The training and simulation systems of the 1990's will be more complex as total trainers and multiple participant systems mature. The Ada language offers the software engineering fundamentals needed to handle the greater complexity and the life cycle advantages to reduce software costs. To use Ada, training system vendors must decide on a method to transfer to Ada.

There are two basic approaches for transfer to Ada technology. At one end of the spectrum is the "generic" approach, which uses the generic, standardized Ada structures for the implementation. This path promises the benefits of modern software engineering, easier maintenance, and greater portability. This path also requires the cost of quality Ada training, the risk of using a new technology, and possible performance degradation.

At the other end of the spectrum is the "proprietary" approach, which depends on other non-Ada, more traditional support systems for the real-time implementation. This approach promises a more appealing transition since risk of new technology is lowered, but there are tradeoffs such as the predicted higher life cycle costs and the loss of the software engineering advantage offered by Ada.

Neither approach is superior in all cases, but each has its advantages and disadvantages, which are classified and weighed in this paper. Analysis is based on the application speed, efficiency, portability, determinism, software training, and maintainability. A survey of the philosophy of some of the real-time Ada systems currently available on the market is presented. Systems are evaluated based on the cost/benefit areas established in the paper. Developers of Ada real-time training and simulation systems can use these guidelines to plan their approach early in the project to ensure that the requirements will be met in a cost-effective manner.