Research On Data Collection Optimization Algorithm Of Rechargeable Sensor Network Based On Data Utility

With the rapid development of micro electro-mechanical system,wireless communications,low power embedded system and wireless charging technology,wireless chargeable sensor network technology is gradually becoming the key technology in many fields,such as military monitoring,environmental monitoring,logistics management and so on.In general,there are two ways to transmit the collected data of the node to the base station: multi-hop transmission between sensors and mobile sink.Due to the many advantages of the mobile sink,data collection based on this approach is becoming a research hotspot in wireless sensor networks.How to effectively design the data collection algorithm to plan the sink path will greatly affect the wireless sensor Network performance.Aiming at the high delay characteristics of mobile sink data collection,this paper presents a data collection method based on sensor node data utility.Then based on the sensor node data utility,heuristic algorithm and online algorithm are proposed.The heuristic algorithm plans the sink path according to the sojourn point data utility,while online algorithm pan the path according to the distance between sojourn points because it is able to know in advance the datay utility of all sojourn points.Because of the limitations of heuristic algorithms and online application scenarios,we presents a mobile sink data collection strategy based on q-learning algorithm in large-scale wireless sensor networks,this strategy combines the experience of collecting previous sojourn points with the current state of the sink,considering the local optimum and random exploration in the current environment,making mobile sink continue to learn the environment and feedback to adjust the collection path during the process of collecting sojourn point data,which make mobile sink bring back more efficient data when it returns to the base station.Finally,we verify the performance of heuristic algorithm,online algorithm and q-learning algorithm proposed in this paper through simulation.The experimental results show that the proposed algorithm has a good performance in different specific application field.