Research Of Packet Size Optimization Algorithm In Wireless Sensor Networks

Wireless sensor networks (WSN) is a self-organized measuring and controllingnetwork composed of many tiny sensor nodes with communication and computingability. It can be used to detect or perceive a certain regional information andtransmit the information to the observer in real time. Due to its wide application inmilitary, industry, environment protection, et al, it has attracted extensive attentionfrom many countries and research institutions. In most of applications, sensornodes are powered by batteries, which are difficult to be replaced or recharged.Consequently, the energy efficiency of WSN is always the research focus. Thecommunication between sensor nodes causes the main energy consumption of WSN,and packet size which is the major communication parameter will influence theenergy consumption efficiency (ECE) of sensor node directly. Therefore, it is a keytopic to optimize packet size so as to maximize ECE.The main research contents and innovations in this paper are as follows:This paper investigates the existing energy consumption efficiency (ECE)model, and a novel model is constructed to reveal the relationship between ECEand packet size, wireless transceivers, communication protocols andcommunication channel so as to complete the model in theory.This paper investigates the basic Simulated Annealing algorithm, and proposesa packet size optimization algorithm based SA, combining the characteristic ofECE. Meanwhile, computation cost is defined to evaluate the performance of thealgorithm, combining the method of numerical analysis. Finally, we demonstratethe validity of the algorithm in extensive simulation experiments.This paper investigates the basic Simultaneous Perturbation StochasticApproximation algorithm, and combining the characteristic of basic SPSA and ECE,presents the SPSA based packet size optimization algorithm. In addition, averagequality of solution and computation efficiency are defined to assess theperformance of the algorithm in statistical significance. Finally, we demonstrate thevalidity of the algorithm in extensive simulation experiments.