Special Issue on the Telecommunications Description Language David M. Nicol Department of Computer Science Dartmouth College Hanover, NH 03755-3510 nicol@cs, dartmouth, edu The field of parallel discrete-event simulation (PDES) was launched almost exactly twenty years ago and since then hundreds of papers on PDES have been published. Still, PDES technology has not made much of an impact in critical commercial application areas. I have my own opinions on why this might be so. One is that PDES has not been demonstrated on industrial strength simulation problems, and is likely viewed as a risk by those who have them. I also believe that PDES is a niche activity--only those with simulation problems too large and complex to do at all on a serial machine are willing to deal with the attendant hassles. Telecommunications simulation seems to be in that niche. Two years ago, the Telecommunications Description Language (TED) and applications of it grew out of that recognition, thanks to funding by NSF (CCR-9625894), and later DARPA (N66001-96-C-8530) for a consortium of people in industry and academics. Simply stated, our goal is to make possible parallelized simulation of large telecommunication networks. As a member of the consortium who happens now to have a soap-box to stand on, I thought to assemble an issue that describes TED and its uses. There are, I think, two important things to bear in mind when looking at these papers. First, TED was designed to avoid known performance pitfalls in parallel simulations (some of us think overly so, but the more important issue is that TED succeeds in dodging some serious problem areas). Second, that it exploits parallelism and delivers performance on a variety of simulation problems with some real complexity to them. In short, I think we're on to something here, and hope to convince you all that high performance parallel simulation of telecomnmnications systems is viable and that systems like TED can deliver the goods. We start the issue with a tutorial on the TED meta-language by those who have designed and influenced it most. This is followed by a report on how researchers at Bellcore used TED to develop models of the PNNI protocol suite. This project in particular demonstrates how application needs helped the tool evolved. A version of this paper appears in the proceedings of the t997 Winter Simulation Conference (which holds the rights but allows selective republication). We next look at how University of Massachusetts experts in multicast protocols used TeD to investigate reliable multicast. An important lesson demonstrated by this group is that despite our best efforts, some models are tough to parallelize. The fourth paper is by wireless experts at Rutger's WINLAB, on their use of TED to develop a wireless network simulation test-bed. A version of this paper will appear in the Proceedings of the 1998 Workshop on Parallel and Distributed Simulation (which holds the rights). The last paper describes our work at Dartmouth on developing a library of TCP models in TeD, by way of transforming existing ns models. Our experiences point out additional ways in which TeD might be usefully extended. Finally, I extend my thanks to my partners in this project who took the time to craft these papers, to our sponsors at NSF and DARPA who like us believe in the potential of PDES, and to you the readers.

/lp/association-for-computing-machinery/special-issue-on-the-telecommunications-description-language-Nfo7FmQwq0