Dynamic Algorithms for Sensor Scheduling and Adversary Path Prediction

Abstract

In this thesis we describe three new dynamic, real time and robust sensor scheduling algorithms for intruder tracking and sensor scheduling. We call them Tactic Association Based Algorithm (TABA), Tactic Case Based Algorithm (TCBA) and Tactic Weight Based Algorithm (TWBA). The algorithms are encoded, illustrated visually, validated, and tested. The aim of the algorithms is to efficiently track an intruder or multiple intruders while minimizing energy usage in the sensor network by using real time event driven sensor scheduling. What makes these intrusion detection schemes different from other intrusion detection schemes in the literature is the use of historical data in path prediction and sensor scheduling. The TABA uses sequence pattern mining to generate confidences of movement of an intruder from one location to another location in the sensor network. TCBA uses the Case-based reasoning approach to schedule sensors and track intruders in the wireless sensor network. TWBA uses weighted hexagonal representation of the sensor network to schedule sensors and track intruders. In this research we also introduce a novel approach to generate probable intruder paths which are strong representatives of the paths intruders would take when moving through the sensor network.