Efficient Design and Implementation of Energy Detection-Based Spectrum Sensing

摘要

Cognitive radio allows opportunistic spectrum access by finding and using temporally and spatially unused radio channels to enhance the utilization of scarce radio spectrum. Various spectrum sensing techniques have been proposed to detect the availability of unoccupied radio channels. In this article, we first briefly discuss the merits of various spectrum sensing techniques. We then present an efficient hardware architecture for energy detection-based spectrum sensing. We discuss various design parameters and features, such as transform length, window selection, window overlap, ensemble averaging, and dynamic thresholding for efficient realization of the designed spectrum sensing module. Then, an efficient hardware architecture for an optimal energy detection-based spectral estimator is presented. To reduce the computational complexity, we utilize a polyphase filter structure. For the high-performance implementation of the proposed spectrum sensing architecture, we use cutset retiming, and for the compact realization, we use the time-multiplexing technique to reuse hardware resources. We model our spectrum sensing module in both floating-point and fixed-point representations using our custom-developed library of numerical operations. The synthesizable parameteric model of the energy detection-based spectrum sensing module is developed in Verilog hardware description language. The architecture of the designed spectrum sensing module is implemented on a Xilinx Virtex-7 field-programmable gate array and operates at 327 MHz. The cycle-accurate bit-true hardware simulation results are verified against its fixed-point software simulation results. An ASIC architecture of the designed spectrum sensing module is estimated to occupy 5.65mm(2) and dissipate 9.10 mW from a 1.05-V supply while operating at 200 MHz in a standard 32-nm CMOS technology for a 200 MHz bandwidth.