We infer that executing this algorithm with appropriate partitioning of the dataset in a parallel environment provides better classification accuracy. Our parallel cumulative ranker (PCR) algorithm can rank the attributes of a dataset for cost-effective classification of network traffic. Distributed feature selection in intrusion detection systems using parallel computing can help build a cost-effective defense module. Network traffic classification to detect DDoS attack traffic in real-time is challenging in the context of current high-speed networks and systems, and voluminous Internet traffic.
The effectiveness of Granger Causality helps us in confirming TCP flooding attacks by analyzing causal information, in near real-time. During TCP DDoS flooding attack, the number of spoofed IP addresses and the number of open ports used by the attacker usually follow a causal relationship. A triggering relation can be assumed to be a causal and temporal relationship among the events. Malicious software events are usually stealthy and thus challenging to detect. Self-similarity nature of network traffic, we differentiate the legitimacy of any traffic irrespective of the network traffic protocol. Self-similarity exists in ethernet traffic. Traffic and low-rate attack traffic resemble bursty legitimate traffic and normal The challenge is to find the difference between DDoS attack traffic of all typesĪnd legitimate traffic at the earliest, even though high-rate attack We believe that such comparisons help build defense modules that are better than state of the art. We perform detailed pros and cons analysis of existing methods and systems and compare them with our work throughout the thesis. We also enumerate issues and research challenges in this evolving field. We present a comprehensive study of methods and systems introduced recently for the protection of network resources from intrusion. thesis work addresses the problem of detecting anomalies in network traffic at an early stage in near real-time with a low false alarm rate and preventing known attacks and their variants, using statistical and machine learning techniques. The presence of a large number of resources organized densely is a key factor in attracting attacks, including DDoS attacks. Intensive computing infrastructure, but possibly also a more insecure digital world. The future of the Internet is predicted to be more complex, with more – Alan Turing, Father of Computer Science.ĭefending a computer network 24/7 from intrusion is an important problem of “A computer would deserve to be called intelligent if itĬould deceive a human into believing that it was human”
0 kommentar(er)
