Research Of Directed Diffusion Based On Clustering And Inquiry

Routing protocol is one of the core technologies of wireless sensor networks (WSN), whether it designed in reason influences the performance of wireless sensor networks directly, so designing dedicated routing protocol according to the features and application environment of wireless sensor networks has become the research focus and emphasis.Directed diffusion (DD) is a typical data-centric routing protocol which laid the pivotal foundation for wireless network routing design and is leading the way in data-centric protocol design. However, in this protocol routing establishment mainly depends on the way of flooding, it wastes a lot of energy and time; in addition, as the scale of the network be bigger and bigger, the performance of directed diffusion will drop sharply. In order to address the issues in directed diffusion, directed diffusion based on clustering and inquiry (DDBCI) is proposed in this paper, in this protocol clustering idea and inquiry method was introduced into directed diffusion in order to educe the energy consumption of interest diffusion and initial gradient establishment.The clustering algorithm of DDBCI use node's energy as a parameter of cluster head election, it clusters the network into clusters evenly by the way of iterative cluster head election. Clustering algorithm divides the network into a two-tier communication network, and divides the nodes as three types:cluster head node, gateway node and cluster member node. Single-hop low-level network is composed by cluster member nodes and cluster head nodes, in low-level network cluster member nodes connect to their cluster head node directly. Multi-hop high-level network is composed by cluster head nodes and gateway nodes.Inquiry method of DDBCI is achieved by member information table. Each cluster head node keeps a member information table to record the information of the cluster member nodes in the cluster. When a cluster head node receives an interest message from the high-level network, it doesn't diffuse the interest message into the cluster blindly but inquires the member information it keeps to ensure whether exists some cluster member nodes in target area of the interest message. If exits, the cluster head node will diffuse the interest into the cluster, otherwise, it doesn't do it.This paper describes the process of DDBCI in detail, and simulates it with OMNET++; the experiment result shows that DDBCI has higher energy efficiency than directed diffusion and can keep good performance as the network scale expands.