Open Collections

Abstract

In this thesis, adaptive sectoring of a CDMA cellular network is investigated, and the aim is to maximize the uplink capacity by utilizing the mobiles' spatial information. One important feature of the algorithm developed is that it does not depend on tracking individual mobile, but rather on the statistics of mobiles. The distribution of mobiles is modeled as a spatial Poisson process, whose rate function quantizes the mobile concentration and is inferred with a Bayesian estimator based on the statistics of network traffic. In addition, the time dynamics of the rate function is assumed to evolve according to the mobiles' mobility pattern and it is formulated using the Influence model. With the knowledge of the mobiles' spatial distribution, the interference and thus the outage probability of different sector partitions of a cell can be computed. The adaptive sectoring problem is formulated as a shortest path problem, where each path corresponds to a particular sector partition, and the partition is weighted by its outage probability. In simulation examples, a hot spot scenario is simulated with the adaptive sectoring mechanism, and it is observed that load balancing between sectors is achieved and which greatly reduces the effect of hot spot.