基于改进PSO算法的传感网覆盖问题研究

发布时间：2018-11-09 07:35

【摘要】：近年来，无线传感器网络（Wireless Sensor Network，WSN）技术的快速发展使其得到了各行各业的广泛关注。其中，无线传感器网络覆盖质量的优劣关系到整个系统正常工作时效率的高低。评价无线传感器网络覆盖质量的参考标准有很多，如网络覆盖率，网络连通性，网络能耗性，通信延迟率等。本文以网络覆盖率作为评价无线传感器网络性能的指标，并以提高网络覆盖率为主要目的，对无线传感器网络覆盖进行优化。粒子群优化算法（Particle Swarm Optimization，PSO）是Kennedy和Eberhart通过观察鸟群的群体觅食行为，模拟鸟群觅食过程中带有择优选择的信息交互机制，而提出的一类群集智能计算方法。PSO算法以其简单易行的优点得到了广泛的利用，也因其易早熟收敛，搜索精度不高等缺陷，有大量的工作对其进行了改进处理。无线传感器网络一般是由大量的传感器节点组成，所以无线传感器网络覆盖优化属于多目标优化问题。在PSO算法中，粒子群的每个粒子都携带一定的信息，这些信息对应无线传感器网络覆盖优化问题的潜在解。而每个粒子都具有多维度特性，可以用于对应部署的传感器节点。在粒子的进化过程中，较差粒子通过向较优粒子学习而改善自身的解（即传感器节点的部署位置），算法经过多次迭代之后最终会得到最优解。PSO算法所具有的多粒子多维的特点以及较强的信息交互能力，，使其适合解决无线传感器网络覆盖优化这类多目标动态优化问题。标准PSO算法虽然已具有一定的信息交互能力，但其存在易早熟收敛，优化能力较差等问题，所以在解决无线传感器网络覆盖问题中的优化性能并不理想。之后出现的自适应PSO算法与VFPSO算法在一定程度上改善了标准PSO算法的缺点，但在优化无线传感器网络覆盖率问题中易早熟收敛的现象仍然制约着覆盖率的提升。针对早熟收敛问题，本文在自适应PSO算法和VFPSO算法的基础上对二者加以改进：在自适应PSO算法惯性权重的进化度中加入全体粒子历史最优平均值前后代的比较，使进化度的计算依据更加全面；在VFPSO算法中引入维度选择机制，使随机扰动对早熟问题的干预更加高效。与PSO算法类似，BBO算法（Biogeography-based Optimization，BBO）也是一类具有较强信息交互能力的多目标优化算法，其改进算法（VF-BBO算法）在传感网覆盖问题中对覆盖率的提升效果较好，因此将VF-BBO算法作为两种改进PSO算法的性能对比算法。通过改进前后的算法仿真对比，并结合VF-BBO算法作为覆盖率优化的参考，改进后的两类PSO算法较好的解决了早熟收敛问题，使无线传感器网络覆盖率提升了5%~15%。
[Abstract]:In recent years, with the rapid development of wireless sensor network (Wireless Sensor Network,WSN) technology, it has received wide attention in various industries. The coverage quality of wireless sensor network is related to the efficiency of the whole system. There are many reference standards for evaluating the coverage quality of wireless sensor networks, such as network coverage, network connectivity, network energy consumption, communication delay rate, etc. In this paper, the coverage of wireless sensor networks is taken as the index to evaluate the performance of wireless sensor networks, and the main purpose of this paper is to improve the coverage of wireless sensor networks to optimize the coverage of wireless sensor networks. Particle Swarm Optimization (Particle Swarm Optimization,PSO) is a mechanism of information exchange between Kennedy and Eberhart, which simulates the selective selection of birds in the process of foraging by observing their foraging behavior. The PSO algorithm has been widely used because of its advantages of simplicity and ease, and has been improved by a great deal of work because of its shortcomings such as premature convergence and low searching accuracy. Wireless sensor networks are generally composed of a large number of sensor nodes, so wireless sensor network coverage optimization is a multi-objective optimization problem. In the PSO algorithm, each particle of the particle swarm carries certain information, which corresponds to the potential solution of the coverage optimization problem in wireless sensor networks. Each particle has multi-dimensional properties and can be used for deploying sensor nodes. In the evolution of particles, poor particles improve their solutions by learning from better particles (that is, the deployment position of sensor nodes). After many iterations, the PSO algorithm has the characteristics of multi-particle multi-dimension and strong information exchange ability, which makes it suitable to solve the multi-objective dynamic optimization problem of wireless sensor network coverage optimization. Although the standard PSO algorithm has some information exchange ability, it has some problems such as premature convergence and poor optimization ability, so the optimization performance is not ideal in solving the coverage problem of wireless sensor networks. The following adaptive PSO algorithm and VFPSO algorithm improve the shortcomings of the standard PSO algorithm to some extent, but the phenomenon of premature convergence in the optimization of wireless sensor network coverage still restricts the improvement of coverage. Aiming at the problem of premature convergence, this paper improves the adaptive PSO algorithm and the VFPSO algorithm on the basis of which the evolutionary degree of inertia weight of the adaptive PSO algorithm is added to the comparison of the offspring before the historical optimal mean of all particles. Make the calculation basis of evolution degree more comprehensive; Dimension selection mechanism is introduced into VFPSO algorithm, which makes the intervention of random disturbance to precocious problem more efficient. Similar to PSO algorithm, BBO algorithm (Biogeography-based Optimization,BBO) is also a kind of multi-objective optimization algorithm with strong ability of information exchange. Its improved algorithm (VF-BBO algorithm) can improve the coverage of sensor network better. Therefore, the VF-BBO algorithm is regarded as the performance comparison algorithm of two improved PSO algorithms. Through the comparison of the algorithms before and after the improvement, and combined with the VF-BBO algorithm as the reference for the optimization of coverage, the two improved PSO algorithms solve the problem of premature convergence and increase the coverage of wireless sensor networks by 5%.
【学位授予单位】：江南大学
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：TN929.5;TP212.9

【参考文献】