High Speed Circuit Switched Data (HSCSD) is one of the steps towards 3G services and mobile Internet access. Mobile infrastructure is available now for this technology and requires implementation of a software upgrade to existing GSM networks, which can be carried out remotely on most base stations. It will also require a corresponding software upgrade in the core network (MSC). HSCSD can be used effectively for applications such as file transfer and in certain circumstances video-conferencing but is probably not sufficiently cost effective for voluminous data and Internet use, due to its reliance on circuit rather than packet switching.

HSCSD does not do much to help spectrum capacity problems that the mobile operators are facing. GSM at the 2G level is based on circuit switched systems.

HSCSD (High Speed Circuit Switched Data) extends these systems to not only circuit switch voice conversations but also data connections. By aggregating four time slots (or channels) of 14.4kbps (through compression of 9.6kbps) through software downloads to the radio network, HSCSD can improve the data rates to potentially 57.6kbps which is roughly the speed of today’s landline modems.

However, due to the fact that HSCSD uses multiple time slots, it nails up bandwidth and hence reduces voice capacity over the network. Moreover as an increasing portion of traffic becomes data in origin, this problem would be accentuated.

For a GSM operator, HSCSD is a relatively modest investment (compared to alternatives) of US$20-30 million and easy to integrate into the existing architecture. Is it an attractive option? Today, operators have a number of technology options for migration to 3G services. This market is largely untested and demand estimates vary wildly. If demand is poor, then low end solutions (such as HSCSD) are likely to give operators a better return, but if demand proves to be at the high end of expectations or above, then HSCSD will quickly become noncompetitive.

It is our view that most mobile operators will opt for General Packet Radio Service (GPRS) implementation rather than HSCSD, despite HSCSD being easier to implement and available earlier. We believe that the plethora of applications and services that will be created in the mobile data environment will be Internet related and the Internet is packet based.

However, we also expect some mobile operators to adopt HSCSD as a way of "experimenting" with data services to keep ahead of the service curve. Orange recently announced a data based mobile service to be launched in October 1999. This will be based on HSCSD technology, supplied by Nokia. The Finnish operator, Sonera, has also deployed HSCSD solutions supplied by Nokia, but more recently chose Ericsson for GPRS. Many vendors have not published their respective HSCSD customers. We have outlined some public announcements in the table below. Ericsson states it has over 15 HSCSD contracts.

Another drawback with HSCSD is that the system will allow a subscriber to get higher data rates in a particular cell by allocation of the multiple time slots described above, but when handover is carried out to another cell it can not guarantee the same number of time slots potentially translating into slower speeds in the other cell.

As well as software upgrades, HSCSD will also require new terminals to be purchased. Nokia has introduced a High-Speed PC card for wireless networks (claimed to be the first product on the market). The Nokia Card Phone 2.0 is a PC Card with built-in GSM phone supporting HSCSD and data transmission speeds of up to 43.2Kbit/s without data compression. It weighs 58g and is 5mm thick and will be launched in Q4 1999. It operates on GSM900/1800 MHz networks and offers live video and Internet access over GSM using the higher bit rate.

The mobile division of Singapore Telecommunications (Singtel Mobile) launched HSCSD services across its GSM-900/1800 networks based on Ericsson gear. The solution delivers data at speeds up to 38.4kbps compared with 14.4kbps today. Currently data accounts for only 5% of Singtel’s traffic today. Business customer services will include 2 way e-mail, Internet and intranet usage. Data compatible handsets are available from Nokia and Ericsson.

Sonera has also launched "Hispeed Data" encompassing the entire coverage area of its network.

GPRS is, in our view, the primary migration path to faster data rate technologies using packet switching rather than circuit switching. GPRS requires operators to upgrade existing base stations and core network at a relatively modest expense. For subscribers, it means buying new GPRS enabled handsets to make use of the increased data rates and "always on" connection capability. GPRS will give developers and operators the opportunity to evaluate the network performance, create new services and applications and gauge likely demand and price points.

