Planning 3G networks is going to involve a steep learning curve for engineers used to narrowband GSM. Fortunately, rapid and cost-effective optimisation of UMTS network designs is on its way as new software optimisation products become available. But to be truly effective, which elements of network design must these products most closely address?
European wireless operators rushing to launch service on third generation UMTS networks face a multitude of challenges, both technical and economic. Most have paid extraordinarily large sums for access to the dedicated spectrum. Now they must, at reasonable cost, deploy networks with sufficient coverage, quality, and capacity to attract large numbers of customers in a suddenly worrisome business climate. UMTS operators do not have the benefits enjoyed in build-outs of earlier generation GSM-900 and 1800 systems, where easy access to funding and pent-up demand for service made capital efficiency less of a priority than time to market. With greater economic constraints, 3G system planners need to pay more attention to engineering detail in an effort to optimise their networks. Using traditional engineering methods, such meticulous network design takes time and is itself costly. The good news is that new software products, soon to be available, should bring dramatic changes in the way system engineering is performed, allowing for rapid and cost-effective optimisation of UMTS network designs.
Optimisation software systems have been available for frequency planning in narrow-band systems such as GSM for the last few years, and have proven to be highly effective in improving quality and capacity of these networks. UMTS systems, like the earlier generation CDMA networks in use in North America and Asia, do not employ frequency planning. Instead, the same broad radio channel is used in each base station and sector of the network.
To meet capacity requirements, a network may employ several UMTS RF carriers. This approach does, however, come with a price tag attached. Firstly, only two to three UMTS carriers are available in the segmented frequency band, and, for practical reasons, one of them may be exclusively used by the micro-sites layer, leaving only one or two carriers for the macro layer. Additionally, once a carrier is deployed it should have a contiguous coverage, (that is it should be deployed in all sites in the region). Therefore, optimisation of CDMA and UMTS networks requires a different approach from that applied to previous generation networks, and effective software optimisation products must address the elements of network design that most significantly affect overall performance and efficiency.
Within the UMTS RF carrier, different User Equipments (UEs) are separated by the different orthogonal codes used for digitally mixing, or “spreading”, each traffic data stream. At any moment during a call (or packet data transmission), the sector transmitting the signal intended for a UE will be the one that provides the strongest signal (as received by the UE). When a UE is in a position where it receives signals from two or more sectors at roughly the same level, they will all simultaneously transmit the signal for that UE, via a process called soft handover. Coverage is provided where the RF path loss between the UE and at least one sector is sufficiently low to allow for successful reception of transmitted data. Further reducing path loss lowers transmitted power levels on both links, thus enabling higher throughput for serving sectors. On the other hand, when a number of UEs have comparable path loss to more than one sector, the resulting mutual RF interference tends to reduce total channel data throughput.
Localised demand
Regardless of the air interface technology used, the two primary goals of network design are: provision of reliable coverage everywhere within the specified service area; and provision of localised channel capacity (or total channel data throughput rate) to meet localised demand.
In designing new networks, coverage is usually the dominant issue, and the goal is generally to provide adequate coverage with as few base stations as possible, thus minimising both initial capital outlay and deployment time. However, care must be taken to assure that localised capacity can be efficiently expanded to meet growth in traffic demand following a successful service launch.
As demand for packet data service grows in a given area, throughput capacity can be added either by increasing the number of deployed base stations or by increasing the data throughput available in each existing base station.
Of Course, adding base stations takes time and additional capital, while increasing the data handling capability of existing base stations can be achieved quickly and at relatively low costs. So obviously operators will prefer to grow capacity by increasing per-base station throughput, at least once coverage objectives have been met. This requires minimizing the mutual interference between individual sectors.
Another challenge that the carrier must deal with is the selection of sites among the existing GSM sites. Clearly not all 2G sites will be implemented in the early stages of UMTS network roll out and special consideration needs to be given as to which sites are the most suitable.
While optimisation goals for narrow-band and UMTS networks may be essentially the same, the methods used to achieve – or at least strive for – optimisation are quite different. In narrow-band systems like GSM there are three design aspects that influence optimisation: frequency planning-- the assignment of discrete RF channels to each base station and sector; settings of operational parameters such as handover thresholds and base station transmit power levels; and physical configurations such as the location, type, height, and orientation of base station antennas. There are also operational parameters that must be taken into consideration in a UMTS network, but their adjustment is aimed more at proper system operation than network optimisation. Therefore, it is mainly in the area of physical configuration that true UMTS optimisation is achieved.
