Worst-Case Delay Control in Multi-Group Overlay Networks

This paper proposes a novel and simple adaptive control algorithm for the effective delay control and resource utilization of EMcast when the traffic load becomes heavy in the multi-group network with real-time flows constrained by the (sigma,rho) regulators. The control algorithm is implemented at the overlay networks, and provides more regulations through a novel (sigma,rho,lambda) regulator at each group end host who suffers from heavy input traffic. To our knowledge, it is the first work to incorporate the traffic regulator into the end host multicast to control the heavy traffic output. Our further contributions include a set of theoretical analysis and results. We prove the existence and calculate the value of rate threshold rho* such that for a given set of K groups, when the average rate of traffic entering the group end hosts rholowbar > rho*, the ratio of worst-case multicast delay bound of (sigma,rho,lambda) regulator over traditional (sigma,rho) regulator is O(1/Kⁿ) for any integer n. We also use the computer simulations to evaluate our novel algorithm and regulator that have been proved efficient in decreasing the worst-case delays