A Study Of Efficient Fft Pruning Algorithms For Cognitive Radio Networks

In today's communications environment, the demand for high speed wireless applications continues to put a strain on the radio spectrum resources. This demand has driven the development of dynamic spectrum access (DSA) techniques that enable more efficient utilization of the scarce frequency resources. Cognitive radio (CR) technology is constantly being improved to enable dynamic sharing of idle spectrum between different classes of users. Orthogonal frequency division multiplexing (OFDM) is an ideal transmission technique for spectrum sharing as it allows different users to simultaneously access different parts of the spectrum. Non-contiguous OFDM (NC-OFDM) then enables a single user to collectively access non-contiguous portions of the available spectrum. This thesis seeks to make a contribution to the improvement of the efficiency of the NC-OFDM transceiver as used in CR networks, by improving the efficiency of the Fast Fourier transform (FFT)/IFFT blocks that are critical components of any OFDM system.To this end, FFT pruning as a means of improving the efficiency of the FFT is studied. Various FFT pruning algorithms suitable for arbitrary zero input distributions such as those typical in CR networks are examined to determine their suitability for use in NC-OFDM transceivers. A previously proposed FFT pruning algorithm is modified to ensure that it produces outputs that are the discrete Fourier transform (DFT) points of the input, for arbitrary zero input distributions. It is shown by simulation that this algorithm calculates the FFT of the inputs in a shorter time than the conventional FFT, and other FFT pruning algorithms used for arbitrary zero input distributions. A suitable FFT pruning algorithm for CR networks is therefore developed.