Hardware-driven compressive sampling for fast target localization using single-chip UWB radar sensor

To design an energy-efficient UWB ranging system, we propose a compressive sampling (CS) technique tightly coupled to a recently proposed hardware. Our goal is to design a system that is robust to high noise and consumes less energy while providing reliable localization. In this work, we first introduce a representation of UWB signals as group sparse signals with the number of groups corresponding to the number of objects in the environment. Also, we design an efficient measurement system that is constructed using low-density parity-check (LDPC) matrix, in order to satisfy several constraints imposed by the hardware: non-negative integer entries in measurement (sensing) matrix, constant row-wise sum of non-zero entries in the matrix, and a unique structure characterized by Kronecker product. To enhance performance, we propose a window-based reweighted L₁ minimization that outperforms other existing algorithms in our simulation. The result shows that our proposed method can achieve reliable target-localization, while using only 40% of the scanning (sampling) time required by the sequential scanning scheme, even in highly-noisy environments.