Publication date: 01 April 2008
The global demand for high-capacity wireless broadband access networks is on the increase, with multiple signal formats competing for market share. These include WiMAX, HSPA, 3G-LTE, UMB and TD-SCDMA.
But regardless of which is used, a new approach to network planning and optimisation is required to balance the technical and commercial constraints. After all, there is little point in building a technological marvel if it is to become an economic millstone for shareholders.
Considering that Sprint (XOHM) in the USA has budgeted in the order of $3 billion to deploy WiMAX, the cost of poor or ineffective planning can have a significant cost impact on a network build.
This highlights an important question: can network planners afford to use outdated and inaccurate propagation modelling and optimisation tools for their WiMAX rollout?
This article looks closely at WiMAX network planning, although it could equally apply to any wireless broadband access technology. WiMAX is the new IEEE standard that includes 802.16d for fixed wireless deployments (including backhaul) and 802.16e for mobile applications. Higher capacity WiMAX standards, such as 802.16m, are still in development. The actual frequency band depends on regulatory conditions in each country, but WiMAX is generally deployed between 2GHz and 6GHz, with the mobile version at the lower end.
Network planning for WiMAX is a multi-dimensional task that includes the traditional engineering work of site and frequency planning, traffic and interference analysis and load balancing. For high-speed data networks, planners must also consider other metrics, such as meeting expectations for quality of service for the different classes of users.
For operators like Sprint, the goal is to maximise revenues while minimising capital expense and operational costs; as such it is necessary to consider the economic impact of the network plan. Forward thinking is vital to determine the economic and technical feasibility prior to network deployment.
Planning an urban WiMAX network is difficult because of the complexity of city environments. This is particularly true for access at street-level where signals from many basestations (BS) must combine to achieve the desired coverage level.
The old intuitive approach to network planning is based on relatively basic tools that were not designed for non line-of-sight systems. The result is usually a sub-optimum deployment that suffers from reduced capacity and serious interference problems. Fixing these is a costly and time-consuming business.
In a competitive market, delivering the required level of service with fewer basestations, or from less expensive sites, is critical to long-term profitability. This starts with having an accurate model of the RF propagation environment. Only then can the network be accurately modelled and optimised, and its performance predicted.
Network planning is only as good as the RF model used to predict performance. The first step is to make sure that the modelling tools are adequate for the job. This might entail a process of model calibration (or tuning) to ensure that electromagnetic parameters for any GIS data (e.g. terrain, building and foliage data) are appropriate for the region to be covered.
The best approach is to perform a number of measurements using a single WiMAX basestation and receiving terminal at the expected frequency of operation. The sampled data can then be used to calibrate the model for the environment to which it is being applied, and thus improve the accuracy of its predictions (i.e. minimise the error between the model predictions and actual measurements).
At the start of a project, the number and location of basestations is unknown. Rather than assuming particular values, the network should be accurately dimensioned based on the specific needs of each application:
Rather than let the design tool select basestation locations, it should use a list of possible locations spread over the entire planning area, to ensure that selected sites are realistic and obtainable. Each site can then have assigned its expected installation, rental and operational cost to allow algorithms to reduce the capital expenditure of the total network.
Other site-specific data should also be included. For example, if a hardware vendor and operating frequency have already been chosen, then antenna gains, transmit powers and RF sensitivity levels can be used to configure the planning tools. Alternatively, the tools can be used to determine the impact of frequency and hardware changes.
In particular, the antenna gain and directivity can have a significant impact. Other parameters include the number of basestation sectors, the number of operating channels and the frequency reuse pattern.
Once set up, the propagation links from each possible basestation to each customer or control site should be determined. This might include as many as 100 basestations and 1,000 control sites, resulting in 100,000 point to point prediction studies. Using this matrix of data, the planning tools can then be used to determine the best basestation sites.
These can be ranked and feedback used to confirm that the desired sites are actually available. In practice, the process will be iterative and should converge on the minimum number of required basestations.
A detailed network plan should then be developed based on the actual basestation locations, their heights and antenna patterns. An estimate of the expected coverage and capacity can then be made. Features such as a ranked view of the basestation sites as a function of capacity, coverage and cost can be extremely useful in the decision-making process (Figure 1).
