Just when you need it the most, the internet connection on your mobile phone tends to go kaput. Don’t blame it on Murphy or anyone else, though. Telecom service providers who promise way more than they can deliver might also be at fault.

While Telecom Regulatory Authority of India (TRAI) has benchmarks for quality of service (QoS), it does not have the power to impose penalties if telecom companies violate them. Worse, the regulatory body has to judge these companies by the data the telecom providers themselves provide. What would you say are the chances of them ’fessing up to bad service?

Fortunately, there’s a study to show the gulf between in the QoS that is advertised and submitted to TRAI and the actual QoS available to users. Compiled Indian Institute of Technology Delhi (IIT-D) and Consumer Unity & Trust Society (CUTS), a consumer rights body, this report examines both the statistics as well as user experiences. While IIT-D conducted a study of mobile data connections in 20 locations across Delhi, Jharkhand, Madhya Pradesh and Rajasthan, CUTS carried out a survey of 730 mobile data users in West Bengal, Rajasthan and the National Capital Region (NCR).

In 2013 and 2014, Linux netbooks with three dongles – each of a different telecom provider – were used to study different parameters of mobile data connections, both 2G and, wherever available, 3G. One of the 20 sites was IIT-D itself, and three in Rajasthan (all homes of team members). The other 16 were chosen with the help of non-government organisations – 11 in MP and five in Jharkhand. At each location, three months of continuous data collection was performed using locally-produced SIM cards.

The QoS was evaluated on the following:

• availability, or the fraction of time for which connectivity was available,
• throughput, which is commonly referred to as bandwidth and refers to the amount of data passing through the network in a given time,
• latency, the lag between sending and receiving something.

In case of bothavailability and throughput, the higher the better. Unfortunately, the report notes that in places like Janakpur in MP, availability was as low as 35 per cent across service providers. Predictably, the national capital boasts of impressive figures – over 70 per cent for BSNL (2G), 80 per cent for Airtel (both 2G and 3G) and 90 per cent for Idea (2G). Surprisingly, the figure for Reliance is less than 40 per cent. In fact, Reliance is the only provider that does better in rural areas. While Airtel and BSNL are inconsistent, Idea performs consistently well across the board.

The observations for throughput were similar to those for availability, with Idea outperforming its rivals. However, none of the companies were able to provide throughput figures that were anywhere close to their advertised values. For instance, while Idea 2G boasts of a speed of up to 236kbps, the actual values were found to be in the range of 80-160kbps. Among 3G providers, Airtel fared the best, but again, the observed figures (800kbps – 2.3 Mbps) fell woefully short of the advertised claims (up to 21.1Mbps). The report describes the keywords “up to” as a “misleading sales pitch which does not guarantee any performance”.

Unlike the other two, the lower the latency, the better the QoS. Also known as lag or delay, latency is the amount of time between sending something and receiving it. Low latency is important when it comes to using certain apps like Skype, which can deal with low bandwidth but not lag. In IIT-D’s tests, Idea and Reliance clocked the lowest latency scores.

Worse than not matching performance with advertising, the numbers don’t tally with those submitted by the telecom companies to TRAI. Service providers report values for every cellular circle to the regulatory body. Airtel, for instance, claimed cent per cent availability in all the circles where the IIT-D study was conducted. But the figures observed in Jharkhand were in the range of 42 and 58 per cent.

Aaditeshwar Seth, assistant professor at IIT-D who led the technical analysis, said this is because the companies measure data inside their networks. “The measurement methodology should be made more rigorous,” Seth told Newslaundry. “Data should be collected at the users’ end.” While telecom companies are likely to blame the lack of towers, the report suggests that existing infrastructure is good enough to provide better QoS. “Some providers did markedly better than the others,” said Seth. “Network reconfiguration should be able to improve network stability.”

The customer survey showed that one of two respondents chose a telecom service provider based on the QoS of its mobile internet connection. Barring those from NCR, a bulk of them were unaware of TRAI. Critically, almost everyone unanimously agrees that penalties should be put in place for breaching QoS parameters. Perhaps, like the report suggests, it is time to empower TRAI and punish those who fail to keep their promises.