Reconfigurable Polyphase Filter Bank Architecture for Spectrum Sensing

Abstract

This paper presents a brief tutorial and background on implementing filter banks for spectrum sensing. It discusses the advantages of this approach over standard FFT-based spectral estimation. A general architecture for implementation of filter banks on FPGAs is then presented, exploiting heterogeneous resources. It takes advantage of the fact that subband filters run at a reduced sample rate, and hence can share the same computational resources. We show how to facilitate this sharing of resources by saving state for each subband. The architecture is fully reconfigurable, allowing sensing parameters to be changed at runtime. Resource usage figures are given, showing the efficiency of the architecture. We finally discuss how this architecture can be adapted to signal reception.