Bounds of performance measures in large-scale mobile communication networks

In this paper, we propose a new method for evaluating performance measures in a mobile communication network having a large number of base stations. Each base station has its own zone where its radio wave reaches. These zones are overlapping with neighboring zones, and handovers become possible. However, the existence of overlapping zones makes the analysis of a network difficult. To overcome this difficulty, we introduce an aggregated process on a smaller state space. For each of the important performance measures, using this aggregated process, we can construct a Markov decision problem that derives an exact upper (or lower) bound. Further, under some condition, we can specify a situation at which the upper (lower) bound is attained. This situation is analyzed with a far smaller model, and we can get the upper (lower) bound numerically or even by simulation.