Scheduling real-time traffic in ATM networks

In this paper we study the problem of scheduling real-time traffic in high-speed ATM networks. Scheduling can be performed at different points in the ATM network, such as, at the multiplexers and switches. In our model the arriving traffic cells are assumed to have weights associated with them which determine their importance measure (for example, cells belonging to one traffic class may have different weights than cells belonging to another traffic class). Further, a cell is said to have been lost if its deadline Is violated. We develop a simple procedure, S-OPT, to minimize the cell loss rate for single-class traffic. Using this procedure we develop an optimal algorithm, M-OPT, that minimizes the weighted loss rate at a network node. Finally, we provide a heuristic algorithm, M-HEUR, which significantly reduces the complexity of M-OPT and closely approximates its performance. In each case the scheduling decisions are based only on the current information in the queue. Using numerical results we show that our scheduling algorithms significantly outperform well known algorithms such as first-in-first-out (FIFO) and static priority (SP).