Traffic-Aware Efficient Routing Protocol For Software Defined Networks

Software Defined Networks(SDN) is a promising networking architecture. Information delivery latency is an important measurement of routing protocol, since it is very necessary for the network service efficiency and user experience. However, the existing routing protocols face the following challenges: 1. optimal goal singleness, most of the existing routing protocols only consider the forwarding path length, unfortunately, the shortest forwarding path cannot guarantee the delivery latency minimum. 2. The links in the network often shared by multi-flows, so, the new data flows may increase the delivery latency of the existing data flows. 3. Recently, the bandwidth of link will be assigned fairly to the data flows, which delivered by the link. But this kind of bandwidth assignment patterns cannot meet the individual requirements, which means the deadline of data flows in this paper, meanwhile, the delivery latency will be stretched.To address the above challenges, based on the Software Defined Networks, this paper discusses how to reduce delivery latency of data flows in the networks employing routing scheme and scheduling scheme. Finally, this paper designs a traffic-aware efficient routing protocol. The detail researches is as following:1. Designing an efficient routing scheme to reduce the single data flow delivery latency. This paper re-modes the routing problem, and formulizes the delivery latency of single data flow. And then this paper proposes a polynomial time algorithm to reduce the single flows latency by optimizing multiple traffic factors. Experiments show that this scheme outperforms current scheme with up to 90% improvement on efficiency.2. Proposing a latency-aware routing scheme to reduce the network-wide data flows delivery latency. For avoiding the influence to the existing flows, this paper considers both the network traffic factors and the data flows individual requirements, and designs a polynomial time algorithm to reduce both the latency of new data flows and the influence of other existing data flows, meanwhile, guarantee the individual requirements of as more as possible flows. Experiments prove that this scheme outperforms current scheme with up to 75% improvement on efficiency, and improves about 20% ratio of flows which meet their deadlines.3. Designing a low latency bandwidth assignment scheme. To overcome the challenge of the fairly sharing of bandwidth resource, based on both the network traffic factors and the data flows individual requirement, this paper modes the problem, and formulizes the data flows delivery latency, and then proposes a polynomial time algorithm to re-allocate the bandwidth resource for the data flows to reduce the network-wide data flows delivery latency. The analysis points that this scheme outperforms the fairly sharing scheme with up to 10% improvement on efficiency.