DIS Workshop in Transition to ... What?
by Roy Latham

The 15th Workshop on the Interoperation of Distributed Interactive Simulation (DIS) was held in Orlando, FL, in mid-September. This is the last of the Workshops to be held under the DIS banner. A new workshop series will begin in March, devoted to what is sometimes called DIS++, which is quite a different animal.

DIS was mainly concerned with building battlefield simulations in which participants interacted by exchanging data over a network. Much of the discussion focused on the mechanisms for solving common problems and on defining standard data formats for implementing the solutions. Progress was slow, as one might expect with a standards effort with a thousand or more participants per workshop, in a constantly changing roster. Many of the participants knew few of the basics of simulation; it took about two years for the group to collectively figure out that the extrapolation of vehicle positions was really required to account for time differences in broadcast position data. Despite the slow pace, standards were hammered out and they are reasonably good ones. Perhaps as important, many people learned the basics of interactive simulation.

The DIS workshops, after so long a time, got through the basic problems and were working on the more difficult ones. The tough problems include transmitting weather and environmental information among the participants, and keeping everyone's database concurrent as objects are built, destroyed, or otherwise modified. Although many tough problems were short of resolution, the Department of Defense (DoD) decided to move on.

Someone at DoD correctly noticed that a fortune was being spent on simulations at every level of system development. Detailed simulations are built to verify the innards of new system designs, high-level simulations are built to test the strategies of warfare, and there are strategies at every level in between. Yet rarely can the simulations, as written, communicate with each other. There are benefits of having such communications. For example, the engineering model of a new tank suspension might be tried in the synthetic environment of a battlefield simulation. The perceived problem is how to facilitate communication among the many simulations to obtain the benefits of the interaction between simulations.

DIS focused on solving a specific set of problems and codifying the solutions in specific protocols and data formats. That approach will not work if one is trying to interface anything to anything. There are too many interface problems and specific formats to attempt such an effort. So instead of a straight-away solution, one must seek a metasolution-a solution that facilitates solutions rather than being a solution in itself. The name for such a thing is the High Level Architecture (HLA), or sometimes DIS++. If this sounds a bit ethereal to you, you have company.

At the last workshop, I met people in two camps: those who were perplexed by HLA and those who claimed it would not work. To be sure, there is a third camp of those who are enthusiastic about the prospects; they are just such a small group that it is unlikely one will encounter any of them.

The skeptics may be offended that they were not involved in the HLA development process. Unlike DIS, HLA was not invented in a public forum. DoD let a few development contracts to design the basic approach. Undoubtedly, development proceeds faster when done that way, but it also threatens to leave out the requirements of the broader community. In addition it builds the skepticism of many of those who were not involved. The slow way of wringing things out in a public forum certainly has its disadvantages; the end product may serve many people adequately and few people well.

While it is too soon to say if HLA will win over its skeptics, at this point there are certainly few people who believe it will work well in the case of a real time interactive environment, the environment that DIS was designed to treat. Moreover, the problems of DIS are far from being solved, even though the "politically correct" story is that all that is left is to put a bow around the DIS package before moving on to the grand challenges of DIS++. What will happen is that DIS-related problems will continue to be attacked, but without the benefit of an official forum. There are many mechanisms for continuing the work on DIS, including the Advanced Distributed Simulation Technology (ADST) II program and the DIS Systems Engineering Integrator (DIS SEI) program, as well as individual development contracts such as the Army's Close Combat Tactical Trainer.

An interesting footnote to the DIS and related standards effort is the confirmation, one more time, that large projects are effectively exempt from paying any more attention to standards than the programs choose. Standards are only enforceable on "small guys." CCTT poured much effort into developing SEDRIS, an alternative to the DIS-recommended and officially adopted Simulator Interchange Format. To many outsiders, it seems that SIF could have been easily modified to meet CCTT requirements without undertaking a major new effort. In any case, little attempt was made to involve the larger simulation community in the development of SEDRIS.

Another major program, the Defense Advanced Research Projects Agency's (DARPA) Synthetic Theater of War (STOW) worked to develop new databases in the outdated database format used for the original SIMNET.

If a major point of standards is to save money by facilitating reuse, exempting large projects has the effect of undermining much of the good that could be done. However, all big projects are important projects, and important projects are too important to be held up by slow-moving standards processes.

What will happen to the DIS++ Workshops? No matter how successful the HLA architecture, it is a metasolution, and there are not nearly as many people as interested in metasolutions as in real solutions. Nor are there as many people as capable of contributing. The internet DIS-TRANSITION reflector (i.e., e-mail discussion group) has recently been dominated by the discussion of two topics: (1) what to name the new set of workshops, and (2) how more academic papers might be introduced and refereed. The focus is not on simulation, rather on problems about how data is represented and communicated. The problems are valid and important ones, and the simulation community will no doubt benefit, but it is a different focus from DIS. Part of the transition process will be to a new set of players, probably a small, more academically oriented group than the people who were interested in DIS.

The people interested in the simulation problem, rather than the metaproblem, have to make a transition as well. For example, in the DIS workshops there was a special interest group dealing with the problems of presenting human figures in interactive simulation. This is as interesting and important a problem as any in simulation, but, alas, it deals with a problem not a metaproblem. The people who work on the human figure problem are widely scattered and are in need of a forum. Should they try to hold out as an isolated outpost of problem-solving in the DIS++ workshops, despite the political "incorrectness" of seeking solutions where only metasolutions should be sought? Possibly solutions could be sought covertly under a banner of, say, the "Special Interest Group for Applying HLA to Human Interactions." Or perhaps a new forum should be sought as part of the main simulation conference, the Interservice/Industry Simulation and Education Conference (I/ITSEC).