re: A good hard look at GPRS
>> An Early Taste Of 3G?
Dr William Webb
PA Consulting Group
Mobile Communications International
Issue 88
02 February 2002
GENERAL PACKET RADIO SERVICE HAS TAKEN A LONG TIME TO ARRIVE. NOW THAT IT IS HERE, HOW IS IT PERFORMING AND WHAT NEEDS TO BE DONE TO SECURE A PROSPEROUS FUTURE FOR THE TECHNOLOGY?
GPRS was first discussed during the original GSM standardisation work in the 1980s. However, it was then postponed until a latter phase due to the pressing need to publish the basic GSM standard.
In the mid 1990s, standardisation under the GSM 'Phase 2+' initiative got under way in earnest, with the standards becoming stable around 1999 and the first network becoming available in 2000. As is often the case there were some teething problems with these new networks, principally surrounding the stability of the network under heavy load.
Since then, just as with the initial launch of GSM, the problem has been primarily one of terminal availability - for much of 2001 only Motorola was shipping sizeable numbers of GPRS-enabled handsets.
Ominously, the reason that Nokia was not shipping GPRS terminals sooner was because it was more concerned with waiting for the market to emerge than it was with seizing the competitive advantage afforded by being early to market. Such reticence suggested that the Finnish vendor did not see GPRS as an instantly popular service. Ironically, pleas from Nokia's competitors were heard suggesting that the market would not take off at all until Nokia delivered a product. The provision of packet data to GSM is important for a number of reasons.
For the subscribers it offers an always-on service without the need for dial-up connection as well as potentially higher data rates than the 9.6kbps currently available.
For the operators it offers the potential for increased revenue from new data services and a more efficient use of the spectrum for data transmission than is currently available.
Having said that, it should not be forgotten that GSM already has packet data capability in the guise of SMS. SMS is of limited value for transmission of large files but can work well for simple always-on applications, such as sports results, and has been proving increasingly popular, even at the incredibly high charging rate of around £1000/Mbyte.
More recently, GPRS has been seen as the potential saviour of WAP.
As has been well documented, WAP has proved disappointing to many users. While GPRS will not solve all the problems, the removal of the need to dial up, the increased connection speed and the improved efficiency of spectrum use - which should lead to lower costs - might go some way towards making the mobile internet a more compelling concept.
With these drivers in mind, 2001 saw substantial GPRS roll-out. By September last year there were 87 commercial GPRS networks in the world. In the US, AT&T wireless, Cingular and Voicestream all had commercial networks with the intention that something like 40 per cent of the US GSM coverage would be GPRS-enabled by the end of 2001 and close to 100 per cent by the end of 2002.
It seems likely that by the end of 2002, most of the GSM operators in the world will also be supporting GPRS throughout the majority of their networks. This represents a substantial vote of confidence in GPRS by almost all of the GSM community.
Be that as it may, the technology has not escaped an industry hype machine that has spawned disappointing realities in the past.
Initial discussions of GPRS centred around the high data rates that could be achieved - up to 171.2kbps in principle (although only under specific conditions). The actual data rate available depends on two parameters - the number of traffic frames available and the coding scheme used. A GSM carrier has up to eight traffic frames that could be used for GPRS. However, in practice most mobiles are only capable of accessing between two and four frames, and the operator would be unwilling to provide all eight to one user because of the congestion this might cause. The highest data rates are achieved with no coding, but in most cases this would lead to unacceptable error rates. In practice, data rates of 9kbps to 13kbps per time slot are more likely, providing sufficient error protection for most applications. This suggests practical data rates of between 18kbps and 52kbps and to date most operators appear to be quoting data rates within this range.
Less clear is the range of services that will be offered. So far, most operators appear simply to be offering GPRS as a bearer service, charging through some combination of monthly fee and per Mbyte tariff. End users are then expected to run services like WAP and computer data access over the GPRS bearer rather than the circuit-switched carrier used today. So, to date, GPRS appears to be a bearer aimed at enhancing existing mobile data applications rather than one poised to enable a host of new services.
As such, it requires significant understanding of mobile data by subscribers; they need to judge for themselves when GPRS is appropriate for their needs and be able to configure their phone appropriately - a task that only a small percentage of the public would likely find straightforward. Of course, this may change as time goes on.
Also unclear are the number of subscribers to date. Most industry analysts guess that the current numbers are "very small" but that they might grow to around ten per cent of the subscriber base in due course. Much also depends on the initiatives deployed by operators - in October 2001, Telia announced three months' free GPRS subscription in an effort to bring subscribers onto the service. With these kinds of offers many subscribers may just be trying out the service, so it will not be until well into 2002 that subscriber numbers will start to tell a useful story.
