Abstract:
This paper presents detection method based on 1-order AR model. Because of the low precision 1-order AR model owns, it proposes a method by substituting high-order AR model for 1-order one to make it more precise. Because setting the best parameters of the high-order AR model is an NP-complete problem, it applies genetic algorithm to search the better ones. Experiment shows that the high-order AR model is more efficient and precise for network anomaly detection.
Key words:
network anomaly detection,
AR model,
genetic algorithm
摘要: 介绍基于1阶AR模型的检测方法,用高阶AR模型代替1阶AR模型以增加其精度。鉴于高阶AR模型最优参数的确定是个NP完全问题,利用遗传算法搜索其参数的近似最优解,通过实验验证高阶AR模型对网络异常检测的有效性和精确性。
关键词:
网络异常检测,
AR模型,
遗传算法
CLC Number:
HUANG Qiang-sheng; CHENG Jiu-jun; KANG Qin-ma. Network Anomaly Detection Based on High-order AR Model[J]. Computer Engineering, 2010, 36(3): 174-176.
黄强盛;程久军;康钦马. 基于高阶AR模型的网络异常检测[J]. 计算机工程, 2010, 36(3): 174-176.