A method to determine timeslot duration in a TDMA-based MAC protocol for VOIP over wireless

摘要

What is described is an approach to assigning timeslots of appropriate duration when using a TDMA based MAC protocol for forwarding VOIP traffic. It is known that a typical user speaks only 35% of the time in a voice session. Existing VOIP systems exploit this fact and reduce the amount of data generated by employing Voice Activity Detection (VAD) algorithms. The packets generated when a user is silent are shorter than the packets generated when a user is speaking. Hence there is a need to assign timeslots of different duration based on user speech activity. When the user is speaking we assign an Active Speech Timeslot (AST). When the user is listening a Silent Speech Timeslot (SST) that is long enough to transmit packets generated by a Comfort Noise Generator is assigned.