ARTIFICIAL NEURAL NETWORKS WITH VERTICAL HANDOFF PREDICTION BASED ON USER BEHAVIOUR

Abstract

Wireless Data Network governed by radio waves deploys wireless medium for data communication. Mobility is the
major challenge in integrating the wireless nodes. Handoff in mobile nodes demand uninterrupted data transmission
while preserving the network integrity. Handoff process consumes lot of network resources, increases the network
traffic and is also susceptible to data loss. Handoff prediction will foresee the handoff that is likely to occur in
future so that the handover operations are done beforehand. This paper gives general overview of the vertical
handoff prediction using Artificial Neural Networks (ANN) and Association Rule Mining (ARM). The proposed
methodology uses ANNs to determine whether a handoff is necessary with the current network parameters, which is
confirmed by Apriori algorithm. A detailed comparison is given between the ANN-Apriori and Support Vector
Machine (SVM)-Apriori hybridization. The results indicate that the former performs better than the latter in terms
of accuracy in prediction handoff.