GPRS (General Packet Radio Service) is an evolutionary specification upgrade from 2G (GSM) to 2.5G, also referred to as Phase GSM+. The move to packet enables many users to use the same bandwidth in a cell and increases the data speeds from the 9.6kbps today to potentially up to 115kbps by aggregating up to eight 14.4kbps channels. From an operator’s perspective, adoption of GPRS means more efficient use of the limited spectrum.

In our data networking handbook published in November 1998, we discussed the difference between circuit switched and packet switched systems. We described why packet switched technology was the most efficient option for handling data. The same is true in a wireless network. Third generation networks, and most 2.5 Generation technologies except HSCSD, are based on packet switching technology.

The GPRS phone communicates with the GSM base stations, but unlike circuit-switched data calls which are connected to voice networks by the mobile switching centre, GPRS packets are sent from the base station to what is called a Serving GPRS Support Node (SGSN). The SGSN communicates with what is called the Gateway GPRS Support Node (GGSN), a system that maintains connections with other networks such as the Internet, X.25 networks or private networks. A GPRS network can use multiple serving nodes, but requires only one gateway node for connecting to an external network such as the Internet.

Therefore, to integrate GPRS into an existing GSM network, it is necessary to introduce these GPRS support nodes, the SGSN and GGSN into the existing 2G network. The SGSN is the node within GSM infrastructure that sends and receives data to and from the mobile stations by means of packet switching / routing. It also keeps track of the mobiles within its service area (call authentication/mobility management). Traffic is routed from the SGSN to the base station controller, and to the mobile station via the base station. The GGSN can then provide access to Internet service providers (ISPs) as well as the allocation of Internet protocol (IP) addresses.

In addition, a Packet Control Unit (PCU) is added to the existing BSC. The increase in traffic (data) will increase the load on the BTS and additional sites will be required. There may also be a slight detrimental effect on existing voice quality and coverage which may necessitate further investment in traditional network infrastructure.

Although GPRS can offer speeds of up to 115kbps - over ten times better than the 9.6kbps offered by current second generation mobile networks - it is also possible to offer a multiple slot configuration which can achieve lower rates while still improving upon current performance. For example, a three-slot configuration could offer rates of around 33kbps. Timeslots (channels) within a cell can be aggregated for higher data throughput rates but bandwidth per channel is shared among all concurrent users of the GPRS service within that particular coverage cell. The maximum utilisation of this shared packet network is realistically thought to be around 75%. In practice 70kbps is more likely due to the potential number of users in the same cell.

In a GPRS network, a permanent connection is made to the terminal, which eliminates the need to log on (and off) repeatedly. The effect of this is to reduce the costs for the operator and for the user. Testing and trials are being undertaken at the moment and a number of contracts have been awarded. These are listed in the table below. It is also very difficult to tell who is winning the core network GPRS business and who is winning the radio network portion. There are many other contracts which have not been announced yet, however we believe that in most cases the original GSM supplier will stand a very good chance of winning GPRS deals if it has these solutions in its portfolio. There are a few exceptions when new entrants could break into market. These are when the base stations are very old and can not accept the software download (and hence need to be replaced by new base stations) or when there is a multivendor approach already adopted by the operator at 2G level.

GPRS and circuit switched GSM services can co-exist successfully in the same network allowing mobile operators to leverage existing services whilst adding new ones. In our opinion, GPRS will have greater support from GSM operators than HSCSD as it provides faster data rates, packet capability and is optimised for short bursty traffic such as Internet access.

We believe that GSM operators will want 2.5G solutions that rely on their existing outside plant, towers, antennas, and network backbones as well as their site leases and maintenance crews. In our view GPRS can be envisaged as a long-term investment as it leverages this "legacy" investment and rolls into higher capacity systems such as 3G when the applications and services require it. GPRS is built on top of the existing GSM network infrastructure and requires new network elements at the mobile switching centre and the base station controller. In most cases no new equipment is required at the Base Transceiver station.

The Internet is a packet-based architecture and if we make an assumption that the mobile Internet occurs, then mobile systems will need to become more optimised for Internet applications. GPRS is a mobile packet system. Users will ultimately pay for what is being transferred over the network rather than the time that they spend on it.

