The activity of N speakers accessing a packet voice concentrator has usually been modelled as an N-state Markov birth-death process. The Markovian model is not very accurate when N ≤ 10. In this article we use a three-state single speaker model in setting up an analytical model of a small packet voice system.
Get full access to this article
View all access options for this article.
References
1.
WeinsteinC. J., ‘Fractional speech loss and talker activity model for TASI and packet-switched speech;IEEE Trans. Commun., COM-26, pp. 1253–1257 (Aug., 1978).
2.
BradyP. T., ‘A model for generating on-off speech patterns in two-way conversation;Bell Syst. Tech. J., 48, pp. 2445–2472 (Sept., 1969).
3.
WeinsteinC. J. and HofstetterE. M., ‘The tradeoff between delay and TASI advantage in a packetized speech multiplexer;IEEE Trans. Commun., COM-27, pp. 1716–1720 (Nov., 1979).
4.
GaverD. P. and LehoczkyJ. R., ‘Channels that cooperatively service a data stream and voice messages;IEEE Trans. Commun., COM-30, pp. 1153–1161 (May, 1982).
5.
YatsuzukaY., ‘A highly sensitive speech detector and high-speed voiceband data discriminator in DSI-ADPCM systems;IEEE Trans. Commun., COM-30, pp. 739–750 (April, 1982).
