Handoff management in wireless networks using predictive modelling

Abstract

Handoff management is a key element in mobile networks to sustain an ongoing session of a user. In this paper, we propose a handoff management technique to reduce handoff delay and call dropping. This handoff management technique incorporates a prediction component which will predict a UE's next cell and best handoff time. The prediction component is realized using a predictive model which uses earlier recorded mobility pattern of a user to make the prediction. Neural Network (Multi-layer Feedforward Network) has been used as predictive model. A data cleaning component is added using sequence mining technique to filter only relevant patterns from a large raw data to input to the predictive model. This process significantly improves the prediction accuracy and reduces the computational complexity involved with MFNN.