Congestion Control with Deterministic Service Delay Guarantee

To meet the ever-increasing demand for mission-critical applications, the deterministic service delay guarantee from the application prospective has become an important metric for congestion control in end-to-end communications. In this paper, we propose a two-timescale congestion window control (TCWC) mechanism with delay-aware priority based on TCP Vegas. Different from the existing works only considering the network delay, we formulate the network utility maximization problem for congestion control by adding additional flow queuing delay at the source node to guarantee the deterministic service delay constraints. By designing a virtual queue, we transform the delay constraint to the time-averaged queue stability, and solve it in each time slot according to the Lyapunov drift-plus-penalty method. Then we obtain the adjustment strategy of congestion window according to Lagrangian duality theory. For further guarantee the service delay, we apply extreme value theory (EVT) to evaluate the priorities of different flows, which determines the update rate of congestion window. Finally, simulation results show that our algorithm can significantly reduce the average service delay and achieve better delay guarantee compared to traditional TCP Vegas.