Research On Traffic Control Technology Of Cognitive Network Based On Service-Aware

With the help of cognitive function, cognitive network can sense network status, make plan and decision based on the sensory information, and then take proper action accordingly, which make it manage and control the increasing complex network better, thus improve network QoS.A traffic control scheme of cognitive network based on service-aware is proposed in this paper, which involves mainly two aspects, one is queue management, the other is packet scheduling. The rationalization of the proposed method is validated through simulation.In queue management, this paper analyzes the AQM requirement of cognitive network and the principles of PI controller, on this basis an improved PI algorithm is proposed, which is based on input rate changes and queue length changes. Besides, neural network is used for the parametric learning in improved PI algorithm. The rationalization of the improved PI algorithm is validated through simulation.In packet scheduling, this paper analyzes the packet scheduling requirement of cognitive network, then designs a two-level packet scheduling strategy. On the basis of DWRR, the ADWRR_PQ is proposed for multi-queue scheduling. Within the same queue scheduling, the concept of scheduling factor is introduced in order to ensure the overall network average delay.