The commercial roll-out of HSDPA (High Speed Downlink Packet Access), the upgrade to UMTS 3G, is starting, with operators like O2 shortly to go live with full services. As with most wireless technologies, the prospect of actual deployments is throwing cold water on the marketing claims and turning attention to how this network will perform in the real world. With carriers accelerating their upgrades in order to meet accelerating demand for mobile data services, some are concerned that HSDPA is being pushed out before it is proven and that, as with WiMAX, early systems may involve disappointing compromises.
Motorola has run one of the largest HSDPA trials and has published a set of guidelines, based on those trials, for how cellcos should get the most out of the technology in the first generation. The conclusions do not generate great confidence that HSDPA will be, at least in its early stages, as straightforward a route to the full promise of mobile internet and data services as operators had hoped.
Motorola makes the critical point that HSDPA will need to offer performance close to fixed broadband if services are to gain the all-important positive reaction from early users. This is particularly important if HSDPA operators are to put up a good fight against incoming challengers using other technologies such as WiMAX, and if they are to extend their reach to convergence. For many cellular-only operators, HSDPA is the first chance they have to deliver services uniformly to fixed and mobile users, and so compete with the converged carriers with their existing in-home penetration with wired fixed lines. This is also an advantage offered by WiMAX, and the drive to control the converged fixed-mobile platform is making indoor penetration ‘ still a weakness for 3G networks ‘ a priority, but one that is unlikely to be addressed until the next release of HSDPA in a year or so’s time.
Motorola offers three important recommendations for implementing HSDPA in a way that will not disappoint early adopters. First is the need to supply sufficient processing power to reduce latency. The vendor’s trials found that HSDPA can suffer from delays when using open web applications such as browsers, because of a mechanism called ‘state switching’ that moves a user from a high to a low speed state based on activity. This makes for efficient use of the network but unacceptable performance for true mobile broadband. In markets where HSDPA is being used to achieve true broadband ‘ rather than, for instance, to support more users per cell at lower speeds, another key application ‘ Motorola believes carriers must avoid switching users down to the low speed state, which means intensive processing capabilities at the base station. ‘Network operators should ensure that they have sufficient processing power at the base stations to schedule the highest possible number of calls at launch of the HSDPA service’, the study suggests.
The second recommendation is to adopt handsets that include a signal processing function known as an equaliser, which enhances performance when the user is moving and can increase average data rates by 40%. Only a few handsets currently support this capability, but again, it is essential if the operator’s primary motive in upgrading is to support full broadband data rates ‘ over 1Mbps ‘ while the user is fully mobile.
Another problem is that video streaming performance degrades as more users become active, with as few as four active users being enough to cause video to freeze. To avoid this, close attention must be paid to scheduling mechanisms to ensure that video takes priority, otherwise users will be disappointed with the quality of video applications, which are often the most important aspect of an advanced service to be marketed to consumers.
None of the findings of the trial are too surprising, but they do highlight the strain on the network and its processors that mobile broadband will cause. There is no cheap and simple way to implement mobility at broadband speeds, yet this will be essential to support the next generation of applications, many video based and personalized, and to give mobile operators a chance of providing the converged fixed-mobile service alongside their wireline counterparts. If neither of these elements are essential to the business model in the next couple of years, operators may do better to hold off on HSDPA until they really need these capabilities, or to use it to support more users per cell, and therefore gain better margins, while not promising more than a basic 3G level of service.
The WiMAX world is learning similar lessons. Its technical challenges are different, since it comes from a heritage that is strong on data delivery, IP support and indoor penetration, but weak on mobility. But like HSDPA adopters, WiMAX providers know that true mobile broadband may be glamorous, but only if it makes a profit, and until the technology matures, there will be more margin to be made from delivering conventional services in a more efficient manner.