Q-Flag: QoS-Aware Flow-Rule Aggregation in Software-Defined IoT Networks

Software-defined IoT (SDIoT) is a promising approach to address the requirements of IoT, such as network management, quality of service (QoS), and resource utilization. The advantages of SDIoT are facilitated by the separation of the data-and the control-planes using flow-rules, that allow fine-grained control over individual flows. However, the number of flow-rules that can be placed at the switches is limited, leading to scalability issues in SDIoT. Existing approaches to flow-rule management either do not consider the impact on quality-of-service (QoS) or are applicable only to a particular topology. In this paper, we propose a QoS-aware flow-rule aggregation scheme for generic network topologies, which aims to achieve satisfactory trade-off among flow-rule compression and its impact on the QoS of IoT traffic flows. Specifically, the proposed scheme adaptively aggregates flow-rules while considering different QoS requirements of IoT applications in the network, and the flow-rule capacity of the switches. The proposed scheme consists of the following components a) a path selection heuristic to increase the total number of flow-rules that can be accommodated in the network, and b) a multi-arm bandit based flow-rule aggregation scheme capable of reducing the number of flow-rules, while maintaining adequate performance in terms of QoS. Experimental results using IoT traffic show that, on average, the proposed scheme is capable of reducing the average end-to-end delay and QoS-violated flows in the network by 22% and 30%, respectively, compared to the state-of-the-art schemes.