Paper review: Jason Oh
Reviewer: TCP Congestion Control with a Misbehaving Receiver
- This paper attempts to address the vulnerabilities of TCP congestion control to misbehaving receivers.
- The main contribution of this paper is considering the possibility of misbehaving receivers. Although a number of papers have addressed vulnerabilities to problematic senders, few or none have looked at receivers. The authors make a good case for why the problem is important: the population of receivers is extremely large and has both the incentive and the opportunity to exploit this vulnerability.
- TCP congestion control has an as of yet unexplored vulnerability to receiver misbehavior. This vulnerability is a considerable problem that will get only larger.
- This paper rates a 4 because it explores an important and neglected area of TCP congestion control.
- The authors introduce three separate receiver attacks and analyze their impact on current TCP implementations. They then consider the notion that, since it is impractical to prevent every receiver from misbehaving, the best defense to these attacks is to remove the incentive to misbehave. The authors then examine possible changes to TCP that would make receivers be able to only decrease the flow of traffic coming into them. The theory behind the deterrent is good and the authors provide examples. Although they don't actually go forward and implement the changes to TCP, they present convincing arguments that the changes would work.