Abstract

A TCP flow increases its sending rate linearity until a packet loss is detected. The lost packets can be seen as the cost paid by the method of probing available network bandwidth. A router can decide whether to drop a received data packet only based on its locally available network resource without considering the sender of the packet is responsible or not. Therefore, such a scheme does not motivate end users to use congestion control mechanism. Accordingly, TCP flows have to endure packet loss. In this paper, we propose a novel congestion control mechanism named as probe follow congestion control (PF-CC) mechanism, with which users can be classified into two categories: probers and followers. Probers are those users that are responsible for probing available network resources. Followers are those users that regulate their behavior based on the probing results by probers. Thus, a follower has a small probability of packet loss since there is no available network resource. The routers drop a packet by considering the behavior of its source. More specifically, as an example, a router can reward those responsible users by rarely dropping their packet and punish those irresponsible users by dropping their packets severely. Therefore, the PF-CC mechanism can motivate end-users to conduct end-to-end congestion control. Specifications of the mechanism are presented and also its performance is evaluated using extensive simulations and analysis.

Full Text
Published version (Free)

Talk to us

Join us for a 30 min session where you can share your feedback and ask us any queries you have

Schedule a call