Modelling and Mitigating Spectrum Sensing Non-cooperation Attack in Cognitive Radio Network

Abstract

Collaborative spectrum sensing (CSS) is known to improve spectrum sensing performance in Cognitive Radio Network. In CSS, secondary users participate by sharing their local sensing results. They participate in the sensing process at their own cost, i.e., they expend some amount of energy and time for sensing and sharing. But, a selfish user may refrain from collaborating in the spectrum sensing process in order to save up energy, which results in improper sensing. While this problem is widely known, we call this as the spectrum sensing non-cooperation (SSNC) attack for easy reference. In this paper, a collective action prisoner's dilemma game is used to model the SSNC attack. To handle this attack, repeated game punishment mechanisms, namely Tit-for-Tat and Grim strategies are used. In addition, modified Tit-for-Tat and modified Grim strategies are proposed to handle this attack in the presence of reporting channel error.