2G RF tools optimize capacity in PCS networks

A new generation of RF tools is emerging that can save time and money in optimizing newly launched PCS networks to their capacity potential. Forward projections of network performance under increasing traffic load are required to accomplish optimization, based on actual switch data.

PCS networks will be capable of handling more than 10 times the capacity of their analog cellular counterparts at maturity-in terms of offered traffic load density in erlangs per square kilometer. This is not only because of the additional channels per sector afforded by using digital technology, but also because of the smaller cells that can be used and the distributed switching and call processing of the base station controller (BSC). Networks will be optimum if the capacity matches the geographic pattern of offered traffic load.

The increased capacity will be achieved with increased capital costs. Mistakes in building more costly infrastructure, compared with the competition, may limit the viability of the network in any given market.

Many of us have been surprised that initial PCS launches often require more cells than originally expected to cover an area. Other cost variables also affect PCS network expansion. They depend on the geographic patterns of traffic load with subscriber growth, expanded services and rate plans. These patterns are difficult to predict in advance because of competition and rapid technology advances.

The roadmap to avoid costly mistakes for phased capital expenditures in infrastructure requires forward projections of network performance under increasing traffic load. Such projections will use actual traffic data from the switch as a basis and will define decision points for capital investment in a portion of the network. Decision milestones include:

* when to increase capacity with a cost-effective antenna change only. * when to add an additional carrier to an existing sector. * when to add an additional sector to an existing site. * where to add an additional site(s). * where to add an additional BSC. Such projections are beyond the capability of existing RF planning tools that incorporate empirical COST 231 models. A new generation of tools is needed to optimize existing PCS networks. Capabilities required in the new tools include: * the ability to predict coverage in small cells in the near-zone (foreground) of the base station and/or large diffracting objects. COST 231 models predict in the far-zone only. For a street-level user, the foreground boundary is at a range of 1km for every 100 feet of antenna height at PCS. Most of small cell areas are in the foreground. * the ability to predict performance for users in buildings, elevated above street level, due to building penetration. * the ability to make a true statistical prediction of C/I. * the ability to compute the data throughput rate available by considering modulator performance to time-variable interference, envelope fading, and dispersion. * the inclusion of post-processing algorithms to characterize the reflective properties of clutter by morphology. * an inherent prediction accuracy that is significantly less than the expected C/I variability of a link within the target cell due to interference (or noise rise within a CDMA cell/sector). * a sigma (S) of less than 6dB is required; cell-edge C/I S of 8.5dB is expected; the CDMA noise rise to the instability point is 6.5dB. *use of both high-resolution terrain database, with the ability to add manually derived clutter from field data, and maps.

A predictive propagation modeling technique has been developed by the Communications Research Centre of the Canadian government, Ottawa, Ontario, that performs a quasi-exact diffraction calculation in two dimensions when high-resolution terrain and clutter profile data is available. The calculation engine employs the Fresnel-Kirchoff method developed by Dr. James Whitteker. It is a physical-optics-based wavefront model, and it separately computes the direct and specularly reflected Huygens source fields in a chained calculation, quentially, point-by-point, in range away from a base station transmitter according to equally spaced points in the terrain/clutter database.

The Fresnel-Kirchoff model has been validated against measured data obtained by the U.S. government and is estimated to have a prediction accuracy within a 5dB standard deviation using 1-arc-second terrain data, 250m U.S.G.S. land-use-land-cover data in environments no more dense than suburban morphology.

High-precision predictive RF tools have recently been introduced by Map Info [DB Planner] of Troy, NY, and by Wave Concepts International [Athena] of Dallas, that embody the Fresnel-Kirchoff technology and that are being tailored for the PCS optimization market. The first U.S. license for the technology is held by Bibey Engineering of Arlington, VA. A tool by AMCI of Omaha, NE, embodying the technology has been in private use for several years.

To illustrate the power of the Fresnel-Kirchoff method in predicting diffraction properties along radials, The plots shown in Figures 1-4 on pages 40 and 42 were created by the Athena Profiler. The graphs show both field strength for 90% reliability in dBu and the terrain profile in meters on the same graph. The Profiler permits obstacles to be manually added to the terrain height. These plots, each given as 43kB JPEG images, have a lot in common:

* 150 points calculated, 125 points displayed (2.5km range), uniformly spaced. * 20m increment (point-to-point resolution). * 100-foot antenna height. * Allgon 7200 antenna, 0 degree downtilt. * mobile antenna four feet above ground or clutter. * C-block PCS frequency. * 2W from transmitter, 2dB losses. * location; residential Santa Barbara, CA, 3-arc-second terrain data linearly interpolated to the 20m interval. * light vegetation; three 15-meter mature trees per acre. *surface roughness of 2m RMS.

It is clear that this degree of resolution would be impossible to obtain with any empirical model that depends on averaging of terrain height to enable an above-ground performance calculation. In these profiles, the effects (foreground peaking and nulling) of specular reflection have been minimized through the choice of the 2m surface roughness parameter.

Terry S. Cory, PE, Cameron Crum and Ramsay Decker. Cory is the CEO and Crum is vice president of technology for Wave Concepts International, Dallas. Decker is a radio propagation consultant in Cedar Rapids, IA.