Research On QoS-aware Routing Strategy In Software-Defined Networking

Over the past decades,with the rapid advancements of the Internet,many new types of networking applications and services have emerged for end users.These applications generate their own flows characterized by different quality-of-service(QoS)requirements in terms of link delay,bandwidth and packet loss ratio.However,the current best-effort delivery model in the Internet is not capable of satisfying these heterogeneous requirements.Software-Defined Networking(SDN)has emerged as a promising network architecture.Its core idea is to separate the routing decision function and the data forwarding function in traditional network forwarding devices and form a control plane and a data plane.The SDN architecture,as well as the prevailing protocol between the control plane and the data plane,i.e.,OpenFlow,provides fine-grained technical support for QoS provisioning.Based on the observation,this article mainly studies the QoS-aware routing strategies for different types of applications in SDN networking,and the following two schemes are proposed:(1)QoS-aware multicast routing scheme for the scalable video streaming service.Scalable video streaming is a common multimedia streaming service,and this paper designs a comprehensive QoS-aware multicast scheme for this service.Firstly,we present the designed SDN-based multimedia streaming multicast system architecture.Then,we propose a timeefficient tree construction algorithm for route scheduling.Furthermore,we design a layer switching strategy based on the video distortion model to mitigate network-induced video quality distortion.Finally,we implement a prototype system and conduct various experiments.The experimental results show that the proposed scheme can effectively guarantee users' quality of experience and improve multicast scalability,while meeting the QoS requirements of the scalable video streaming service.(2)QoS-aware unicast routing scheme for the VoIP(Voice over IP)service.VoIP is a popular end-to-end voice call service over IP networks,and this paper proposes a unicast routing scheme for this service.Firstly,we use the SDN logical centralized control plane to collect historical traffic data in the data plane to construct the dataset.Then,we train a designed deep learning model,namely DeepRouting,which employs a recurrent neural network to learn traffic engineering characteristics from the dataset.Afterwards,we use the trained DeepRouting model to perform online route scheduling for VoIP service requests.The experimental results show that the DeepRouting model is able to perform online route scheduling in a time-efficient way and achieve reasonable QoS performance,thereby verifying the effectiveness and feasibility of the proposed scheme.