Assuming that the demand for data and Internet services over mobile increase rapidly, then we believe that GPRS will provide the platform for mobile operators to migrate towards higher data rate network deployment. The majority of mobile operators in the next 12-18 months will deploy GPRS as their preferred lower data rate 2.5G technology. Initially the most advanced operators in each market will invest in the infrastructure, forcing competitors to move down the same path. Pricing will clearly be an important factor in creating demand for these new services. We expect that pricing will relate to the amount of data traffic delivered, rather than to the length of the call as on circuit switched networks. In our opinion, GPRS as a commercial service will be available from Q4 2000 if handsets are available.

`We also believe that GPRS will be adopted by operators who wish to demonstrate commitment to 3G services, which may help in securing a 3G licence. GPRS has the advantage over HSCSD because of its packet capability to connect users at all times while on a call.

Some operators do not believe that the two lower data rate technologies are mutually exclusive. While most operators appear to be backing GPRS, there are alternative strategies. German operator E-Plus is in favour of High Speed Circuit Switched Data (HSCSD) for a variety of applications including file transfer, corporate access/teleworking, online e-mail, e-cash, video/audio on demand, and remote healthcare. However, E-Plus does recognise that GPRS is better suited for Internet and Intranet use. It seems possible that both may be used in parallel, each one being deployed for whichever service it is best suited for.

These terminals will need to have different operations techniques if they want to support both circuit and packet services. Moreover the design will be affected by the types of timeslot overlap. If the timeslots do not overlap, this means that only one transceiver will be needed. However if timeslots overlap leading to greater throughput, then additional radio equipment will be needed which will increase the cost, complexity and power requirements as well as heat generation.

GPRS handsets will probably be available in the second half of 2000. We believe that in their initial format they will only support a limited amount of timeslots (Type 1) which will mean that the full 115kbps speed will not be available in the initial handset releases, not allow for broadcast type applications (Phase 1) and subscriber will be registered for both circuit and packet services (Class B). The initial programme outlined for terminal delivery released by the ETSI standardisation committee is outlined below:

What this means to the mobile subscriber is that in the initial release of GPRS handsets the full 115kbps data access speed will not be available. But more importantly, terminal vendors are making sure that they are in line with the specifications adopted by infrastructure players to enable full interoperability across different networks. What this means is that a Nokia GPRS handset can be used on an Ericsson GPRS network.

Prices for the above are not yet available and it is also not yet possible to determine the level of subsidy that may be offered by the operators (likely to vary significantly between markets). However, for the moment we are assuming a premium to existing, high-end, 2G product of 25% for initial products. We do not expect GPRS versions of low end digital products initially.

EDGE technology will be deployed in the same spectrum as today’s GSM/TDMA services, however, it will be much more expensive than either HSCSD or GPRS as it requires a different modulation scheme in the radio portion. This in turn requires major infrastructure hardware upgrades to the radio portion of the network. Each cell will need to be upgraded with an EDGE transceiver unit (hardware) and software upgrades will be needed for BSC and BTS, which can be done remotely. The technology could be deployed initially in the hot spot areas and then roll out gradually to gain the coverage element.

Another school of thought is that EDGE could be used on a nation-wide basis as it has more or less the same advantages as 3G technology at a lower cost and when the urban centres need more capacity 3G "islands" could be deployed there. These deployment issues will become clearer once guidelines for 3G licences are released.

In our opinion EDGE will be used by:

1. Mobile operators without 3G licences to be competitive on data rates.
2. Mobile networks in the suburban and rural areas (this will depend on licence coverage specifications).
3. In the US for TDMA upgrades.

Infrastructure upgrade costs will be USD300-600mn depending on mobile penetration and coverage.

EDGE uses a different and more efficient modulation scheme: 16 quadrature amplitude modulation (QAM), rather than the gaussian modulation shift keying (GMSK) scheme used over the radio interface by GSM and GPRS. The 16 QAM system opens up more bandwidth per radio carrier or cell.

