Robustness improvement against noise uncertainty by cooperative spectrum sensing

Multiuser collaboration can significantly improve the performance of spectrum sensing in cognitive radio networks. By admitting the noise power estimation error and fading/shadowing effects in practical implementation, this paper presents another look at the energy detection based cooperative spectrum sensing. The worst-case collective probability of miss detection due to the noise uncertainty is derived under the spectrum utilization constraint, and the numerical results over Log-normal shadowing channel confirm that the performance deterioration can be compensated for if more secondary users take part in the cooperative sensing, since there will be a higher chance for a user with its instantaneous SNR beyond the SNR wall. Further, the effect of noise uncertainty on cooperative spectrum sensing is quantized by the equivalent "SNR wall" as in the local sensing. Typical K out of N fusion strategies are considered and compared in terms of the equivalent SNR wall reduction relative to the local case. As is demonstrated by the results, the robustness gain to noise uncertainty brought by cooperative spectrum sensing can be adjusted to any required level if sufficient users and a reasonable fusion strategy are met, and among the three typical fusion strategies, OR rule performs the best, Major rule is the next, and AND rule is the worst.