Research On Clustering Algorithm Of Industrial Wireless Networks

The rapid development and sophisticated of wireless communication technology, microelectronics technology, sensor technology and computer technology have promoted the development of low-cost, low power and short distance wireless communication technology of wireless sensor networks. Industrial wireless network technology is a network monitoring and control technology that is used in industrial field environments, its appearance is due to the rapid development of wireless sensor network technology.The planar structure and the hierarchical structure are commonly used network organization structures in industrial wireless network. In the planar structure of the network, all sensor nodes have equal status, often referred as peer to peer structure. While in the hierarchy structure, the network consists of two different types of nodes:ordinary nodes and cluster head node, cluster head can be used to form the higher-level network structure. The disadvantage of the planar structure is that it can not be applied in large-scale industrial wireless network structure, because in this type of structure, each node must save and maintain the routing information to all other nodes. In large-scale network, if there are mobile nodes in the network, the control overhead to maintain the routing information of these dynamic changes will become very large.In industrial wireless network, the hierarchical network structure compared with the planar network structure, have good scalability and flexibility, the use of hierarchical network structure can significantly improve the overall performance of the network. Therefore, the rapid development of industrial wireless technology and the diversification of industrial wireless network application environment, make the research of network clustering algorithm become a research focus in industrial wireless network. The rational and efficient of clustering algorithm, makes the network resource scheduling and management, routing, channel access control and power control more easily achieved, and good cluster structure provides the basis for data fusion, which helps to reduce network data traffic, saving the energy consumption of the entire network, thus extending the life cycle of the entire network. This paper studies the clustering algorithm of the WirelessHART network. First, gives an analysis of typical wireless sensor network clustering algorithm, and compare the performance of these clustering algorithm. Secondly, the idea of K-means algorithm is introduced into the WirelessHART network clustering operation, the sensor nodes in the WirelessHART network are regarded as data objects, using the K-means clustering analysis method to cluster all the nodes in the network according to our guidelines, so that we can achieve the clustering of industrial wireless network.In the light of the inherent shortcomings of CAKM:(1) the choice of initial cluster centers have a great influence on the performance of the algorithm, for the initial cluster centers randomly selected usually results in the K-means algorithm have different clustering divided, even though have the same data set, and often can not get ideal clustering result. (2) CAKM algorithm uses a gradient method to solve the extreme value of the objective function, easy to make the algorithm into a local extreme point, not the optimal solution of the problem. We propose using particle swarm optimization algorithm to optimize the CAKM clustering algorithm, and a mixture of K-means and particle swarm optimization algorithm, KMPSO clustering algorithm is proposed.Finally, in order to validate the performance of the CAKM algorithm and the KMPSO algorithm, according to the simulation parameters in this article, setting up network simulation environment, using MATLAB simulation tool to verify the performance of the algorithm. Simulation results show that, for FND, HND and LND, CAKM algorithm compared with LEACH algorithm, respectively improved 12%,12.2%and 20.9%; KMPSO algorithm compared with LEACH algorithms respectively improved by 23.1%,26.5%and 24.1%; KMPSO algorithm compared with CAKM algorithms respectively improved by 9.9%,12.8%and 2.7%. In the life cycle of the network, the total amount of data receive by the gateway, CAKM algorithm compared with LEACH algorithm, improved by 11.3%; KMPSOM algorithm compared with LEACH algorithm, improved by 24.8%; KMPSOM algorithm compared with CAKM algorithm, improved by 12.2%. Compared with LEACH, while the CAKM clustering algorithm and the KMPSO clustering algorithm select the cluster head nodes, they take into account of the geographic location of the nodes in the network and the remaining energy, reducing the chance of low energy node to be elected as the cluster head node, thus can balance the energy consumption of the whole network and extend the network life cycle. The KMPSO algorithm combines the advantages of PSO and K-means algorithm, compared with CAKM algorithm, can find the better global optimum clustering structure, and the life cycle of the network increased by more than 10%.