Impact of incomplete knowledge on scanning strategy

Security is a fundamental problem facing wireless systems employing spectrum sharing, and thus scanning algorithms are used to detect malicious or illegal activity in such systems. A crucial issue in designing such algorithms is incorporating knowledge about the environment, as well as what knowledge an adversary might have, into the scanning algorithm to improve detection performance. In particular, if such knowledge is initially incomplete, it becomes desirable to adapt one´s knowledge based upon the results of the scanning activities, so as to further improve detection performance. To obtain insight into this problem, we suggest a Bayesian game-theoretical model of bandwidth scanning with learning. We show that such knowledge could change the structure of the strategies employed from distributing effort among all the bands, to band-sharing or even band on/off strategies and improve detection performance. Also, we have shown that a lack of information for the scanner compare to the adversary makes the scanner strategy more sensitive to the information he has.