Quantized cooperative spectrum sensing for Cognitive Radio

Cognitive Radio (CR) is proposed to maximize spectrum utilization. When the licensed user is not using the spectrum, CR will then use it to communicate among each other. Hence, spectrum sensing, a procedure to observe primary user´s existence, is important to CR. In this paper, we investigate the energy detection based cooperative spectrum sensing technique as it offers low complexity and significantly improves the sensing performance for CR. Conventional soft decision combining (SDC) scheme assumes that CR´s local observation is perfectly forwarded to the fusion center. This forwarding procedure requires bandwidth, and increasingly so with more sensing nodes. By carefully applying quantization to this local observation, we could significantly reduce the number of bits required for this communication and thus reduce the overall system overhead. Uniform and non-uniform quantization approaches for SDC are investigated. We further propose a low complexity scheme based on an approximation of the cumulative density function. Simulation results show that these schemes can achieve a good detection performance comparable to the conventional SDC scheme, with reduced communication overhead.