Rank Reversal Effect on Multicriteria Ranking for Cooperative Spectrum Sensing under Shadowing and Noise Uncertainty

Abstract

Rank reversal is the main challenge of multi-criteria ranking method. Currently, CSS adopts the multi-criteria ranking to determine the quality of SUs. However, the effect of rank reversal to the performance of CSSs based on multi-criteria ranking have not been investigated yet. Therefore, in this paper, we evaluate the rank reversal effect on the performance of CSSs based on multi-criteria ranking under poor environment such noise uncertainty and shadowing effect. As a result, AMT can avoid the rank reversal occurrence and presents rank reversal robustness under decision threshold adapting strategy. On the other hand, based on fixed threshold, AMR suffers from rank reversal. Moreover, we compare the performance of AMR and AMT to IMA and K-of-N under different number of SUs. It can be noticed that AMT and IMA do not suffer from the number of SUs by determining the optimal number of SUs to determine the channel availability. On the contrary, the performance of K-of-N and AMR degrade under an increasing in the number of SUs once the unreliable SUs are taken into the account in order to determine the channel availability.