An Aggregation Technique for Network Traffic Described by MMBP Models

In this paper, an efficient technique is presented to model the aggregation of traffic described by Markov Modulated Bernoulli Processes (MMBP), which has been used to describe the burstiness of network traffic. For the aggregation of multiple MMBP traffic, a recursive algorithm is derived to find the parameters of the aggregated traffic models based on the buffer sizes and the number of inputs. By aggregating multiple traffic inputs into one, the complexity of network performance modeling is significantly reduced. Compared to methods by simulations, the proposed technique results in satisfied accuracy with much less computations, which are demonstrated by simulation examples.