Analysis of Asynchronous Time Multiplexing of Speech Sources

Asynchronous time multiplexing is proposed as a more efficient alternative to synchronous time multiplexed PCM. The analysis here evaluates the asynchronous method when extremal coding is used with the speech sources. Assuming a Poisson relation for the occurrence of multichannel samples, the buffer action is analyzed using an iterative probability calculation and computer experiments. Using chosen parameters, the over-all digit rate is then evaluated. For these parameters a 100-source asynchronous system requires only one-third the digit rate of a synchronous multiplex. About a 5-db loss in signal-to-noise may be expected, due to the extremal coding. Two additional features are: 1) 20 per cent of digit rate is available for nonreal time data; and 2) the usual elastic overload of asynchronous systems is present. These theoretical results indicate that the asynchronous method should supplant synchronous multichannel cases if bandwidth is sufficiently important to justify additional equipment.