Channel management models in cellular mobile systems

Abstract

Channel management is one of the most important problems in the design of cellular mobile systems. Several channel management models have been suggested in the literature. Using fixed channel assignment (FCA), to reuse radio spectrum efficiently, frequency planning considering traffic demands is required. We describe two representative channel assignment problems MSP (minimum span problem) and MBP (minimum blocking problem) for frequency planning and suggest efficient solving procedures for these problems. As a cellular mobile system evolves, cells will become smaller. Decreasing cell size exacerbates the spatial and temporal volatility of traffic and makes optimal frequency planning infeasible. Dynamic channel assignment (DCA), which is more flexible than FCA, will play a more important role in such a system. We suggest an efficient borrowing channel assignment (BCA) model which is a kind of DCA. For MSP, we suggest an efficient general heuristic algorithm which uses a GOS updating scheme. This scheme generates a sequence of target GOSs that converges to the required GOS in a self-adjusting manner. We also suggest a simple procedure to further improve the channel allocation solution obtained by the algorithm. Computational experiments show that our algorithm provides solutions with much smaller span than existing other ones. MBP is mathematically formulated as a nonlinear combinatorial problem. Using the piecewise linearization technique for convex functions, MBP can be converted into a linear combinatorial problem. To begin with, we deal with the simple minimum blocking problem (SMBP) considering only the co-channel interference constraint. To treat SMBP more conveniently, we introduce the concept of pattern and reduce the problem. Using Lagrangean relaxation and subgradient optimization techniques, we obtain high-quality solutions with information about their deviations from true optimal solutions. Computational experiments also show that our method is very ...