Barrier Construction Strategy For Confident Information Coverage Based On Reinforcement Learning

Barrier Coverage as one of the coverage issues of Wireless Sensor Networks.In Barrier Coverage field,the sensor nodes are deployed in a strip area and their sensing ranges overlap each other forming a barrier to protect the monitoring area from intruder.Reinforcement Learning is one of machine learning mechanism which is used to solve the problem of selecting the optimal sensor node to realize the goal by an autonomous agent than can sense and act in its environment.Learning Automata(LA)is a sample model of reinforcement learning mechanism.Confident Information Coverage(CIC)is a new coverage model of sensor node which fully considers the collaborative sensing ability among nodes and mines the spatial characteristics of the environmental variables monitored.Consequently,it is the focus of this thesis that designs a reasonable and effective barrier construction strategy by basing on Learning Automata model and Confident Information Coverage model.This thesis systematic introduces wireless sensor networks and barrier coverage,based on deeply analyzing the barrier construction algorithm in current barrier construction research literatures,aiming at the problems existed in other barrier construction algorithm,this thesis provokes a new barrier construction strategy.The specific work is summarized as follows:(1)We define and formulate the Confident Information Coverage Barrier path Construction(shorted as CICBC)problem with the objectives of maximizing barrier paths and minimizing IoT nodes in each barrier path.(2)In order to solve the CICBC problem,we devise a distributed CICbarrier path construction schema based on learning automata mechanism.CBLA models the network by constructing a Coverage Graph(CG)which is the topology of the randomly deployed nodes.Each IoT node equips a LA.CBLA consists of four important phases: initialization phase,learning phase,monitoring phase and repairing phase.CBLA randomly selects source IoT node from CG,and selects the next optimal IoT node by learning to build the barrier path while guaranteeing the network connectivity.(3)A series of comparison experiments are performed to evaluate the performance of proposed CBLA by comparing with other two peering approaches DBS and Random.Experimental results verify the effectiveness and advantages of the designed CBLA scheme,and show that our proposed CBLA algorithm outperforms DBS and Random solutions in terms of the number of barrier paths and the number of nodes in each barrier path.