Research On The Rate-Based Anycast Congestion Control Strategy

With the fast development of network technology, Internet has displayed a lot of new trend, and people has also brought forward the new request to Internet. The extended with the anycast technology on Internet, and the condition that consumer quantity exponent level increases, we must resolve bottleneck chain in communication process as moving downwards.Based on analyzing the problems anycast communication face , this thesis puts forward the main valuation target in anycast congestion control, while are extensibility and TCP-friendly. We analyze and discuss the fair weight distribution an unfair weight distribution in anycast communication congesting control, and propose three influence factor of weight distribution, which is the local switches history record, bandwidth, route distance. Having compared by AWD, DBWD, DHWD, an improved unfair weight distribution method that refers as Distance-History information-Band Weight Distribution is suggested. The simulation experiments show that DHBWD method is possessed by flexibility, validity and applicability.For occupying bandwidth on TCP flows and anycast flows in the anycast communication fairly, we study the problem of congestion detecting and congestion control, and propose a rate-based anycast congestion control strategy(RBACC) which satisfies extensibility in congestion detecting of RBACC and TCP-friendly, the congestion signal is divided four states, which is constant of the first stages state, increase of the first stages state, decrease of the first stages state and the second stages state. By the congestion signal and TCP throughput model, the congestion control span has adjusted a rate linearity field in according to the ideal of AIAD and AIMD.In order to evaluate the performance of our proposed RBACC strategy, we perform same simulation experiments aiming at the TCP flows and concurrent and eruptive anycast flows. The experimental results show that RBACC strategy is extensible and TCP-friendly. It can solve the problem of bottleneck chain balance the load of network, and improve network transmission function.