Low complexity 2-level (FFT-GOERTZEL) spectrum sensing method for cognitive radio

Abstract

Energy detection in frequency domain is a preferred technique for the spectrum sensing and the accuracy of frequency estimation depends on the Discrete Fourier Transform (DFT) size. Instead of computing full length (N) DFT of the whole data, a new two level (coarse-fine) technique for energy detection is proposed. In the first (coarse) level, time averaging of smaller size (L<<N) data blocks of the whole data and its DFT are computed and Neymen Pearson based detection is performed to determine the presence of energy in the subbands. In the second level (fine), Goertzel algorithm is applied to determine the fine estimates in those subbands. Matlab based experiments are performed to verify the performance of proposed method in terms of probability of correct detection and false alarm at different noise levels. Simulation results show that this method can be applied for non-uniformly occupied spectrum also. The complexity of this approach is evaluated and it is 51% computationally more efficient for the considered case. Different windowing methods have been evaluated to be used for better detection performance. It is shown that optimal Discrete Prolate Spheroidal Sequence window helps in minimizing the spectral leakage to the adjacent bands, hence reducing false alarms and improving frequency estimation for Cognitive radio. Adaptive thresholding methods have been applied to the proposed detection method for increasing the reliability of spectrum sensing and combating the problem of noise uncertainity. Directions to extend the work further for case of Multipath fading environment and spatial sensing have also been mentioned in the thesis