The Study Of Queue Scheduling Algorithm On IP DiffServ

Nowadays IP network developed rapidly the initial purpose is to provide connectivity for transferring data between computers, so the application is relatively simple, just some simple data services. With the commercialization of the internet, the application is gradually diversified. At this point, the traditional IP network has shown the weaknesses of quality of service. To this end, the concept of IP QoS has proposed, in order to make sure IP network can provide QoS guarantees for different application.The nature of IP QoS Control is the management of resources, namely the control of the buffer queues, link bandwidth, distribution and use of network resources, in the network transmission ,the queue scheduling plays a significant role and is one of the core technology of IP QoS Control, is also an important means of network congestion. As to IP QoS, the IETF put forward The Integrated Service Model early, this model is not suitable for the core network, generally used in the edge of the network. Afterwards in order to overcome the limitations of The Integrated Service Model, the IETF in 1998, put forward The Differentiated Service Model. This paper comprehensively study the basic scheduling algorithm of The Differentiated Service Model and analysis their advantages and disadvantages, according to the characteristics of The Differentiated Service Model, we put forward a improved algorithm named DWRR_PQ.Its basic idea is that when the packet arrival DiffServ domain, edge router will firstly make a classification & scheduling on the packet,. If the packet is marked as EF business, it must go through the flow control of the token bucket filter, if the token bucket has the token with the flow requirements, it will pass the PQ scheduling and immediately forwarded out, otherwise the group will be relegated to treatment by re-labeling, and while being transported to other queue to go through the DWRR algorithm to scheduling; if the coming packet is marked as AF or BE business, then it must firstly through the DWRR algorithm for scheduling, the DWRR algorithm processing will be sent to the PQ algorithm in low priority or middle priority queue waiting for priority algorithm. DWRR_PQ is rooted in the DWRR algorithm and combines the advantages of DWRR and PQ algorithm, it provides a low-latency and low-jitter service for a number of demanding real-time application, but also to provide fair bandwidth services for time-sensitive application, to avoid the possibility of long-term un-served. At last, by using the OPNET simulation software we simulate the performance in throughput latency, and jitter of DWRR_PQ algorithm, confirmed the superiority of the DWRR_PQ algorithm.