【摘要】 对网络入侵信号的准确检测,提高检测概率是保证网络安全的基础,传统检测方法难以实现对较低信噪比下的攻击信号的高效定位和检测,无法有效预测入侵信号的局部特征点,导致重采样,虚警概率和漏检概率较高。提出一种基于局部特征压缩采样的网络入侵信号检测算法,构建低信噪比下网络入侵信号模型,采用多普勒频移模糊搜索对入侵信号进行平滑处理,得到入侵信号的局部特征预测量和测量值,为了使得入侵检测适合线性实时处理过程,提高在低信噪比下的检测性能,采用卡尔曼滤波对结果进行修正,采用局部特征压缩采样判断入侵信号的联合特征,实现了对网络入侵信号的局部特征压缩采样检测。仿真结果表明,该算法检测性能较好,检测概率优于传统算法,展示了较好的应用价值,确保了网络安全。更多还原

【Abstract】 The accurate detection of network intrusion signal, improve the detection probability is the foundation of the network security, the traditional detection method is difficult to realize the efficient location of attack than the detection signal and low signal-to-noise, unable to effectively predict the local features of intrusion signal, resulting in re sampling, the probability of false alarm and false detection probability is higher. Put forward a kind of local feature compression and network intrusion detection algorithm based on signal sampling, construct an intrusion signal model under low signal to noise ratio of network, using the Doppler frequency shift fuzzy search intrusion signal smoothing, obtain the local characteristics of the intrusion signal quantity prediction and the measured value, in order to make intrusion detection for linear real-time processing process, improve the low SNR detection performance, using Kalman filter to revise the results, using local feature compression sampling determination combined features of intrusion signal, the realization of the local characteristics of the network intrusion signal compression sampling detection. Simulation results show that this algorithm better detection performance than the traditional algorithm, the probability of detection, showing a good application value, to ensure network security.更多还原