Channel allocation in wireless integrated services networks for low-bit-rate applications

Abstract

This work addresses issues related to the design and performance of a wireless integrated services network with emphasis on a tactical framework. We propose an asynchronous transfer mode (ATM)-like protocol architecture for the mobile network, which is an extension of schemes proposed in the literature. A medium-access-control (MAC) scheme, based on slot reservation by the remotes, is proposed for the network. Traffic models for low-bit-rate applications, suitable for low-capacity channels, such as a multiple-access (macro cell) wireless network, are presented. New bi-directional speech-conversation and bursty data models are proposed. The issue of scheduling in wireline integrated services networks is thoroughly addressed and new algorithms are proposed. An analytical scheme to obtain the required (static) capacity for homogeneous sources based on their Markov-chain characterization is provided. A necessary condition for optimality of a scheduling algorithm is the balance of cell-loss-probability (CLP) ratios to values approaching 1 from below, on the boundary of the admissible region. The balanced-CLP-ratio (BCLPR) algorithm satisfies this condition but ignores the deadlines of the cells. The shortest time to extinction (STE) with BCLPR (STEBR) algorithm, proposed here for the first time, utilizes the earliest-deadline- first concept while satisfying the necessary condition. A proof is provided to show that the STEBR decisions are optimal at each service slot given that no information about future traffic arrivals is available. Simulation results indicate that STEBR admits more sources and yields larger normalized channel throughput (by up to 4%) than STE.