Analysis of a Soft Real-Time Random Access Protocol

A communications network that is part of a soft real time system will need to transmit messages within a bounded delay, but may allow some messages to miss this bound and be dropped within a maximum prespecified rate of message loss. This paper presents a media access protocol for soft real time systems implemented on a slotted radio channel with binary feedback. The protocol is based on the Gallager FCFS and Capetanakis splitting algorithms, but incorporates strict delay bounds using packet laxities. Also presented is an analytic model for this protocol by examining the probable lengths of the collision resolution intervals given the current lag at any time. Both analytic and simulation results are obtained to study the maximum input traffic rates that can be sustained for various laxities, delay bounds, and message loss rates.