作者: Tomás Navarrete , Carlos Orozco , Juan Manuel García
DOI:
关键词:
摘要: We present an approach to anomaly detection based on the construction of a Hidden Markov Model trained processor workload data. Based load measurements, HMM is constructed as model system normal behavior. Any observed sequence measurements that unlikely generated by then considered anomaly. test our taking real data mail server construct and we it under several experimental conditions including simulated DoS attacks. show some evidence suggesting this method could be successful detect attacks or misuse directly affects performance.