Game-based Secure Sensing for the CRN

To enhance the security of spectrum sensing in centralized cognitive radio networks, we introduce a novel trust game-based model. We aim at encouraging the secondary users to send correctly their local sensing outcomes to the data fusion center. Our proposal ensures both the attacks detection and the punishment of malicious users. Indeed, our proposed punishment function intends to exempt the malicious secondary users with faulty spectrum sensing from participating in the spectrum sensing phase. Extensive simulations illustrate that the proposed game-based model outperforms the AND-rule and OR-rule models in terms of correct decision, missed-detection, false alarm, error probability and throughput.

I. Introduction

To enhance the efficiency of spectrum utilisation and resolve the spectrum under-utilization problem, the concept of cognitive radio [1] was invented by J. Mitola, III which allows the Secondary Users (SUs) to share the spectrum with the Primary Users (PUs). One of the most important feature in cognitive radio network is the spectrum sensing which decides the channel availability. In literature, many spectrum sensing techniques are implemented such as the energy detection [2], cyclo-stationary based detection [3], matched filter [4], etc. However, the presence of malicious users can degrade the network performance by affecting the spectrum sensing. In the Spectrum Sensing Data Falsification (SSDF) attack [5], the malicious users launch harmful attacks [6] intentionally or unintentionally (hardware or software failure, shadowing problems, etc). In the cooperative spectrum sensing, the malicious users aims to affect the final decision taken by the Data Fusion Center (DFC) by reporting faulty spectrum detections. The major malicious users intention is to deprive selfishly the SUs from accessing to the white space (available channel).