Parametric Optimization of Packet Transmission with Resending Packets Mechanism

Abstract

The data transmission process is modelled by a Markov closed queuing network, which consists of two stations. The primary station describes the process of sending packets over a lossy channel by means of a finite and single-channel queue. The auxiliary station, being a multichannel queuing system, accumulates packets lost by the primary station and forwards them back for retrial. The transmission rate at the primary station and the retrial rate at the auxiliary station are in the specified ranges and are subject to optimization in order to minimize the time of successful delivery and the amount of network resources used. The explicit expressions for these characteristics are derived in the steady-state mode in order to formulate the problem of bi-criterion optimization. The optimal policies are established in two scenarios: the first problem is to minimize the average time of successful transmission with limited resources; the second problem is to minimize the consumption of network resources under the constraint on the time for successful transmission. The set of Pareto-optimal policies is obtained by solving the problem of minimization of the augmented functional. The quality characteristics of approximate solutions that do not take into account the service rate in the auxiliary system are analyzed.