An Adaptive Loss Differentiation Algorithm Based on Queue Management

It is important for TCP to provide a mechanism to differentiate between congestion losses and wireless random losses in wired/wireless hybrid networks, because the latter may cause a severe throughput degradation if TCP sender misinterprets such losses and reduces its window size. Being different from other approaches based on end-to-end metrics, BQM is an efficient loss differentiation algorithm based on queue management. This paper points that BQM discriminator can not be fit to different network states and proposes an adaptive BQM, whose key idea is to adjust the primary parameters according to network events and track the dropping pattern on routers. The simulation results show that the accuracy of A BQM discriminator is higher than that of BQM in different network states and A BQM can adjust itself adaptively according to network states.