Device testing – Testing for the future

A report from Broadband-Testing & Mobile Test Labs calls for a change in test methods.  Mobile Europe runs some extracts

All mobile phones have to pass conformance testing for certification and type approval for each country in which they are sold but, as Dean Bubley, analyst with Disruptive Analysis explains later in this report, this is no longer enough.

The problem is that the certification and type approval process does little other than confirm that a mobile handset works at a fundamental level and is compliant with industry-wide certification standards, plus a given set of country-specific requirements. It does not actually give any clues as to how well that handset actually performs.

Meantime, in the US, operators have looked to take matters into their own hands.

It started with Verizon Wireless looking not just to have the best quality network but also the best quality handset range and finding that the set standards – there was no formal certification in the US at the time for handsets that used Verizon?s CDMA technology – were too low to guarantee that a handset performed adequately. Not relying on the industry standards bodies, Verizon set its own bar higher and required handset vendors? products to meet strict performance levels before they would be considered for resale on Verizon?s network. Verizon effectively developed its own handset certification process, using performance as a key criterion.

To compete effectively with Verizon, AT&T – its rival for the title of largest US carrier – then also raised its handset acceptance criteria, testing well beyond the basic requirements of the industry?s certification process. Unlike Verizon, AT&T is a GSM/WCDMA operator, so this led to the bar also being raised for new handsets that use this technology.

And not without good cause… Two well-documented recent cases of issues with major smart phone releases concern the initial launch of the iPhone 3G (since corrected with firmware upgrades, as shown in this report) and AT&T postponing the launch of the Blackberry Bold for several months because of significant issues found during testing. In Europe, Orange actually withdrew the Bold from the market for a while in the autumn of 2008. For handset vendors and operators, the consequences of releasing a handset to the market that simply doesn?t work well enough are enormous, not just in terms of the costs at the time but also in terms of the bad PR and loss of confidence by the general public – its customers – this creates.

Several leading Asian operators have also put handset performance standards requirements in place to qualify phones for resale, but EMEA still lags behind in this respect. As we move towards 4G/LTE, the issue of performance is set to become more important and fundamental than ever. So, wake up Europe, let?s get in line with what is happening in the rest of the world…

Despite the impressive rise of mobile broadband and other forms of wireless data service, ordinary voice telephony still accounts for the bulk of revenues for most cellular operators.

As well as its critical importance financially, voice performance also remains one of the main ways that customers can differentiate between service providers. Dropped calls, poor voice quality and other glitches can affect an individual user's propensity to churn – and can also contribute to a wider, word-of-mouth loss of reputation amongst the user's social network. Voice services are also the most regulated aspects of mobile telephony, with operators typically scrutinised for service availability and customer satisfaction.

"Some of the factors affecting voice are clearly network-related in origin – spotty coverage, insufficient capacity, issues with core network switching and so on. But generally, these are well-known and (fairly) well-managed by the operators themselves. What is less obvious is the contribution that the handsets themselves make to the voice performance equation and thus user experience. Given identical network conditions, not all handsets perform equally well."

While there are certain clear conformance tests that phones must pass, these do not show the full picture. Their real-world behaviour is less predictable, set against variable network conditions, handover between cells and technologies, sensitivity in low-signal environments and tolerance of differences in network setup.

If we take an Internet-based analogy, while mobile browsers pass "conformance" tests on different handsets, there are significant differences in how successfully these are implemented – the Safari browser on the Apple iPhone clearly offering superior performance in most users' eyes to most of the competition, for example. In this example, the difference is largely driven by manufacturers' user interface and touch screen technology, underpinned by their optimisation of the browser software and graphics capabilities of the phone's application processors. For mobile voice, the performance is driven more by the baseband (‘modem') chip and the associated protocol stack and interface software. Certain aspects of voice user experience are driven by higher-layer software like the phonebook application or the dialler itself, but the actual performance of the voice stream is quite lower-level.
"Disruptive Analysis believes that additional scrutiny of handsets' voice performance is well-justified. Despite the enthusiasm for new forms of mobile data service, there is also likely to be innovation in „mobile voice 2.0? services, which could drive an important part of future operator revenue growth."
But LTE, like some other network technologies such as WiMAX, is all-IP. Voice services will either need to be transported as "pure VoIP", perhaps driven by an IMS or NGN application core – or, alternatively – reuse existing circuit-switched voice protocols, tunnelled via an IPSec connection. (This latter approach is similar to the GSM-over-IP voice used in the older, WiFi-based UMA standard).
In any of these scenarios, a range of additional performance characteristics will need testing and optimising: 
– Management of the vagaries of IP connectivity, such as variable latency, packet loss, jitter and so on. 
– "Vertical handover" between VoIP and circuit calls, when the phone moves out of LTE coverage and back to 3G or 2G networks. Also known as "circuit fallback", this introduces a huge set of performance-related issues, from the total dropping of calls, through to delays or perceptible differences in voice quality. 
– In some cases, there may be an attempt to run VoIP over HSPA+, or even Evolved EDGE 3G/2G networks, rather than using „circuit fallback?. 
– In some cases, deployment of LTE may proceed in parallel with "refarming" of some older 2G spectrum bands. There is a possibility that narrow LTE channels may be squeezed into parts of existing GSM frequencies, with the two coexisting as operators slowly look to migrate users and radio allocations away from older and less-efficient technologies.

In addition, other new network technologies will become widespread over the next five years. Femtocells (small ‘Access Point-sized cellular base stations) should provide improved indoor coverage and capacity. But again, these will present a new set of challenges for handsets when it comes to voice – especially during transitions between macro and femto domains as users hand off from one to the other. Although in theory all 3G phones should work with a 3G femtocell, this does not mean that all will work equally well.

For further info contact:
www.broadband-testing.co.uk