In non technical speak, assume spectrum is equivalent to a page of A4 paper. With the current technology we can use ten lines of the sheet. With EDGE we will be able to use 25 lines. Hence we will be able to write more on the piece of paper. From the subscriber perspective this means that it will be able to send more information. From an operator’s perspective this means it does not need to obtain more spectrum (more A4 paper) to increase data rates (send more information).

In the US, CDPD has been deployed as a data enhancement to TDMA networks since the early 1990s. It has not been very successful for a number of reasons such as lack of applications and services that mobile end-users require. It increases data rates to potentially up to 19.2kbps.

EDGE technology was initially designed for TDMA networks and then the equipment manufacturers realised it could also be deployed in GSM implementations. IS-136HS (High Speed) is the next phase in evolution for TDMA networks that increases data rates to 384kbps. It is a packet based upgrade that will require new modulation schemes on the radio side increasing spectral efficiency.

The changes that will occur in the CDMA network for it to become "packetised" would be the addition of a data infrastructure (router-type) device at the Base Station Controller. An alternative is to deploy additional hardware at each Base Station which would prove very costly to the operator.

The CDG (CDMA Development Group) has also been working on a specification that will increase data rates further to potentially up to 144kbps known as CDMA2000 phase 1. This upgrade will be more expensive than IS-95 release B as it will require new cards to be placed on the base stations. 1XRTT terminals will be capable of operating on existing IS-95 networks due to utilisation of the same spectrum.

1XRTT in technical circles, brings the first phase of 3G services to market and lays the foundation for a broad array of high-speed wireless information services. In November of 1998, Bell Atlantic Mobile announced that it will begin phased introduction of cdma2000 high-speed wireless data capabilities in its cdmaOne network. In Australia, Telstra is launching a cdmaOne network later this year and has already announced its plans to trial wireless Internet, multi-media and packet data services enabled by cdma2000. Sprint PCS recently demonstrated interactive, high-resolution wireless videoconferencing at 128 kilobits per second (kbps) using the cdma2000 platform. 1XRTT provides 144 kbps packet data in a mobile environment and a potential two-fold increase in both voice capacity and standby time, advanced packet data services and greatly extended battery life. All of these capabilities will be available in an existing 1.25 MHz channel in existing spectrum, allowing all operators to take advantage of these 3G capabilities without the burden of re-farming or securing new spectrum.

Qualcomm is developing a mobile data technology that works with IS-95 version A. Higher Data Rate (HDR) could potentially allow PCS CDMAOne mobile operators to offer higher speed data rates up to 2.4Mbps within the standard spectrum allocation. We believe that this technology is similar to EDGE for the TDMA environment in that it uses a different modulation scheme within the existing spectrum to increase data rates.

PDC will leapfrog straight into 3G systems without the evolution phase seen in GSM. In our opinion, Japanese operators such as NTTDoCoMo need to alleviate network congestion by gaining more spectrum capacity and the forthcoming 3G spectrum gives them this capability. We also believe that the Japanese telecoms industry is attempting to push its manufacturers to be more significant players in the global mobile market and this explains their strategy. However, it now looks as though Japan’s timetable is slipping and Europe may be launching 3G in a very comparable timeframe. The 3G system that is currently being rolled out will in itself be an evolution, the initial data rates are between 64kbps and 144kbps and hence could be seen as a 2.5G function.

The success of wireless has been one of the major technology stories of the 1990s, particularly for the European players - Nokia and Ericsson. Asian manufacturers have made relatively sustained but broadly unsuccessful efforts to penetrate the infrastructure and handset markets. We believe that both Japanese and Korean manufacturers are striving to make a greater impact in the build- out of 3G systems.

Although European manufacturers have been at the forefront of 3G development, the most aggressive plans for the introduction of services and systems to date have come from NTT DoCoMo in Japan. The aim of the Japanese, in our view, is to erode the advantage European companies have gained through the success of GSM. However, the Ministry of Posts and Telecoms (MPT) in Japan was recently reported to have delayed submission of recommendations for 3G standardisation.

This service has proven very successful with over one million subscribers signed up in six months. Moreover, introduction of these services has led to ARPU increase of 13%.