Adaptive bandwidth-constrained media-specific forward error correction for voice over Internet Protocol

In response to time-varying network loads, the proposed adaptive bandwidth-constrained media-specific forward error correction (FEC) mechanism for voice over IP (VoIP) attempts to maximise the perceived voice quality at the receiver by switching between coding algorithms while maintaining a constant source bit rate. Its inherent fixed bandwidth nature alloys the worries of network overloading caused by additional redundancies. Besides that, during low network traffic, the mechanism adopts a better speech coding scheme with a lower level of redundancy to achieve better voice quality and to reduce the overall latency. Typical voice traffic accumulated through Internet measurements is characterised by using the 2-state Markovian (Gilbert) loss model. To evaluate the overall performance of the adaptive mechanism, the cumulative distribution function (CDF) of perceived voice quality is adopted. Preliminary results clearly show that the proposed adaptive mechanism produces a consistent and enhanced voice reception subject to the primary-encoder limitation while maintaining a constant bandwidth.