Simulation of optimal selection algorithm for wireless sensor cluster head node Bayesian statistical network

Abstract

This paper proposes a routing algorithm of cluster tree network and further combines the hierarchical structure of clustering with that of neural network, and designs a data fusion algorithm based on clustering routing protocol. Then, aiming at the difficulty in selecting the weights of neural network, a weight optimization neural network based on particle swarm optimization algorithm is proposed and applied to multi-sensor fusion. The simulation results show that the number of cluster heads of ACEC protocol is more concentrated on the expected value and has good stability. The algorithm selects cluster head nodes by non-uniform clustering and dynamic threshold, which ensures the balanced distribution of cluster head nodes in the network, reduces the network energy consumption and prolongs the service life of the network. The success rate of ancec protocol is similar to debug protocol, but with the increase of transmission time, LEACH protocol and debug protocol do not consider the link quality factor when forwarding data, so the communication link quality is uneven when selecting the next hop relay point in each round, so the data transmission success rate has a relatively obvious downward trend. The fusion result is clearly better than the poor two sensors, but inferior to the best sensor. This is due to the low SNR of sensors SNL and SN2, so their recognition effect is relatively poor, which also conforms to the rule of multi-sensor fusion. The results show that the method based on qdpso-bp network fusion is feasible.