关键词: DDoS检测, DDoS防御, 实时检测, 自适应检测, Spark Streaming

Abstract: Distributed denial of service (DDoS) attack is an important security threat, the constant improvement of network speed to the traditional detection methods has brought new challenges. Represented by Spark and so on, the big data processing technology brings new opportunity to the completion of high-speed safety detection. This paper proposes an adaptive technique for real-time DDoS detection and defense using Spark Streaming framework. Based on source cluster grouping in sliding windows, and the deviation of proportionate to the source cluster of groups, this paper detects out the DDoS attack traffic trace, and realizes the adaptive rapid and precise detection of DDoS attacks through sensing legitimate network traffic. The experimental results show that the technique can greatly improve detection capabilities, in order to ensure the security of network service performance and the extensibility of detection, this paper provides a feasible solution.

Key words: DDoS detection, DDoS defense, real-time detection, adaptive detection, Spark Streaming