Wireless Sensor Network Cluster Optimization Algorithm

Due to recent advances in micro-electro-mechanical systems (MEMS), wireless sensor networks (WSNs), which are mostly used to collect data, have received a lot of attentions both in military and civil fields. Generally, such systems consist of mass tiny nodes with sensing, data processing and wireless communication modules on them. These nodes collaborate with each others in a self-organized way, to send the sensed information in the vicinity back to the sink node through the network effectively. Since this kind of system is able to cover large sensing fields while guaranteeing high quality, it is widely used in military sensing, security, environmental monitoring, traffic surveillance, medical treatment, building and structures monitoring and even anti-terrorism.However, sensor nodes are usually small devices with limited on-board energy supply and it is impossible to replace the batteries for them, the network lifetime become one of the key performance index of WSNs. So far clustering is supposed to be an efficient way to perform energy management and extend the network lifetime. In this paper we begin with the clustering in WSNs and try to enhance the performance of the network:First, we analyze the unique characteristics that make wireless sensor networks different from other traditional wireless networks, e.g. Ad-Hoc. It is followed by the analysis of current routing algorithms in WSNs, including flat routing and cluster-based routing. We try to conclude the pros and cons of these routing algorithms and design our novel approach for clustering nodes based on them;Second, to prolong the lifetime of the network and enhance the efficiency of energy utilization, we propose a new cluster-based routing algorithm, which is called time delay based routing algorithm. The main idea of our approach is to use the cluster head (CH) declaration delays to characterize the qualification of each node to be a cluster head, where the delay time of each node differs from each other according to its residual energy. Our approach manages to achieve the load-balance by guaranteeing the fairly uniform and optimum cluster distribution while incurring low overheads.Third, as the sensor nodes in WSNs are usually densely deployed, their overlapped sensing areas produce huge number of redundant data in the networks, which not only increases the energy consumption of the network dramatically but also makes the probability of MAC collisions much higher in the wireless channel, thus lowers the efficiency of energy utilization. In this paper we propose a k-neighbors sleeping algorithm, which lets a part of the nodes being in inactive states to decrease the redundancy of the collected data and prolong the network lifetime.At last, we conclude our work and give some suggestions for the future work.