Resource Allocation Model for Estimating Non-Uniform Spatial Loads in Cellular Wireless Networks

Abstract

Carriers of cellular wireless networks often partition the service territory into small bins in order to monitor and provide adequate service throughout the territory. For example, an area of 50 kilometers by 50 kilometers, served by 500 Base Transceiver Stations (BTS’s), may be partitioned into 40,000 bins of 250 meters by 250 meters. Carriers estimate the signal strength from every BTS to every bin. The carriers also collect information regarding the carried load and lost call information at every BTS. This information is useful for planning purposes, including the evaluation and modification of a frequency assignment plan. It is also useful for operational purposes, including balancing loads among the BTS’s. However, effective planning and control would be further enhanced by having even more detailed load information, specifically, estimates of the offered load initiated at every bin. In this paper, we propose an equitable resource allocation model to derive such load estimates. The model uses as input offered load estimates at each BTS. Service probabilities that assign the load generated at a bin to multiple BTS’s are derived using signal strength information. Demographic data is used to estimate a demand target for each bin. The model uses a performance function for each bin, which represents the weighted, normalized deviation from the demand target. The objective function is a lexicographic minimax objective, where the loads at the BTS’s are viewed as resources to be allocated among the bins. The model has an intuitively appealing interpretation, and the relations between this model and a model for point-to-point demand estimation in wire-line networks will be presented. A specialized algorithm can readily generate estimated offered loads for problems with a very large number of bins.