Bayesian Compressive Sensing For Wireless Network Localizaiton

With the rapid development of wireless communication network technologies, the location-based services have caused great concern. Wireless network localization technologies need to sample, process and store large amounts of data, however, existing localization technologies have inadequate of time-consuming and energy-intensive. In recent years, the emerging theory of compressed sensing can combine sampling with compression process, and ignoring the vast majority of redundant data, thus improving the effect of wireless network localization.This paper first elaborated the basics of CS and Bayesian Compressed Sensing (BCS), and we have analyzed the application of wireless communications network based on CS from the application layer, network layer, physical layer. For underwater acoustic communication and complex indoor environments, we analyzes the feasibility that Bayesian compressed sensing are explored to localize the abnormal sensor nodes of underwater acoustic communication network and the mobile terminal indoor.In this paper, Bayesian compressed sensing and regression algorithm are combined to predict the underwater acoustic communication network traffic, which can make use of the relevant of traffic, and can obtain high prediction accuracy with collecting a small amount of traffic, so that we can compare the actual wirh forecat traffic to find the abnormal sensor node with high accuracy.In addition, this paper proposed Bayesian compressed sensing algorithm to localize the wireless network indoor. Emulation results show that this method which based on the sparse prior of mobile terminals can achieve higher positioning accuracy with the same number of measurement. This paper put up with a new way of thinking that combine the Gaussian mixture model with Bayesian framework to achieve wireless network localization, which has better noise immunity and can reduce the number of measurement with the high positioning accuracy.