The Research Of Clustering Topology Control For Multiple Basestations In Wsns With Q-learning

The development of the Internet of things is dependent on the wireless sensor network(WSN). It provides a new platform for the access information. The routing technology as the basis of its network has attracted a great deal of research. The energy of nodes is not renewable in WSNs after one-time seeding. So, it is a primary design goal of routing protocols that reduces the network energy consumption and prolongs the network lifetime in sensor network. The core routing protocols of WSNs is clustering routing protocols. The multi-level clustering tree topology structure as the basis of clustering routing protocol which is widely researched and applied due to its efficiently energy and easy to maintain.This paper firstly introduced an algorithm, called ETBG(Energy-Aware Topology Protocol Based on Gradient) algorithm builds a Gradient field in network according to the transmission range of the sensor nodes. Clustering is defined as process of dividing the network into groups in the same gradient levels. Cluster tree is built according to the different gradient levels and the concept of maximal independent set. Such an algorithm can reduce the delay time and the number of hierarchical levels when it clustering and transmits the data packet. The ETBG algorithm fails to find the optimal path because it not thinks about other various factors. It just considers the energy and distance when it selects cluster heads and the weight of single cluster head when it builds cluster tree. In large-scale WSNs, there is a problem of "energy hole" of single base station ETBG algorithm which is seriously shortened the lifetime of the network.To solve this problem, the Multiple Base Stations Clustering Topology Control Based on Q-learning(CTQL) algorithm has been proposed. The number of base stations is determined according to different scenarios and the weight of nodes is calculated by the method that multiple attributes decision-making based on ordered weighted averaging(OWA) operator in this algorithm. Our algorithm uses a method of graph theory and the gradient of directional diffusion to clustering. To achieve the clustering topology control, our method exploits Q-learning, incrementally learning at each node’s sufficient network knowledge to indentify the best path to base stations. Evaluation of the resulting the CTQL algorithm demonstrates its ability to significantly increase the lifetime of network in comparison to the ETBG,and the CTQL algorithm with multi base stations also increase the lifetime of network in comparison to single base station. This paper give the maintenance and updating algorithm based on CTQL algorithm, in order to make the network build new communication when topology has changed quickly. The network has been ensured normally operation.