Exploiting Focused Time Delay Neural Network for mobility prediction in Ad-Hoc Networks

Abstract

Ad-Hoc networks are flexible and easy to install and trends to deploy in several application such as Mobiles, Vehicles, Smart Phones, Sensors, Robots and Hospitals etc. But to sustain continuous established link connection between nodes is the major challenges, due to randomly node movement that directly effects on network performance parameters. However an accurate future node position estimation or mobility prediction before leaving present position of a node can sustain continuity link connection and can improve network performance parameters. In this paper we exploited Focused Time Delay neural network (FTDNN) to predict mobile node position, as a node positions trajectory is a kind of time series position and the FTDNN are also suited for time series prediction. The model based data pattern using Gauss Markov mobility model and Real-World data pattern are used to experiment prediction results and we demonstrate the effectiveness of the FTDNN based mobility prediction model and measure the prediction accuracy using RMSE and MAE.