Study On Routing Algorithm Based On Uneven Clusteirng For Wireless Sensor Networks

Wireless sensor network(WSN) is a self-organizing and multi-hop network, which iscomposed of a large number of sensor nodes. These nodes can be used to collect all kinds ofuseful data, such as temperature and humidity, and they are distributed in the monitoring arearandomly. By the data fusion, the whole network can reduce or delete the redundant data. Andwhen the sink node collected the data from the sensor nodes, monitors will get it to analysis.Due to the various characteristic such as cheap nodes, rapid deployment, large-scale and so on,wireless sensor network has broad application prospects in environment, intelligent household,military defense and many other fields, and it has drawn widespread attention of researchersboth at home and abroad.Routing protocol is one of the key technologies in wireless sensor network, which has acrucial effect on survival time of the network. The main target of routing protocol is to: reducethe energy consumption of each node, balance the energy consumption among all sensornodes and prolong the lifetime of the entire network. According to the network topology,routing protocol can be divided into flat routing and clustering routing, the main researchobject of this paper is the clustering routing protocol, especially the uneven clustering routingprotocol. The main idea of the unequal clustering routing protocol is to deal with the "hotarea" problem, caused by multi-hop transmission in the equal clustering networks. It groupsthe sensor nodes into clusters of unequal sizes. The main work and innovation points of thisthesis are as follows:1) Firstly，this paper studies the typical clustering routing protocol(LEACH), as well assome other algorithms based on LEACH. It’s essential to grasp the main idea and operationsteps of them. Next, the key analysis is made about the unequal clustering routing protocol,such as UCS, EEUC, etc, which is put forward on the base of EEUC. At the same time, theconclusion is accomplished on the existing problems about them. Then, the paper analyses theclustering routing protocol combined with swarm intelligence algorithm, such as EBUCP,ACOUC, etc. These algorithms have an advantage in selecting the final cluster heads andoptimizing the multi-hop path.2) In order to mitigate the "hot spot" problem in wireless sensor network, a doublecluster-heads routing algorithm based on uneven clustering(D-EEUC) is proposed. Itenhances the election quality of final cluster heads, though improving the election way oftentative cluster heads in EEUC. For this algorithm, the area is divided into near zone and farzone, and there are two kinds of cluster head are selected in each cluster(a master cluster headand a vice cluster head). In the near zone, the vice cluster head is responsible for forwardingdata to share the energy consumption of the main cluster head. However, in the far zone, thevice cluster head is responsible for collecting data and converged data to reduce the cost ofthe nodes’ communication. In the election process of cluster heads, residual energy, densityand other factors should be taken into consideration.3) This paper proposes an uneven clustering routing algorithm (SFLUCR) based onshuffled frog leaping algorithm (SFLA), this algorithm improved local search strategy of the SFLA. Take the advantage of its good capability of global optimization to obtain a globaloptimal solution. The objective function must take the compactness and uneven of clustersinto consideration, as well the node energy. Use competition radius of EEUC, kick part of theoptimal solutions out of the competition of head cluster to avoid the head cluster becoming todense. Then get the final head cluster. To reduce the time of election, this algorithm takesrotation mechanism of head cluster, re-elect the head cluster only in the specified head clusterupdate round.The Matlab simulation results show that compared with LEACH and EEUC, theproposed algorithms are effective in balancing the energy consumption and prolonging thenetwork lifetime. And they can meet the requirements of solving the "hot spots" problem,through grouping the sensor nodes into clusters of unequal sizes.