Worst traffic passing virtual frame regulation: Analysis with dynamic programming

This paper studies a discrete-time queueing system in which the traffic regulation places an upper bound on the number of arrivals in each of the non-overlapping time windows (virtual frames) of a fixed length. The worst traffic passing through this traffic regulation is characterized. As a criterion for traffic comparison, this paper employs the weighted sum of queueing delays with the weighting factors decaying geometrically over the time horizon. This paper uses dynamic programming techniques and convexity argument to characterize the worst traffic.