Cooperative Spectrum Sensing with Feedback

One of the main objective of spectrum sensing is to detect spectrum opportunity quickly and accurately. Both of the two aspects can be improved by hardware upgrading and software optimization. In this paper, we concentrate on sensing accuracy, which, differs from previous solutions, will be enhanced by modification of network topology. We expand the traditional parallel fusion network (PFN) to parallel feedback fusion network (PFFN), in which local cognitive sensor makes its local decision based on previous global decision and its recent observation. We then derived optimal decision rules for PFFN. Moreover, based on PFFN, sensing history of local cognitive sensor is introduced and studied. Theoretical analysis and experimental results show that PFFN outperforms the PFN in smaller system error probability under the same SNR. We also proved that PFFN and PFFN with sensing history (PFFNSH) emphasize on different aspects when certain condition holds in real world applications. Finally, a general PFFN framework is proposed and discussed.