When the first GSM base stations were installed vehicle-mounted mobile phones routinely failed at cell boundaries. Now, 12 years on, a new technology is being introduced and we are forced to ask, are dropped calls a necessary part of the UMTS learning curve? Achim Grolman of Willtek Communications assesses the problems and how they are being solved.
Compared to American CDMA wireless standards, UMTS defines far more network elements and interfaces. This has been done deliberately as such granularity gives operators, for each entity, the choice between multiple vendors to buy the best products. However, the evolutionary nature of standards means that vendors may initially follow different interpretations of the UMTS specifications. This means that operators face a complex task to make systems work effectively from the outset and that requires a lot of testing and discussions with different suppliers. This will come as no surprise to those familiar with the development of cellular infrastructure as there are no conformance test cases for the network elements as there are for terminals.
In the early 1990s, it took two to three years to get GSM networks going. For 3G, the complexity of integrating telecoms and IT network elements has added to the problem experienced then and, in truth, it is surprising how well the first 3G networks are already operating.
As real network elements are knitted together, problems in the specifications are discovered and resolved, which means specifications are constantly being fine tuned. This process directly impacts on the testing environment as conformance test specifications have to follow these changes; test equipment and test cases then need to be validated. This is why test equipment is often not available before the network elements as test equipment and network elements are developing in parallel in the early stages of a new standard.
Installing the RAN
3G network operators have already installed base stations (or Node Bs to use UMTS jargon) and radio network controllers in major cities. Installation in smaller towns and urban areas is also progressing although much more slowly as there is little revenue from the existing 3G networks to pay for further rollout. 3G is being installed under strict financial controls and that means attempting to optimise the design of the network.
To this end, simulation tools are in place to determine the best locations for the base stations. Yet, to be effective these tools need to be calibrated with real-life tests using constant wave or W-CDMA transmitters, measuring receivers and RF propagation test software. These efforts can minimise problems later, such as the need to acquire additional sites later.
Operators have learned from less-than-optimal network layouts and configurations in the early GSM phase. For example, back then dozens of engineers from different vendors circulated London to find out why their phones lost calls along the M25 in spite of strong carrier signals. Add to this the fact that W-CDMA behaves differently to GSM, making in-fill more complicated, and the benefits of taking the time to plan the network as effectively as possible is clear. However, getting it right in theory and practice are not necessarily the same. Therefore, once installed, radio networks are verified and optimised with the help of test mobiles and drive test software, pinpointing coverage gaps or problems due to cell reselections occurring unnecessarily and/or too often.
Handover techniques
There may only be a few UMTS phones available on the market – first-tier vendors of mobile phones are introducing their first-generation models now – but to be fair, these phones and the networks they are operating on, already support most of what customers are used to from GSM phones: voice, SMS, MMS, WAP and web over packet data channels. But customer expectations are high and, on a technical level, the handover of a call from cell to cell, now highly reliable in GSM, provides a very specific additional challenge.
There are, in fact, several handover techniques in UMTS. Adjacent cells using the same carrier frequency perform a soft hand-off. In this both base stations hold a connection to the phone for some time so ideally, there is no disruption in the call. The softer hand-off takes the phone from one antenna sector to another of the same base station. A hard hand-off (which breaks the existing connection before making the new one) however, occurs when the carrier frequency is changed.
As Europe demands dual mode operation, a special case inter-RAT (Radio Access Technology) hard handover occurs when a call is transferred between UMTS and GSM. It is a new requirement and technically challenging. Indeed, the first successful handover from UMTS to GSM was only achieved by Ericsson less than a year ago and therefore it is a function that has not been supported by early 3G networks.
The core network provides the link between both radio access technologies. When a UMTS call is due to be handed over to a GSM base station, the core network requests information about the new channel from the GSM base station system. While the handover message is usually sent to the phone via a GSM base station, the BSS now sends it to the core network which in turn encapsulates the normal GSM handover message in a UMTS handover message. This way, the GSM protocol stack in the phone can interpret the data and set frequency, timeslot and other parameters accordingly although the message is transmitted over the UMTS interface.
