Implementation of a low area and high-speed spectrum sensor with reconfigurable sensing resolution for cognitive radios

Conventional uniform discrete Fourier transform filter bank (DFTFB) used in energy detection based spectrum sensors provides only fixed sensing resolution. We have recently proposed a progressive decimation filter bank (PDFB) that can provide variable sensing resolutions and adapt to different detection bandwidths for wideband spectrum sensing in cognitive radios. Changing the sensing requirement does not involve hardware re-implementation. In this paper, we present the hardware implementation of PDFB and its comparison with the most popular uniform filter bank, DFTFB, and with the non-uniform Cosine Modulated Filter Bank (CMFB). The implementation results indicate that while providing high degree of flexibility, the PDFB based spectrum sensor is area efficient when compared to its counterparts. It has an area reduction of 39.2% over DFTFB and 57.9% over CMFB.