Transmission and Energy Efficient Clustering Method for Wireless Sensor Networks Using Compressive Sensing

Abstract

The high volume of data transmission in a wireless network can be reduced significantly using Compressive Sensing (CS). Compressive sensing can cause a breakthrough in the wireless sensor network (WSN) field. However, the problem with a high number of data transmissions remains the same as using pure CS. The total number of transmissions does not drop off much significantly. Using compressive sensing, we use a sparse representation of a signal where most of the sample values of that signal are very low in a specific transform domain. Based on a threshold value, we can ignore a certain amount or percentage of these samples and send the remaining data through a transmission network. Later on, we can reconstruct the sent signal, i.e., audio, video, or image, from these transmitted samples with a significant level of accuracy at the receiver end. Hybrid CS can be used instead of pure CS to significantly reduce the total number of data transmissions for the same data or signal. In previously used methods routing trees were being used on these transmissions using the CS method. Now in this proposed implementation, a Clustering method is being used in the Compressive sensing technique. The entire area where the sensor nodes transmit some data are spread or located is divided into clusters. These clusters are hexagonal instead of circular to avoid overlapping cluster zones. In each cluster, its node is elected as its Cluster Head (CH). It can be selected based on specific parameters like battery capacity, power handling capacity, and bandwidth. In that cluster, existing nodes do not need to choose new cluster heads very frequently. One cluster head will function for a very long time, which will, in turn, save time and energy for nodes of running the cluster head selection algorithm over and over again. In each cluster, the number of sensor nodes is less than the total number of nodes, so the sensor nodes send their data to their respective cluster heads without compressive sensing. However, much higher compression is needed when the number of data sent to cluster heads sinks. So the data is sent to the sink from Cluster heads using compressive sensing. First, a relationship needs to be found out between the cluster size and the number of transmissions. This method aims to minimize the total number of transmissions, including both inter-cluster and intra-cluster, as much as possible. The optimal size of a cluster needs to be found. Then a centralized clustering algorithm is required to be implemented using this optimal cluster size. Furthermore, in the end, a distributed implementation of this algorithm has to be devised.