An optimal reputation-based detection against SSDF attacks in industrial cognitive radio network

With the increasing development of industrial wireless technologies and standards, the scarce spectrum in the industrial, scientific, and medical (ISM) band has been extremely overcrowded, which can be mitigated by harvesting more spectrum in licensed bands with the emerging cognitive radio technology. In industrial cognitive radio networks (ICRNs), security is one of the most important problems. Spectrum sensing data falsification (SSDF) attacks is one of major challenges for cooperative spectrum sensing (CSS) in ICRNs. Malicious users send fake decisions to mislead the fusion center for exploiting the spectrum source. And smart malicious users launch attacks with small probability which is difficult to be detected by existing detection scheme. To address this issue, we propose an optimal reputation-based detection scheme with considerations of the attack probability. Moreover, the threshold of reputation value is specially designed to adapt to the varying attack probability which can be estimated by past performance. To estimate the attack probability, we use the maximum likelihood estimation method in certain CSS rounds. Numerical results reveal that the proposed detection scheme performs better than existing reputation-based detection schemes.