A Credibility-Based Congestion Control Method

A novel congestion control method works at the network layer, which is credibility-based and under supervision is proposed. The method supposes that network elements are all well-reputed, which means that most of them will control their transmission rate consciously and the total demanding bandwidth will not exceed the capacity of the network at high probability. The scheme is also capable to deal with situations where a few elements violate the rules and thus lead congestion. The whole network is divided into several domains independently in the scheme, each domain includes a congestion control server (CCS) and several control modules (CM). The CCS serves to collect credit information and to make punishment decisions based on the credit information, while the CM generates signaling packets to calculate the responsibility mark. Violators will get punished according to the responsibility mark. The method is distributed and can adapt to the dynamic state of network, and is quite simple to implement. It relies on sources self-restriction to avoid congestion and the expensive traffic shaping needs only to be done at sources in this way.