As an example of practical UMTS network optimisation, consider the simple case illustrated in Figure 1. Here we see two sectors from different UMTS base stations jointly providing coverage, with substantial overlap, to two areas of high traffic concentration. Many UEs in these two areas have low RF path loss to both sectors. In this situation, the more critical downlink capacity of both sectors is impacted because of these high levels of mutual interference.
If this were a narrow-band system, the problem could be managed for optimisation in numerous ways. First of all, because of their overlapping coverage, sector A and sector B would not be assigned the same RF channels. They could then jointly serve the traffic in the high-density areas without mutual interference. If traffic loading ended up unequal, the sectors would each be assigned the number of RF channels they required. A significant loading imbalance could be addressed by manipulating relative base station transmit power and mutual hierarchy and handoff parameters for sectors A and B. Finally, the antenna configurations of one or both sectors could be modified to change relative coverage and thus relative traffic loadings.
In a UMTS network, adjusting the operational relationship of sectors A and B would require a different strategy. Obviously, since the same RF channel would be employed in both sectors it would be impossible to avoid mutual RF interference by spectral isolation. What is less obvious is that “optimising” network parameter adjustments would be of limited help. This is because adjustments such as manipulating base station transmit power levels will have no effect on the relative path losses between a mobile and the base station antennas and their ability to increase isolation is very limited.
Parameter manipulation
In practice, a certain amount of parameter manipulation may be effective in UMTS optimisation because in most cases UMTS network capacity is downlink-limited. Some “fine tuning” of traffic loading among sectors may be obtained by such parameter adjustments as downlink pilot power. However, major improvements in UMTS network capacity must entail reducing mutual interference between sectors, and that will primarily involve adjustment of the relative path losses between mobiles and nearby base stations.
Of course, it is impractical for a network operator to change the geographic distribution of the traffic or the topography it serves, so controlling the mix of path losses means adjusting the configuration of the base stations relative to the environment. In Figure 2, we see how such a change may be applied to the example discussed above. The antenna beam that defines sector A has been rotated to increase path loss to mobiles that are closer to sector B, and the sector B antennas have been lowered so that surrounding “clutter” increases path loss to mobiles near sector A. The result is reduced coverage overlap, particularly in high traffic areas, and thus decreased levels of mutual interference.
In optimizing an initial UMTS network design, location and configuration of base stations must take into account both anticipated traffic distributions, as we have seen in the above example, and coverage. For example, the changes suggested in Figure 2, particularly lowering the sector B antennas, may cause coverage problems in other areas. Maintaining overall coverage reliability might therefore entail changes to other sectors as well. A skilled engineer can do a reasonably good job of this balancing act over a small number of sectors, but in a network with several hundred base stations the task becomes overwhelming. This fact is even more significant when taking into consideration the increased complexity that will be placed on the network once data services have been added. As a result, most “manually” designed UMTS networks are far from optimal.
Until recently, operators of narrowband networks were also faced with the problem of non-optimal network designs. That situation changed with the introduction of automatic frequency plan optimisation software solutions that have revolutionised design and optimisation of narrow-band networks. The best of these tools work by reducing the frequency planning task to the creation of a mathematical network model, and then applying a very powerful optimisation algorithm to assign the required number of channels to each sector while keeping overall interference to a minimum.
Of course, frequency planning is not applicable to UMTS networks, but the good news is that research is well under way to apply the same optimisation approach to RF configuration management, which as we have seen goes to the heart of UMTS system design. This will require what are essentially three network modules. The first will define base station-to-mobile path loss distributions for given base station configurations based upon geographical database with the option to calibrate it by field measurements. The second module will distribute the traffic across the network according to the service type (voice, data, high-speed data). Finally, the third module will combine the previous modules and create a simulation of the network to define the sector-by-sector parameters such as transmit power, receive levels, and handover status. The optimisation engine will apply these models to advanced processes in order to find optimal network RF configurations that will provide additional capacity and improve performance without sacrificing coverage.
Roni Abiri is vice president, wireless products, of network optimisation specialist Schema (www.schema.com)
* This article first appeared in Wireless Evolution World Focus 2001, and Informa Telecom Group Publication (
www.informa.com)
Credit:
Wireless Evolution
.gif)
