The Research On Leaky Cable Perimeter Intrusion Detection Algorithm

The perimeter intrusion detection system is a system that protects the security zone,promptly and accurately detects the intrusion and reports the intrusion location.Traditional perimeter security systems such as vibration cable alarm system,video surveillance system,electronic fence,etc.have drawback of low location precision,high false alarm rate,poor antiinterference ability.However,the intrusion detection system that based on leaky cable sensor are widely used in military bases,airports,prisons,and other places for its advantages of low false alarm rate,strong concealment,low environmental impact and good terrain adaptability.At first,this paper studied the leaky cable sensor in the leaky cable security system.Aiming at the problem that leaky cable communication system and leaky cable intrusion detection system can not share the same type of leaky cable,a detection system based on RMC leaky cable which is used for communication is proposed.After that,a hardware test platform is built and the test results show that this type of leaky cable could perceive intrusion effectively.And then,in order to solve the problem that the current leaky cable security system lacks the positioning ability,this paper proposes the principle of intrusion target location based on golay code autocorrelation technology after verifying the feasibility of the RMC leaky cable for security.On this basis,system modeling and simulation are carried out and intrusion detection and positioning prototype system is built.The experiment results demonstrate that the system could detect the intrusion disturbance accurately and locate the target preliminarily.At last,as the key indicators of the overall performance of the leakage cable perimeter security system are mainly the positioning accuracy,false alarm and false negative rate,this paper designed a whole leaky cable perimeter digital signal processing process after the deeply study of localization algorithm and artificial intelligence algorithm,SVM in pattern recognition technology is applied in the signal process.The rate of false positives have reduced and the positioning accuracy get promoted by constructing classification model with the intrusion and no intrusion datasets,these datasets include all kinds of interference signals and signals of every sub-bin which are acquired in actual outdoor perimeter area.Due to the limitations of SVM in solving multi-class problem and the full consideration of intrusion signal characteristics in time domain,this paper introduced RNN deep learning algorithm for precise positioning,results from a field test prove the algorithm had high positioning accuracy,low rate of false positives and false negatives.