Friendly Jamming Based Intelligent Privacy Protection Technology Study On Visible Light Communication

With the improvement of computer decoding ability,the visible light communication privacy protection is concerned by researchers.Friendly jamming technology can prevent eavesdropping through cooperative communication,however,it is difficult to optimize jamming strategy in the actual dynamic communication environment,resulting in privacy leakage.Therefore,this paper studies the friendly jamming and beamforming mechanism of visible light communication system to improve its privacy protection level.Firstly,reinforcement learning based friendly jamming scheme for visible light communication is proposed.We use reinforcement learning technology to optimize the friendly jamming strategy with the channel state,energy loss,bit error rate and other information,and improve the secrecy rate and accuracy in the friendly jamming cooperative dynamic communication.On this basis,a friendly jamming scheme based on deep reinforcement learning is proposed,which uses neural networks to extract channel features and compress system state space,and designs memory dictionary to store historical communication experience,strengthen the correlation between similar channel states,accelerate learning process and further improve the system confidentiality rate.Simulation results based on four friendly jammers show that the proposed reinforcement learning based jamming scheme can improve the security rate by 15.4%and reduce the bit error rate by 27.1%compared with the existing robust jamming scheme.Deep reinforcement learning based jamming scheme can further reduce the energy loss by 39.1%and the bit error rate by 83.6%.As for multiple input single output visible light communication system,a reinforcement learning based beamforming control strategy is proposed to solve the problem of the traditional beamforming privacy protection scheme,optimize the beamforming strategy based on channel state,bit error rate,secrecy rate and other information,and promote system confidentiality and reliability of communication.To improve the optimization speed and performance,a deep reinforcement learning based scheme is proposed using the deep neural networks to simulate the high-dimensional continuous beamforming strategy space.Simulation results based on 9 Light Emitting Diodes(LEDs)show that the bit error rate of the proposed reinforcement learning based beamforming scheme is reduced to one tenth of that of the contrast scheme,and the secrecy rate is increased by 91.34%.The deep reinforcement learning based beamforming scheme increases the secrecy rate by 29.7%,and reduces the bit error rate by 33.3%compared with the reinforcement learning scheme.