LAUSR

The received signal strength (RSS)-based target localization is an important field of research with numerous applications in wireless sensor networks. By exploiting the sparsity of localization, the compressive sensing (CS) can be applied to develop an effective localization framework for multiple targets. However, the existing CS-based method considers localization directly from real-valued sensor measurements, which implicitly assume that the measurements have infinite bit precision. When the assumption is violated, the localization performance will deteriorate dramatically, especially at low quantization bit rates. In this paper, we first design a quantizer for efficiently processing the raw RSS measurements and then develop a novel Bayesian CS framework for estimating target locations from the quantized measurements. By modeling the quantization errors as independent random variables, the non-linear localization problem is formulated in a linear CS observation model. Following this idea, we solve the problem from a Bayesian perspective and design some sophisticated prior distributions to guarantee the performance. To address this, we resort to the variational Bayesian inference methodology and propose a novel iterative algorithm for jointly estimating target locations and dealing with quantization errors. The extensive simulation results demonstrate the superiority of the proposed algorithm in comparison with the state-of-the-art methods.