Article ID | Journal | Published Year | Pages | File Type |
---|---|---|---|---|
427500 | Information Processing Letters | 2013 | 4 Pages |
•We introduce and define the burst factor κ as a key parameter to measure source burstiness.•We explain the properties of κ with respect to its intuitive interpretation and calculation simplicity, its scalability, superposition of sources.•We show how κ can be obtained for a set of MPEG-4 video traces.•We demonstrate the importance of κ for evaluating the performance of a buffer with MPEG-4 video sources.•We demonstrate the superposition property when MPEG-4 video traces are used as input.
In this contribution we introduce the burst factor, which is defined in a general setting, and quantifies the long-term variability, or burstiness, of a packet source. By means of a combination of analytic and simulation results, we will demonstrate that this parameter plays an important role in the calculation of the moments of the amount of work in a buffer, both in a scenario where (a superposition of) MPEG-4 video traces are used as input of the buffer, as well as when the (correlated) arrivals are generated by a stochastic model. As such, the burst factor is a key parameter for buffer dimensioning purposes.