Feedback-Based Control Loop Congestion Control Algorithm for Wireless Networks

Abstract

Wi-Fi plays a crucial role in connecting private professional networks by providing varying data rates based on channel quality. Despite advancements in Wi-Fi protocols, there is still a problem of throughput degradation for higher data rate devices when lower data rate devices are around, due to airtime unfairness. The diverse requirements of emerging applications emphasize the necessity for network protocols that cater specifically to these needs. To tackle the problem of throughput degradation in devices that use higher physical data rates, we have designed a control-loop congestion control algorithm that decides the application data transfer rate based on the real-time network context and aggregated airtime feedback from the access point. The design is based on in-band network telemetry and stack programmability of eBPF. Using the designed algorithm, the throughput of the device that uses a higher physical data rate is improved by 58% without compromising the overall network efficiency. The average airtime difference between the end devices of the network is as low as 18%. Other than airtime fairness, the features implemented in the algorithm and monitoring and feedback system in the access point can also be utilized for priority-based airtime fairness and several such potential use cases in wireless networks in the future.