Changing times
It is a complicated process and internal test methods and equipment to analyse software interfaces are proprietary, at least for each first-tier vendor of terminal equipment. Later on, second and third tiers will benefit from third-party tools and software written around generally available GSM/UMTS chipsets.
In addition to testing hardware and software modules within the phone, measurements of the overall performance are required and vendors have to prove that their products are up to standards. Although UMTS protocol testers are available, the majority of the test cases are not, since they take some time to develop. RF and protocol specification go first, then the conformance tests, and at the end of this chain, test case implementations from test equipment suppliers need to be validated. Note also that the Inter-RAT protocol is still being optimised!
One of these optimisations concerns adjacent cell measurements. In order to hand over the call to the best suitable base station at the cell boundary, the network needs information from the phone regarding the received signal strength and quality of the surrounding cells. This is easily possible with the TDMA nature of GSM which leaves enough time for the phone to monitor adjacent base stations. However, UMTS-FDD calls are transmitted and received using permanent carriers. Only with the introduction of the compressed mode where the user signal provided at constant bit rate, can it be transmitted at a higher rate for some time, thus leaving a gap which the mobile can use for adjacent cell measurements on different frequencies. An alternative would be dual-receiver designs where the mobile phone can receive on the assigned channel with one receiver and on the neighbouring cell carriers with another. This approach, however, is more costly in terms of components, weight, power consumption and battery life.
Now that the network knows the best base station to hand over the call to, it provides the phone with information about the new carrier. Depending on the direction in which the call changes from one radio access technology to another, the terminal has 40 to 120ms (without preceding synchronisation) or even up to 220ms (in the unsynchronised case) time to switch technologies, synchronise in frequency and time to the new base station, set up the protocol, and continue the call. One design approach, commonly used in early models, uses separate base band chips for GSM and UMTS which are more difficult to synchronise during handover. However, even the other approach, a combined chip, still poses a challenge in terms of both hardware and software design.
While data connections are regarded as something of a special case now, they are expected to become more common with 3G. The problem there are a few applications will not survive the cell change order procedure because they time out before the mobile has synchronised and set up the link to the new base station. UMTS equipment will support concurrent voice and data connections. The data part, however, will most likely be lost in a UMTS to GSM handover because phones do not support both connections at the same time in GSM. This problem may be solved with voice over IP over both UMTS and GSM/GPRS.
As is becoming more obvious as we move through the 3G structure, it is much more complicated. Furthermore, the absence of protocol testers fully supporting these procedures to test both phones and networks, means all parties involved have to find a different method.
Alternative test route
This began when manufacturers of phones and network elements announced bilateral cooperation agreements years ago to further their 3G developments, but interoperability tests between virtually all manufacturers are required and are being executed now. This can be time-consuming but should not be seen as a luxury as it will avoid situations such as the one that occurred two years ago when a major handset maker had to withdraw its GPRS phone from the market because it did not work with one network vendor’s equipment. Interoperability tests are especially important for standards which are open to interpretation and/or provide many different options for doing the same. In other words, they are vital for new, emerging technologies.
When the first stable versions of the UMTS standard came out, people expected ‘UMTS islands in a sea of GSM’ and to many it was natural to believe (or demand) that from day 1, terminals would be capable of handing over voice and data calls between GSM and UMTS. Reality, however, shows us that the handover has never been the first function to work with a new cellular standard, be it 1G, 2G, 2.5G or 3G. Engineers still need time to optimise hardware and software designs, especially in the handsets. There is much pressure from operators to get handovers working because dropped calls are the first things that customers discover to be wrong with the network and are a major obstacle to success. But this does not mean that UMTS doesn’t work — it simply does not fully support this feature – yet. It is only a question of a few months until new GSM/UMTS phones are available, allowing users to seamlessly roam across air interface technology barriers.