Enhanced CS-Based Device-Free Localization With RF Sensor Networks

摘要

With the inherent sparsity of spatial positioning, the compressive sensing (CS)-based device-free localization (DFL) methods have been proposed in recent studies, which can reduce the sampling requirements. However, model errors and grid errors in CS-based DFL methods are inevitable due to using the fixed dictionary and discretization, but these errors are ignored by most existing CS-based DFL works. To solve these issues, this letter proposes an enhanced CS-based DFL scheme, which first exploits the increment dictionary learning technique to overcome the problem of the dictionary mismatch and then utilizes the quadratic programming approach to mitigate grid errors. Experimental results verify the better performance of the proposed method in terms of localization accuracy compared with the state-of-the-art CS-based DFL algorithms.