PDSMA: Pseudo-Deterministic Statistical Multiple Access for Voice Packets Uplink Scheduling

Statistical multiple access gives opportunity to decrease bandwidth wastes in radio cellular networks when sources activities are variable. Usually, 2 states exponential on-off models are used for voice sources and to study statistical multiplexing in telephony. In this paper, we propose a distributed algorithm based on a 3-states model with downlink activity detection for a refined prediction. The proposed algorithm makes the distinction between silent sources according to their downlink activities to enhance the statistical multiple access for telephony. Then, silent sources are pre-allocated available channels on the uplink according to their respective switching probabilities; hence the designation "pseudo-deterministic" for our proposed statistical multiple access scheme. Performance evaluation (analytical and simulation model) is given to illustrate the benefits of our proposal.