Pattern Classification Based Handoff Algorithms

A new class of adaptive handoff algorithms that views the handoff problem as a pattern classification problem is proposed in this chapter. Neural networks and fuzzy logic systems are good candidates for pattern classifiers due to their nonlinearity and generalization capability. Simulation results show that the proposed algorithms improve the distributions of SIR and traffic compared to the conventional algorithm, increasing the spectral efficiency and quality of service of the cellular system. Adaptive direction-biasing is proposed to reduce the processing load and improve the cell membership properties.