Cell-oriented adaptive admission control for gos support in wireless multimedia networks - Electronics Letters

point in this Letter is that numerical results have shown that, for small numbers of multiplexed sources, our earlier analysis [l] can be highly inaccurate. In this Letter we present a scheme for source aggregation (and therefore the solution of the overflow probability) which is applicable when multiplexing a small number of sources. As in the previous paper, we refer to the buffer fed by the aggregate input process as the aggregate queueing system. Unlike previous work oriented towards a large number of sources [l], we here achieve parameterisation of the aggregate queueing system by adjusting only the output rate, defining C‘ = the equivalent output rate of the aggregate queueing system in packetsls and also: X = buffer length in packets A = mean applied load in packetsls p = utilisation q = decay rate of the burst scale state probabilities. Note that the purpose of using the equivalent output rate C‘ is to parameterise the model in the most minimal way possible, while maintaining accuracy. In the ON state of the aggregate queueing system the total input rate exceeds the service rate of the buffer and the buffer fills (rate of increase of queue level = R C’). In the OFF state of the aggregate queueing system the total input rate is less than the output rate of the buffer, so allowing the buffer to empty (queue level reduction rate = C‘ R). For reasons of simplicity we propose and test the following formula for C‘: