Adaptive QoS Mechanisms for Control and Data Planes of Software-Defined Tactical Networks

This paper defines two quality of service (QoS) mechanisms for the control and data plane of Software-Defined Tactical Networks (SDTN). The motivation comes from the fact that standardized SDN protocols like OpenFlow are not designed to meet tactical network requirements, demanding complementary mechanisms. Therefore, we introduce an Adaptive Quality of Service (AQoS) to reduce the overhead of OpenFlow on the control plane via a clustering approach and a queue mechanism on the data plane to mitigate packet loss and ensure prioritization. Both mechanisms are evaluated in a network emulator hosting emulated tactical radios with REST APIs to expose measurements and parameters to the controller. The clustering mechanism uses these APIs to compute control overhead and packet loss to decide how to split the network into clusters. The queues at the switch adapt the dequeuing rate according to the radio buffer usage and link quality. Multiple experiments have been conducted to test the AQoS mechanisms on both planes using VHF, UHF, and SatCom links. The experimental results show a significant reduction of control overhead of up to ~93% by using multiple clusters and suggest that the adaptive queuing differentiates data flows in heterogeneous networks while significantly reducing packet loss from ~68% to nearly 0%.