Trust-based bandwidth management for critical flows

Abstract

In this thesis, we focus on the bandwidth allocation on the bottleneck link and provide several allocation schemes to minimize the potential damage caused by flooding attacks. Assume that we know which flows are critical flows, and all other flows on the bottleneck link are considered as suspicious flows [4], when the delivery path is under attacks. The proposed schemes try to provide more bandwidth to critical flows and reduce damages as much as possible. Using these schemes, the system can properly allocate the bandwidth among critical flows and suspicious flows on the bottleneck link. At the same time, the congestion of the bottleneck link will be identified and the non-critical flows are tagged as suspicious flows. Once the congestion of the bottleneck link is identified, the bandwidth allocation of suspicious flows is reduced based on the proposed methods. In other words, we can supply proper bandwidth to critical flows, such that critical flows will suffer less damage. Ideally, if a suspicious flow can be determined as an attack flow, this flow will be dropped completely. As a result, the whole system will suffer less damage.