabi- tants of Singapore has contributed to our crowd data. Overview Score (July - September) The evaluation area of our crowdsourcing represents Coverage 97.8 per cent of the built-up area of Singapore. User DL Speed Data Service ALL OPERATORS SHOW HIGH PERFORMANCE, BUT THERE Availability ARE STILL SOME VARIATIONS IN THEIR CROWD RESULTS The overall results of our crowd tests confirm the drive- test results in so far as all three Singapore operators show respectable performance. In an o verall assess ment, M1 leads the field with a narrow margin in the April to June period. In the July to S eptember period, Singtel and M1 are on par. StarHub follows at close distance in both cases. A direct comparison of both periods shows that Singtel‘s performance stayed the same, StarHub was able to improve by some percen- 94,8 89,3 100,0 94,5 91,0 98,7 94,8 91,7 98,2 tage points, w hereas M1 lost a couple of percentage Operator Singtel Starhub M1 points. However, these variances are minor all in all. Shown scores in % MOBILE REVIEW 7
Singtel CROWD Starhub User Download Speed M1 April - June ALMOST PERFECT VOICE AND DATA COVERAGE FOR ALL OPERATORS AND ALSO HIGH 4G COVERAGE 160,7 For all Singapore operators, we observed 100 per cent coverage within our test area for both voice and data Mbit/s 134,4 services during both observation periods. Even the de- 10% EA manding 4G coverage is very high with values between 126,2 faster than 97 and 99 per cent. Also, the results for the so called Quality of Coverage (the percentage of actual availability of the mobile network services) were high for voice 30,2 and data. An examination of the Quality of 4G C overage Mbit/s shows that these values were a little lower in the first 32,3 10% Users evaluation period with Singtel and StarHub improving in faster than 33,0 the second period, while M1 received a slightly better result in the first period than in the second one. When it comes to data rates, M1 shows a slight ad 10,6 vantage in the average as well as the top values (P90) Mbit/s of each Users‘ Best Throughput within the observation 11,1 time. These averages were 11.2 Mbit/s in the first Avg. Users Best Throughput evaluation period and 10.5 Mbit/s in the second one. 11,2 StarHub and SingTel each follow at close distance with StarHub achieving 11.1 Mbit/s in the first and Singtel 10.3 Mbit/s in the second period, SingTel‘s results are CROWD Starhub 10.6 Mbit/s in the first and 10.1 Mbit/s in the second User Download Speed M1 observation period. A possible explanation for the July - September overall drop in these values between the April to June and the July to September period could be a higher 156,9 number of users in the networks leading to a higher Mbit/s utilisation of the network cells and their capacities. 132,5 10% EA Stil, SingTel delivers the highest values for the top faster than 126,6 speeds (P90) observed per evaluation area (“EA“). This applies to both observation periods. The difference to the User‘s Best Throughput is that these top speeds 28,7 have been achieved by varying users. Mbit/s Occasional drops especially in some coverage KPIs 29,9 between the first and the second three-month period 10% Users faster than can be explained by user fluctuations – the number of 30,9 users within a particular observation area is typically not constant between evaluation periods. 10,1 SINGTEL SHOWED NO DEGRADATIONs IN 10,3 Mbit/s BOTH CONSIDERED PERIODS Avg. Users In the evaluation for Data Service Availability, also all 10,5 Best Throughput three competitors shows generally pleasing results. Our analysis identified one hour with limited service availabilty in the StarHub network in March and also a two hour degradation in the M1 network in S eptember. CROWD RESULTS Both incidents only had a very limited impact on the AT A GLANCE users‘ experience of network availability and stability. The crowdsourced results for the coverage of voice and data ser- vices is exceptionally high in the Singapore networks. The Quality DATA SERVICE AVAILABILITY of 4G Coverage was a little lower in Affected hours (h) and days (d) (2018) the first evaluation period, SingTel and Starhub improved regarding this KPI in the second one. On the Singtel other hand, the average data rates Starhub 1h over 1d dropped in all netwoks in the se- cond observation period. In some M1 2h over 1d of the KPIs, M1 showed a slight advantage. The examination of March April May June July August September Data Service Availability showed only minor degradations in the StarHub and M1 networks. MOBILE REVIEW 8
