Design of Multi-Objective Particle Swarm Optimization Routing Algorithm for Heterogeneous Wireless Sensor Networks

• Main Authors: CHUNG, CHIA-CHIH, 鍾佳志
• Other Authors: TSENG, CHWAN-LU
• Format: Others
• Language: zh-TW
• Published: 2019
• Online Access: http://ndltd.ncl.edu.tw/handle/53p35m

碩士 === 國立臺北科技大學 === 電機工程系 === 107 === In the static wireless sensor network, the nodes close to the sink are rapidly depleted energy due to the transfer of a large number of data packets. This problem is called energy hole problem, which will end the network life early. Considering that the nodes in the practical application have different initial energies and will sense the data of different packets, this thesis adds a heavy load node (with higher initial energy and data packets) to form a heterogeneous wireless sensor network. In the heterogeneous network, the relay node needs to transfer more data packets, which leads to more intense energy holes. Therefore, how to improve the node energy consumption imbalance in heterogeneous networks and prolong the network life has become the focus of new routing algorithms. Using multi-objective particle swarm optimization (MOPSO) technology, this thesis proposes a MOPSO based energy balance clustering and routing algorithm (MEBCR) for static heterogeneous wireless sensor networks. First, considering the distance between the cluster head and the member nodes, the distance between the cluster head and the sink, and the residual energy of the cluster head, the MOPSO technology is used to optimize the cluster head selection. Moreover, this thesis analyzes the importance of these fitness functions in the cluster head selection. The best solution is selected according to the hypercubes coordinate in the Pareto frontier. Based on the cluster head selection results, a multi-hop routing tree between the cluster heads is conducted to establish a cluster head transfer path. Finally, considering the energy consumption balance between the groups, particle swarm optimization (PSO) is used to optimize the clustering, which makes the energy consumption in the network uniformity and prolongs the network life. The performance metric network lifetime is defined as number of communication rounds when 50% nodes energy exhaustion. The simulation results show that compared with PSO-UFC, EECR and TEAR, the network lifetime of MEBCR is increased by 13.26%, 10.14% and 18.27%, respectively. The MEBCR also reduced the residual energy standard deviation by 4.29 times, 3.7 times, and 4.2 times, respectively. In addition, under different heterogeneous wireless sensor networks, MEBCR can achieve better network performance than other three routing algorithms.