The number of subscribers, the data rates offered and the average traffic levels will determine the overall load on the existing network. In order to implement GPRS, operators typically take a number of voice channels out of the voice pool and nominate them as GPRS channels. If the GPRS load grew, this would require additional capacity to be installed at many base stations to avoid degrading the voice service. For operators already suffering from capacity problems this might require substantial additional investment in the form of further, smaller cells - an investment that the operator might not be ready to make if they believe that the arrival of 3G will provide rapid relief to these problems. As a result, if GPRS is successful, there may be cases where the GPRS data rates are poor - potentially even worse than the existing circuit-switched data rates.
Alternatively, GPRS success might spur operators to deploy microcellular networks providing much improved capacity and coverage. The next 12 months should see the completion of the GPRS technical picture.
By the end of 2002, most networks should have GPRS deployed across most, or all, of their coverage area. All of the upper tier handsets should also be GPRS- enabled, and even the lower tier increasingly arriving with GPRS numbered among their capabilities. With GPRS also available throughout the US, and many tri-mode phones in existence, the end of 2002 should effectively see GPRS available on a near-global basis assuming that the roaming issues currently preventing widespread use are overcome.
Less clear is how the services and subscriber picture will look 12 months from now. With little data to go by and widespread public scepticism at the ability of mobile phones to provide useful data services, some compelling new products will be needed to drive a rapid uptake in subscription. So far there is no sign of these, and so, there is substantial doubt as to whether the end of 2002 will see a healthy subscriber base.
For the last year or so, GPRS has often been seen as a testing ground for 3G.
With 3G originally scheduled to roll out around 2002, it seemed that there was only a two year window for GPRS, after which the higher data rates and advanced services of 3G would be the logical choice of the business user. Hence the original intention to test a number of applications and services on GPRS ready for full roll-out on 3G.
That thinking is slowly changing as 3G is increasingly delayed and as the initial data rates of 3G drop to around 64kbps - only marginally better than a four-slot GPRS service. Also, with some deploying 3G only in islands of coverage in dense areas, GPRS will be the only packet data bearer in many parts of many countries for the foreseeable future. Compared to 12 months ago, GPRS is now looking much more viable as a long-term wireless data bearer.
Indeed, under one scenario, GPRS remains the only wide-area packet data bearer. Here indoor coverage is provided using WLAN solutions delivering data rates of up to 10Mbps and providing a range of new services, such as multimedia download and video communications, whilst GPRS provides sufficient data rates for outdoor coverage where users are expected to have lower data rate requirements. In this scenario, there is little role for 3G other than perhaps to provide additional capacity.
Regardless of which scenario comes to pass, with the delays surrounding 3G, GPRS is becoming increasingly important for GSM operators as the key data bearer from now until around 2006, when 3G subscriptions are predicted to take off strongly. This makes it important to look at GPRS as the first in a line of new data services rather than a testing ground for the 'real thing' due soon. Conversely, if operators are unable to build a viable subscriber base and range of services around GPRS, it seems unlikely that they will be able to do so around 3G. This will cast increasing doubt around both the future role of 3G and also some of the operators who have made licence bids on the assumption of increased ARPUs.
Unfortunately, the mostly abortive WAP launch means that subscribers will take additional persuasion to try a new data service. Few new services have emerged that are viable within the data range provided by GPRS. With GPRS, GSM operators will have the same 'toolkit' as the Japanese i-mode providers, but for a variety of reasons, replicating the success of i-mode outside of Japan is looking problematic.
Most importantly, GPRS is more than just a new service added to the GSM bearer.
In many ways it is the bridge from voice to multimedia. It heralds always-on connectivity at higher data rates than available today. It is the first instance of a packet-based core network which may form the platform for eventual migration to an all-IP core. For most proposed services it provides near-equivalent capabilities to 3G.
Furthermore, being more network upgrade than new network construction - and as a result technically relatively straightforward - it is relatively inexpensive to implement. If GPRS is not financially viable then it is difficult to justify migrating a network to a multimedia capable IP core and 3G access network, requiring a significant re-think as to the future direction of wireless communications. In years to come, the success, or failure, of GPRS is likely to be seen as a key milestone in the development of mobile communications.
Given that GPRS provides a bridge to the future, operators need to do much more to persuade users to cross that bridge. If they could make the phone the single communicator device, used in the home, office, car and wide area they could substantially increase the value that users gain from the phone. By proactively encouraging home wireless solutions and office IP networks, including embracing WLANs where appropriate, they can dramatically increase the usefulness of the phone, which in turn will increase the application base.
With most GPRS phones also incorporating Bluetooth, operators have most of the network tools they need to set about the task of making the mobile phone become an individual's "remote control on life". A simple starting point might be to work with PDA manufacturers to offer a Bluetooth-enabled PDA and GPRS/Bluetooth phone with remote email and synchronisation package at a subsidised price and monthly fixed price contract. Many more such examples exist.
Launching a service with the unfriendly name of GPRS and with pricing in little understood units of per Mbits seems to repeat many of the mistakes of the past. For the sake of the mobile communications industry, operators need to take some risks with the new tools at their disposal. <<
- Eric - |
