Intelligent cognitive radio jamming - a game-theoretical approach

Abstract

Cognitive radio (CR) promises to be a solution for the spectrum underutilization problems. However, security issues pertaining to cognitive radio technology are still an understudied topic. One of the prevailing such issues are intelligent radio frequency (RF) jamming attacks, where adversaries are able to exploit on-the-fly reconfigurability potentials and learning mechanisms of cognitive radios in order to devise and deploy advanced jamming tactics. In this paper, we use a game-theoretical approach to analyze jamming/anti-jamming behavior between cognitive radio systems. A non-zero-sum game with incomplete information on an opponent's strategy and payoff is modelled as an extension of Markov decision process (MDP). Learning algorithms based on adaptive payoff play and fictitious play are considered. A combination of frequency hopping and power alteration is deployed as an anti-jamming scheme. A real-life software-defined radio (SDR) platform is used in order to perform measurements useful for quantifying the jamming impacts, as well as to infer relevant hardware-related properties. Results of these measurements are then used as parameters for the modelled jamming/anti-jamming game and are compared to the Nash equilibrium of the game. Simulation results indicate, among other, the benefit provided to the jammer when it is employed with the spectrum sensing algorithm in proactive frequency hopping and power alteration schemes.

Strategic Power Allocation With Incomplete Information in the Presence of a Jammer

“…These drawbacks, to a large extent, limit the practicality and applicability of any spectrum sharing protocol that takes the malicious intent of a rational user into consideration. It should also be pointed out that there exist in the literature works on jamming that study practical problems related to the detection or mitigation of jamming attacks in the context of Bayesian games [28], [29]. However, they do not investigate the interactions between players in terms of optimal power allocations, which is an effective technique to mitigate jamming attacks.…”

“…Specifically, the authors in [28] formulate a malicious node detection game and a post-detection game to isolate it in wireless networks. Furthermore, in [29], the authors focus on mitigating the jammer's effect through channel hopping and power alteration from a predefined set. In [30], we investigate a Bayesian power allocation game in which the SU (assumed to have incomplete information of the location of the jammer) and the jammer (assumed to have complete knowledge of the game) strategically allocate their powers over N channels that are assumed to always be available and follow the simplified path-loss propagation model.…”

Application of reinforcement learning for security enhancement in cognitive radio networks

“…(b) RL enables SUs to make decisions based on a series of actions made, with the notion of maximizing long-term reward, which is more efficient. Using RL helps to solve the performance efficiency issue associated with the game-based approaches, particularly one-shot or repetitive games (e.g., potential game and matrix game), that have been applied to address jamming and primary user emulation attacks [24,28,29]. (c) RL enables SUs to explore new operating environment and exploit the knowledge gained so far.…”

“…In practice, an issue is that, obtaining the model is a complex procedure that requires high amount of processing and storage capabilities. The issue of obtaining an appropriate model is faced by many AI schemes, such as game theory (which has been applied to address jamming attacks [24][25][26]) and belief propagation (which has been applied to address primary user emulation attacks [27]). Using RL helps to solve this issue since RL can be applied in a modelfree manner.…”

A Learning Method for Secondary Users to minimize the effect of Jamming in Cognitive Radio Wireless Sensor-Networks State of the art in CR-WSN security

“…The game can now be modeled as a Markov Decision Process (MDP). Most literature considers channel hopping as the primary defense against jammers, but another defense strategy, manipulating power of transmission has been recently explored [11]. The focus is mostly on channel hopping.…”