A Markov Chain Model for Performance Analysis of VBR Voice using Multiple Thresholds

Abstract:
Statistical multiplexing is a popular technique used in digital communications in order to increase utilization efficiency of system resources. Variable bit rate (VBR) voice coding is an example of such multiplexing technique and has been widely used in digital telephony. While fixed rate codec allows only a limited number of simultaneous connections, VBR encoding allows the codec to alter its bit rate dynamically, depending on the instantaneous traffic demand and resource utilization status. In this paper, we propose a scheme to increase the number of simultaneous VBR connections for a system with limited number of channels. When there are more calls, the bit rates of the ongoing connections are decreased and new connections are accommodated by the saved bandwidth. Multiple threshold levels are adopted to increase the number of high data rate connections. The system performance of the proposed scheme is analyzed through Markov chain modelling and the developed analytical model is also validated by simulations. Numerical results demonstrate that the concept of multiple thresholds leads to better performance of the system, represented by a higher percentage of high rate calls.