CAJ | 학술논문

提出采用Lorenz混沌系统异步跟踪优化期望最大化高斯混合模型算法实现对低信噪比下深度伪装的网络攻击信号最优检测。通过提取待检测网络数据流参数向量和正常数据流参数向量的差值为特征，使用高斯混合模型并与期望最大化算法相结合，设计Lorenz混沌异步跟踪检测算法，对网络数据流进行建模和检测。仿真结果表明改进的检测算法能有效去除不是攻击信号的伪峰，相比Hough变化检测算法，能更加正确地检测非法攻击信号，信噪比为-15dB下，不同异步攻击中的检测概率就能达到100%，实现检测性能最优，尤其适用于信噪比极低的深度伪装网络攻击环境中对攻击信号的检测。研究成果为网络安全防御及应用具有巨大的理论参考价值。
제출채용Lorenz혼돈계통이보근종우화기망최대화고사혼합모형산법실현대저신조비하심도위장적망락공격신호최우검측。통과제취대검측망락수거류삼수향량화정상수거류삼수향량적차치위특정，사용고사혼합모형병여기망최대화산법상결합，설계Lorenz혼돈이보근종검측산법，대망락수거류진행건모화검측。방진결과표명개진적검측산법능유효거제불시공격신호적위봉，상비Hough변화검측산법，능경가정학지검측비법공격신호，신조비위-15dB하，불동이보공격중적검측개솔취능체도100%，실현검측성능최우，우기괄용우신조비겁저적심도위장망락공격배경중대공격신호적검측。연구성과위망락안전방어급응용구유거대적이론삼고개치。
The improved and new the Lorenz chaotic asynchronous tracking network attacking signal detection algorithm was proposed based on the expectation maximization algorithm and Gaussian mixture model, it was designed for detecting the attacking signal in low SNR with the optimum detection performance. Combining the Lorenz chaotic system with the asynchronous control, the feature was extracted by calculating the difference value between the normal flow data and the waiting detection network data flows. The Lorenz chaotic asynchronous tracking detection algorithm was proposed successfully. Simulation result shows that the improved detection algorithm can eliminate the false peak in the detection spectrum, and it can detect the abnormal attacking signal more accurately than the Hough transform algorithm, the detection probability can reach 100% when the SNR is-15dB. The detection optimization is realized in low SNR. It adapts to the low SNR and deep disguise network attacking environment. Research result provides huge reference value in network security defense in application and theory.