DDoS Detection

With the development of smart grids, power grid systems are gradually becoming morecomplex. This poses new challenges for ensuring the security of power grid systems.The prevailing approach to network intrusion detection relies heavily on manuallyengineered features, often requiring rigorous expertise and struggling to accommodate adiverse array of attack types. In response to this challenge, we employed a windowingtechnique to segment network traffic data into manageable samples. These samples aresubsequently input into the Informer network for feature extraction and classification,facilitating intrusion detection. Our proposed algorithm simultaneously considers boththe temporal information of sessions and overall attention information, autonomouslylearning features from traffic data. Experimental evaluations using CICIDS-2018network traffic data demonstrate the algorithm’s effectiveness in DDoS attack detection,yielding promising results.