Discrete-Event-Dynamic-System-Based Approaches for Control in IntegratedVoicelData Multihop Radio Networks

摘要

We report accomplishments and new directions in our effort to develop and applydiscrete-event-dynamic-system-based techniques for the transmission scheduling problem in Radio Networks (RN). First, we examine this problem in the context of data traffic in general topology networks. Next, we look at the scheduling problem when processing packetized voice calls, where Grade-of-Service (GOS) requirements are quite different. For data traffic, we formulate an optimization problem for the allocation of transmission time slots to different competing nodes and present a gradient-based algorithm suitable for on-line implementation without any assumptions on the nature of the data traffic processes. Examples illustrating the adaptive features of our approach and comparing it to other schemes are included. For voice traffic, we formulate the optimal scheduling problem as a stochastic discrete resource allocation problem, which is combinatorially hard. We describe a technique for transforming this to a continuous optimization problem and develop algorithms for minimizing the blocking probability of packetized voice traffic.