Pattern classification based handoff using fuzzy logic and neural nets

Conventional handoff algorithms are susceptible to varying propagation environments, traffic intensities, and user speeds due to the lack of parameter adaptation. This paper proposes a new class of adaptive handoff algorithms that views the handoff problem as a pattern classification problem. Adaptive direction biasing is proposed to reduce the processing load and improve the cell membership properties. The paper shows that the desired balance among the system characteristics can be achieved by making appropriate design tradeoffs in a pattern classification based handoff framework.