Research On The Key Technologies Of Cluster-based Wireless Sensor Networks

As a new technology integrated sensor with wireless communications, wireless sensor networks draws more attention to researchers, and becomes a hot topic in information fields. With many advantages such as low energy consumption, low cost, self organization, convenient deployment, flexible network establishment, automatically data gathering and realizable operation under special circumstance, wireless sensor networks is widely used in military fields and lots of civil fields.With the development of wireless sensor networks, number of sensor nodes and size of the network are increasingly enlarged. Traditional flat architecture can not adapt to the network situation. Clustering structure in wireless sensor networks is required to increase energy efficiency, and to realize more effective network management by prolonging the network lifetime. In this dissertation, key technologies of wireless sensor networks with cluster based hierarchical architecture is mainly researched.Typical clustering models, energy consumption models for nodes and data fusing modes are analyzed below. According to the character of road traffic, the design of chain topology and fan topology are presented. Chain topology is used for strip road area, in which sensor nodes are unevenly deployed. The nearer to the sink node, the more sensor nodes should be deployed accordingly. The network is divided into many monitor areas. Information payload frames are adopted, into which collected data monitored from many nodes are placed. Redundant sensor nodes are responsible for the continue forwarding of forwarded frames. Fan topology is used for cross or L-shaped road crossing, while the sink node is far away from it. The application of the algorithms can improve the network energy efficiency after the area type having been choosed. In the research of clustering algorithm, two algorithms are presented. One is an energy efficient strong head clustering algorithm for wireless sensor networks (EESH), another is a dynamic multi-hop clustering algorithm with low energy protection for wireless sensor networks. For EESH, a weight value is created, which integrates energy of nodes, degree and information from neighbor nodes. Based on the weight value, clustering procedure of the network is performed. EESH requires high calculation and storage ability for sink node. It is suitable for the network application with many more node numbers and high network density. For DMC, energy of nodes is divided into four energy levels, weak nodes are protected during the clustering procedure, and a dynamic multi-hop mechanism is established. DMC is suitable for the network application with moderate number of nodes, appropriate node density, and ordinary ability of sink node. The two algorithms above can also realize the evenly consumption of energy of nodes in whole network, and can advance the network energy efficiency. They also can prolong the lifetime of the wireless sensor networks.In the research of network topology discovery mechanism, a cluster merging algorithm with link optimization for wireless sensor networks (CMA) is presented. For the algorithm, every node initiates the cluster formation procedure autonomously and carries out the merging procedure and the optimizing procedure. Later, the topology discovery function is realized to form a hierarchical architecture, and dynamic adjustment is performed during the network operation procedure. The algorithm is applicable to the large area network. The establish time of topology is shorten and the network overhead is reduced in low node density network application.In the research of MAC protocol, a protocol (SC-MAC) based on schedule is designed for clustered wireless sensor network. According to the protocol, wireless channels are assigned by FDMA, and slots in each cluster are assigned by TDMA. By means of scheduling alterable length of slots, energy-saving and low delay character is achieved in the whole network.Finally, a design approach of clustering algorithm simulator for wireless sensor networks is introduced. A platform named CluSim for wireless sensor networks is designed for the simulation of many kinds of clustering algorithms.