Abstract : More and more traffic is migrating to private networks in use in various professional environments. With this, there comes a growing diversity in applications, each of them requiring different quality of service. This poses challenges to properly managing such networks, which can span both wired and wireless segments. To tackle the issue of network management and application demands in such networks, we introduce a network- and application-aware adaptive congestion control algorithm, which provides congestion-free service differentiation to the flows in wireless networks in a decentralized manner. The designed algorithm operates based on the real-time network information obtained from in-band network telemetry and aggregated flow information from intermediate nodes. The algorithm performs three times better than the existing CUBIC congestion control algorithm and twice better in a multi-flow architecture. The designed algorithm is the first step towards adaptive transport and application layer protocols which are the future of private professional networks.
Index terms : Connectivity , Wireless networks , Congestion control algorithm , eBPF , INT , APP-NET