Optimization models for design of mobile communication systems

Abstract

This thesis is concerned with the frequency assignment and the channel allocation for the design of cellular mobile communication systems. Our first model deals with frequency assignment problem (FAP) for a system with nonhomogeneous traffic distribution and channel interference conditions. The second model considers the prioritized channel allocation scheme for handoff calls. For FAP, we provide general framework for connecting it to general graph coloring problem. Based on the lower and the upper bounds on the optimal solution of FAP, we suggest an approach to reduce the complexity of the problem. Then, FAP for ν-complete subset with two kinds of constraints, which in general consists of cells with the heaviest traffic distribution in cellular systems, is analyzed in detail. We show that the optimal frequency assignment to ν-complete subset depends on the relations between the total requirement, the maximum requirement, the number of nodes with the maximum requirement, and the ratio of the co-site and the adjacent channel constraint. Based on these observations, we provide the exact frequency assignment algorithms for all kinds of ν-complete subsets considered. By utilizing the upper bound for the chromatic number of general graph coloring, we devise a new measure for assignment difficulty which can be applied to any of non-iterative heuristic frequency assignment algorithms. Besides, we suggest another heuristic procedure for general frequency assignment problem. Our algorithm exploits the observation that increasing the channel separation between frequencies than necessary may bring the decrease in the ultimate frequency span required. Computational experiment indicates that our algorithm performs better than existing ones in some problem setting. Our second model considers the prioritized channel assignment for handoff calls. Handoff is peculiar to the cellular communication system and the failure of handoff critically degrades the service quality. In o...