Low-Complexity Compressive Spectrum Sensing for Large-Scale Real-Time Processing

To overcome the challenges in high-speed sampling and processing of real-time spectrum measurement, compressive sensing (CS) theory has been implemented in wideband spectrum sensing. Moreover, to take full advantage of CS, the nonconvex l(nu)-norm minimization algorithms are employed to reconstruct the wideband signals from compressive samples. However, solving these algorithms usually leads to relatively high computational complexity and sensing cost, especially when the dimension of wideband signals is high. Therefore, we propose a low-complexity compressive spectrum sensing algorithm that is suitable for large-scale real-time processing problem. The numerical and experimental results demonstrate that the proposed algorithm achieves the fast convergence speed and keeps the same accurate signal reconstruction with reduced computational complexity, from cubic time to linear time.