Having completed the initial network design, detailed RF planning and mapping can begin. For many networks, this is usually the starting point for the planning process using non-ideal basestation sites. If the exact location of each BS is known, the planning tools should be used to predict the RF coverage levels from each site over the entire coverage region.
Particular attention should also be given to locations within a designated coverage radius (e.g. 3km) of each site and full 3D modelling is desirable to ensure accurate non line-of-sight predictions. At distances beyond this radius, analysis can use a simpler model (e.g. a 2D vertical plane model, which is highly accurate for line-of-sight locations).
Other desirable features for WiMAX include the ability to align the antennas of customer premises equipment (CPE). This is vital for non line-of-sight locations where the optimal CPE orientation is unknown prior to deployment and depends on site-specific multi-path scatter. Network optimisation should include load balancing across cells to evenly balance traffic.
Finally, a detailed coverage map can be generated for each BS and sector. Using knowledge of the WiMAX link-speeds and RF sensitivity levels, a projected link-speed map could also be produced to indicate the operating mode to each customer location.
Once the network is rolled out and commissioning begun, further analysis should be carried out. For example, site surveys to check service availability at a given rooftop or street level location could be performed. Predicted RF levels and link-speeds can then be compared with actual data to confirm and characterise the operation of the planning tool.
The stakes are high as the world watches WiMAX roll out. Will it deliver on the promise of high-speed wireless data services? Just as voice networks have evolved to data, so must the network planning tools to avoid disappointment for investors and consumers.
To support these new wireless data networks, ProVision has developed a sophisticated suite of planning and optimisation software tools called ProPhecyTM. The ProPhecy propagation engine and coverage planning tool is a new three-dimensional RF propagation tool that models dense urban environments as well as open rural areas.
Unlike many vendors who offer a generalised product range based on dated and mostly empirical propagation data, ProPhecy has been designed from the outset to meet the needs of WiMAX and similar wireless broadband access technologies. It uses a unique Radar Cross Section (RCS) technique to allow detailed scatter and diffraction analysis to be performed for both rooftop and street-level deployments. Signal propagation into buildings is also modelled.
It supports the detailed modelling of basestation and CPE antennas, can auto-align directional CPE antennas towards the optimum sector, and includes modelling of MIMO antenna configurations. The tool is extremely flexible and can be easily customised to meet specific user requirements.
The ProPhecy optimisation suite is a flexible and dynamic network optimisation tool for site, traffic, and cost optimisation as well as load balancing. It is a powerful network optimisation platform that works with the ProPhecy propagation engine to dimension and build radio networks. It considers financial and technical constraints in its analysis to deliver the most cost-effective network plan.
This new tool selects the optimum location of basestations together with the allocation of radio resources to satisfy coverage, throughput or cost/revenue constraints. The algorithms exploit the link-adaptive nature of the WiMAX physical layer to ensure that capacity (as well as coverage) is maximised.
The network planner works from an initial list of sites that are known to be available and avoids theoretical solutions containing unobtainable sites. The result is in the form of a ranked list of solutions, with ranking based on user-defined metrics (e.g. coverage, throughput, cost, revenue, number of sites etc.).
The combined toolset can reduce deployment costs by up to 20% over conventional network planning tools while increasing user data throughput by up to 30%. With access to suitable sites becoming harder and more expensive, it is no longer enough to deploy first and make adjustments later.
Figure 1: (a), The Ranked Solution Viewer in the ProPhecy optimisation module compares the cost and percentage coverage of the top 20 solutions, and (b), The 3D Solution Viewer plots coverage and capacity against cost for each solution.
Figure 2 shows optimised mean data throughput (aggregated DL + UL) for all users
George R J Green, Director Business Development, ProVision Communication Technologies Ltd., 3 Chapel Way, Avon Valley Business Park, St. Annes, Bristol BS4 4EU, United Kingdom
Tel: +44 1179 711 120, Fax: +44 1179 723 367, Web: http://.provision-comm.com/
