Research On The Attack Strategy Of Underwater Acoustic Communication Networks Based On MAB Theory

With the development of underwater acoustic communication technology,the research on electronic attack of underwater acoustic communication networks is paid more and more attention.Compared with land attacks,underwater attacks often face the challenges of high energy consumption,inconvenient charging,unknown network topology and strong uncertainty,etc.The existing attack methods can't be directly used in the complex underwater environment.In addition,the security mechanisms of underwater acoustic communication system such as adaptive routing and multi-path routing will also hinder underwater network attacks.Therefore,this paper will optimize the target node selection and physical layer parameter configuration for underwater attack,and propose an efficient,low-cost and highly adaptable underwater attack strategy.Firstly,the communication principle of underwater acoustic communication networks and the relevance between network nodes are studied according to the transmission characteristics and network structure of underwater acoustic communication.and then the vulnerability of underwater acoustic communication networks is analyzed for the particularity and complexity of underwater attack environment;the advantages and limitations of existing underwater attack methods are analyzed according to the different network architectural layers.After that,the optimization of underwater attack is studied based on the typical Multi-armed Bandit(MAB)decision models.Secondly,the problem of underwater attack node selection with unknown topology is studied,a distributed hybrid attack strategy is proposed,and an attack model based on MAB learning is constructed.A Virtual Expert-guided Exponential Learning(VEEL)algorithm is designed for the complex calculation and time-varying problem of underwater attack.The algorithm does not rely on prior network topology information,can make full use of experience-based virtual information and probabilistic prediction to effectively improve the exploration frequency and cost of attacks,and has the advantages of real-time learning and self-update.Simulation results verify the well adaptability and robustness of the proposed algorithm.Finally,the optimization problem of underwater jamming attack with limited energy is studied,and an adaptive control jamming attack strategy is designed to optimize the distribution of limited attack energy.In view of the complexity and unpredictability of underwater attack environment,an MAB decision model is constructed to optimize the physical parameters selection of jamming signal.With the broadcast characteristic of underwater acoustic,a multi-link joint optimization mechanism is designed,an attack reward function is constructed through multiplex feedback information generated by underwater acoustic jamming to alleviate the energy constraint problem of underwater attacks;and then through real-time interaction with the environment,an Underwater Jamming Learning(UJL)algorithm based on MAB strategy is proposed to optimize the success rate of jamming attack.