Advanced frequency planning techniques for TDMA and GSM networks

With the increasing popularity of cellular telephony service, providers are having to support more and more users with limited infrastructure. Nowhere, is it more apparent than the limited resource of "spectrum". With the amount of money being spent in acquiring the rights to use pieces of spectrum, it is imperative that this resource be used in the most cost-effective manner to provide excellent quality of service to as many users as possible. We address the classic spectrum utilization problem of frequency planning. We present a new and novel method for solving this problem for TDMA systems such as GSM and North American cellular systems. The past work in this area has focused on either: (1) using an idealized hexagonal grid system or (2) using an idealized modeling of the RF environment. In our approach we make no assumptions about the size, shape, configuration, or the position of the cell sites. Also we use direct drive-test data that reflects all the non-uniformity and unpredictability of the real RF environment. We develop new algorithmic techniques that can handle the complexity of the actual terrain and drive-test data and take advantage of the existing non-uniformity in arriving at the solution. Using direct drive-test data and avoiding idealized assumptions guarantees a good in-field performance of the frequency plan generated by our technique. Our results show significant improvement over solutions obtained from classical reuse patterns and frequency planning techniques. Using our techniques we have created frequency plans that have 75% more RF capacity with little or no degradation in C/I and C/A performance compared to an existing reference plan.