How does Singapore compare to other world capitals? As a successful business capital, Singapore is competing with metropolitan areas on a global scale. Therefore we have compared our crowdsourced throughput results gathered in Singapore with those achieved in Berlin, Cape Town, Jakarta, London and Sydney. A look at the average values of In London, only EE achieves with buildings, whereas cities like the best throughputs a chieved by similar results, while the other Berlin or London contain a higher each individual user participating British operators as well as the share of open spaces such as in our crowdsourcing (see chart mobile networks in Berlin, Cape parks or fringe areas, which are below) shows that the Singapore Town and Jakarta overall deliver provided with less mobile co operators deliver considerably slower data rates. When looking verage and also comprise of high performance also when com- at the P90 value (the threshold less mobile users. pared to their global competitors. below which 90 per cent of the Furthermore, in order to e nsure gathered values are ranging statistical relevance, we had HIGHEST OBSERVED DATA RATES – see also the description of to exclude Jakarta‘s smallest IN SINGAPORE AND SYDNEY our methodology on pages 10 operator Smartfren from this and 11), S ydney even shows comparison. M1, StarHub and Singtel provide slightly better overall results than The analysis at hand concen- high data rates to their customers, Singapore. trates on data throughputs. But which in this comparison are only However, when evaluating the an additional look at the Areas matched by the performances of actual values, it must be taken and Quality of Coverage with 4G the Australian operators Telstra, into account that areas such as services, 3G/2G data services as Optus and Vodafone in Sydney. Singapore are densely c overed well as voice services in the consi- dered cities also shows favour Singapore Berlin Cape Town Jakarta London Sydney able results for 12M Singapore and Sydney – how ever, some com petitors such as London or Jakarta come close in one or the other of the analysed metrics. Deeper insights can be made MobileOne Telefónica Telefónica Telkomsel Vodafone Vodafone Vodafone Vodacom INDOSAT Telekom CROWD Starhub Telkom Telstra Singtel available to Optus Three Three Cell C MTN Users‘ Best Throughput EE XL (AVG) interested parties Mbit/s 10,1 10,5 10,3 5,4 6,1 5,6 3,8 4,7 4,7 4,4 4,6 6,7 4,3 5,5 8,0 4,9 5,5 6,8 9,4 10,3 9,3 on request. Shown values are rounded. MOBILE REVIEW 9
Methodology The methodology of the P3 connect Mobile Review is the result of more than 15 years of testing mobile networks. Today, network tests are conducted in more than 80 countries. They were carefully designed to evaluate and objectively compare the performance and service quality of mobile networks from the users’ perspective. The P3 connect Mobile Review for Alexa ranking. In addition, the Singapore includes the r esults of static Kepler test web p age as Boxes were mounted into the rear and an extensive data drivetest as well specified by ETSI (European side windows of each measurement car as a sophisticated crowdsourcing Telecommunications Standards supporting the smartphones used for the drivetests. approach. Institute) was used. In order to test the data s ervice drivetests performance, files of 3 MB and 1 MB The drivetest was conducted on for download and upload w ere based on crowd data that had 12 measurement days from May transferred from or to a test server been gathered in two periods of 2nd to May 16th, 2018. All sam- located on the Internet. In addition, three months each. P3 decided ples were collected between 8am the peak data performance was to follow this approach because and 10pm. Two d rivetest cars tested in uplink and downlink di- our crowdsourcing metrics are drove a total of 2,100 kilometres, rections by assessing the amount based on three-month periods. covering 99.5 per cent of Singa of data that was transferred within As the drivetests were conducted pore‘s total population. a seven seconds time period. in May 2018, we wanted to factor Each of P3‘s drivetest cars was Another discipline was the in crowdsourcing data from this equipped with arrays of Samsung playback of YouTube videos. It period as well — thus, we have Galaxy S8 Smartphones. These took into account that Y ouTube included the months from April to „Cat 9“ phones are capable of dynamically adapts the video June 2018. On the other hand, we supporting 300 Mbit/s download resolution to the available band wanted to present as current data and 50 Mbit/s upload speeds. All width. So, in addition to success as possible, which is why we also data measurements were done in ratios, start times and playouts included the period from July to 4G preferred mode. Each car car- without interruptions, YouTube September 2018. The considera- ried one smartphone per operator. measurements also determined tion of two crowdsourcing periods average video resolution. All tests of a duration of three months each DATA TESTING were conducted with the best allows us to also have a look at For the web tests, the test smart- performing mobile plan available trends and developments between phones accessed web pages ac- from each operator. the two compared periods. cording to the widely recognised For the collection of crowd data, RANKING AND GRADING P3 has integrated a background As this Mobile Review is intended diagnosis processes into 800+ just as an indication of network diverse Android apps. If one of performance and q uality, we ab- these applications is installed on stained from ranking or grading the end-user’s phone, data collec the operators considered based tion takes place 24/7, 365 days a on the number of total achieved year on this device. Reports are points. However, the p ublished generated for every 15 minutes and percentages give a good indica daily sent to P3‘s cloud servers. tion of their achievements. Such reports generate just a small number of bytes per message CROWDSOURCING and do not include any personal As an addition to the drivetests, user data. Interested parties can P3 conducted thorough crowd- deliberately take part in the data All test phones used in the drivetests were operated and supervised by based analyses of the Singapore gathering with the specific ”U get“ P3‘s unique control system. networks. These analyses are app (see box on next page). MOBILE REVIEW 10
