QoS-based traffic management in software-defined networking

Abstract

The emergence of new applications, such as online gaming, video conferencing, and virtual reality necessitates the underlying network to be capable of fulfilling high bandwidth and low latency requirements. Growth in data traffic from these applications over the Internet increases the congestion in the network architecture. The “best-effort” service delivery model used in the traditional Internet architecture fails to meet the bandwidth requirements of these applications. Software-defined Networking (SDN) solves the congestion problem and allows the network to be dynamic, intelligent and control the network devices centrally. SDN has many advantages compared to traditional networks, such as the separation of forwarding and control plane from devices, global centralized control, and network traffic management. The SDN is a viable approach to fulfill such Quality of Service (QoS) requirements by improving the data delivery performance of emerging applications. In the first work, we propose a policy-based framework to enhance the QoS of traffic flows in the SDN networks. We phrase a max-flow-min-cost routing problem and present a heuristic method to route the traffic flows in the network in polynomial time. In this work, the framework monitors the QoS parameters of traffic flows and identifies policy violations due to link congestion in the network. This approach dynamically implements policy rules to SDN switches upon detecting policy violations and reroutes the traffic flows. The simulation results show that the proposed approach has achieved better results than SDN without policy-based framework and reduced end-to-end delay, average jitter, and QoS violated flows by 51%, 62%, and 28%, respectively. Sometimes, the single path routing fails to meet the bandwidth requirements of traffic flows. Some existing works focus on multipath routing but have few limitations. Firstly, the packet scheduler in multipath routing perform packet distribution without head-of-line blocking delays. Secondly, the data packets may reach out-of-order at the destination due to the uneven packet distribution over multiple paths. To resolve these issues, we propose a QoS-aware dynamic routing scheme using multipath routing that comprises three phases – flow splitting, multipath routing, and flow reordering. In the first phase, we propose a flow splitting scheme to decide how to split the incoming flows to enable multipath routing in the network. In the second phase, we design a cost function for routing splittable flows and formulate a min-cost routing problem as an integer linear program (ILP). We propose a greedy heuristic-based approach to solve the problem in polynomial time. Finally, in the third phase, we propose a flow reordering scheme for received subflows through multiple paths to maintain the desired flow sequence at the destination. We evaluate the performance of our proposed approach against the existing Multi Constraint Optimal Path (MCOP), Multi-path SDN (MPSDN), and priority-based dynamic multi-path routing (PDMR) schemes. The results show that our proposed approach outperforms the benchmark schemes regarding QoS metrics such as network throughput, end-to-end delay, and QoS violated flows. 5G network service includes enhanced mobile broadband (eMBB) applications such as video streaming, video conferencing, and virtual reality which support a high data rate. These applications generate a large amount of data daily and pressure the network operators to offer desired bandwidth during peak times in the network. Existing related work used the Flat-rate pricing (FRP) model because of its simplicity and easy deployment in which the ISPs charge a fixed price from the users for the data usage for a particular period (e.g., month, year). Similarly, the Usage-based pricing (UBP) model was used in which the users are charged additional fees if they exceed their limits. However, it does not work well in controlling peak time congestion as it fails to change the prices according to the varying traffic in real-time. Further, Time-dependent pricing (TDP) has gained attention for handling peak time congestion and considers the time variance for users’ demands, and charges are applied to users dynamically with time. The Internet Service Provider (ISP) plays a vital role in flattening out the fluctuations in traffic demands during peak-time. Therefore, in this work, we study the varying traffic demands and develop a pricing method to allocate bandwidth to the users during peak hours. We model the interaction between ISP and users in the form of a Stackelberg game and solve the Nash equilibrium to find the optimal congestion price during peak time. The simulation results show that our proposed approach significantly controls congestion on the link by performing bandwidth allocation to the users within the limits of the available bandwidth. Further, we propose a congestion technique using Hierarchical Token Bucket (HTB) to manage the bandwidth in the Software-defined IoT (SDIoT) network. We propose a routing scheme to compute optimal routing shortest paths using Dijkstra’s algorithm by selecting the min-cost path based on the priorities of traffic flows. The results illustrate that the proposed approach reduces end-to-end delay by 38%, 44%, and higher average throughput by 29% and 43% in comparison with the benchmark schemes – SDN with HTB and the Delay Minimization method, respectively.