Uninterrupted Video Surveillance in the Face of an Attack
Jagannadh Vempati, Ram Dantu, Mark Thompson · 2018
Distributed denial of service (DDoS) attacks continue to plague businesses and consumers alike, and due to an ever-growing digital landscape, these attacks are expected to grow in size and complexity. Current mitigation techniques ranging from hours to days are completely unacceptable given the cost and inconvenience these attacks place in our society. This paper puts forth three feedback control mechanisms to minimize the effects of DDoS attacks on real-time traffic. The first, called differentiated services code point (DSCP) Markdown, is a passive approach that uses micro firewall rules to lower the priority of out-of-profile packets while a second mechanism actively drops the out-of-profile packets based on rate and burst size parameters. The third technique uses parallel links when feedback is applied to stabilize the network after an attack has been detected. Results from all three techniques have shown to have a positive effect on real-time traffic. The first two approaches were able to stabilize network traffic in real-time, while the parallel links technique resulted in a slight delay. We validate the feedback mechanisms with our model that was generated using the system identification technique. Results show that the feedback architecture provides a fit accuracy with positive results.