A discrete-time stochastic approach to flow control dynamics [in computer networks]

Flow-control mechanism that dynamically regulate the rate of data flow into a network based on feedback information about the network state are considered. A stochastic discrete-time approach that yields models which are realistic and yet tractable and computationally easy to solve is presented. The approach is illustrated by means of a rate-based flow control mechanism where the feedback consists of an exponentially averaged estimate of the bottleneck service rate and queue size. The traditional static exponential averaging technique and a dynamic technique are considered. The static technique guarantees that the averages of the source rate and bottleneck queue size converge to their ideal values. However, the variances behave in such a way that one has to choose between good steady-state behavior and rapid adaptability. The dynamic scheme overcomes this disadvantage