Other crowdsourcing solutions relevant statements, P3 requires which 10 per cent of the values have a very technical user base. a certain number of users and are situated. These values depict Thus, their results are typically measurement values per opera- how fast the network is under skewed towards high-end, heavy tor for each tile and each evalua- favourable conditions. data users. With the integration tion area. If these thresholds are into more than 800 diverse apps not met by one of the considered Data service availability covering different market seg- operators, this part of the map is Another performance indicator ments, P3 has generated data not considered in the assessment considered in the crowd results is which is a fair and equal repre to ensure fair terms. the Data Service Availability. This sentation as o pposed to that of Even more relevant results are parameter indicates the availability classical speed test apps. The accomplished by not only de- of a network and the number of unique crowdsourcing t echnology termining the mere network outages or service d egradations allows P3 to collect data about coverage but also considering its respectively. real-world customer experience quality. The parameter Q uality of In order to differentiate net- in a truly passive way – wherever Coverage reveals whether voice work glitches from normal varia- and whenever customers use and data services actually work tions in network coverage, we their smartphones. in the respective evaluation area. apply a precise definition of ser- P3‘s crowdsourcing data set P3 does this because not in each vice degradation: A degradation is the most realistic, s ince it is area that allegedly provides net- is an event where d ata connec- the most diverse that is c urrently work reception, the mobile servi- tivity is impacted by a number of available in the market in terms ces can actually be used. For this cases which significantly exceeds of locations, geography, times, reason, the percentage for Quality the expectation level. To judge devices, subscriptions, n etworks, of Coverage is always a little lower whether an hour of interest is an technologies and s martphone than the corresponding coverage hour with degraded service, the usage patterns. P3 applies value. We specify these values algorithm looks at a sliding win- advanced big data analytics to each for the coverage of voice dow of 168 hours before the hour distill the essence of informa- services (2G, 3G and 4G combi- of interest. This ensures that we tion from the bulk data. By ana ned), Data (3G and 4G combined) only consider actual network ser- lysing data according to prede- and 4G only. vice degradations differentiating fined m etrics, P3 can provide them from a simple loss of net- information for the optimisation of ASSESSMENT of data throughputs work coverage of the respective networks and also show whether Additionally, P3 investigates the smartphone due to prolonged networks live up to the expecta data rates that were a ctually indoor stays or similar reasons. tions of their customers. available to each user. For this In order to ensure the statisti- purpose, we have determined the cal relevance of this approach, a rating of network Coverage best obtained data rate for each valid assessment month must fulfil For the assessment of network user during the e valuation period clearly designated prerequisites: coverage, P3 lays a grid of 2 by and then calculated the a verage A valid assessment hour con- 2 kilometers over the whole test of these values. In addition, we sists of a predefined number of area. The so-called evaluation have determined the so-called samples per hour and per opera- areas generated this way are then P90 values for the top throughput tor. The exact number depends on sub-divided into 16 smaller tiles. of each evaluation area as well as factors like market size and num- In order to ensure statistically of each user‘s best throughput. ber of operators. A valid assess P90 values specify the threshold ment month must include at least in a statistical distribution, below 90 per cent of valid assessment which 90 per cent of the gathe- hours (again per month and per red values are ranging – or above operator). Participate in our crowdsourcing Everybody interested in becoming a part of our global crowdsourcing panel and obtaining insights into the r eliability of the mobile network that her or his smartphone is logged into, can most easily participate by installing and using the “U get“ app. This app exclusively concentrates on network analysis and is available under uget-app.com or via the adjoint QR code. “U get“ checks and visualises the current mobile net- work performance and contributes the results to our crowd- sourcing platform. Join the global community of users who understand their personal wireless performance, while con- tributing to the world’s most comprehensive picture of the mobile customer experience. MOBILE REVIEW 11
You can also read