Preserving Avionics Software Investment Through Interface Abstraction

Modernization of fielded avionics software has proven to be a difficult task. As the software evolves through its life-cycle, it invariably becomes larger and more complex and represents an increasing investment in the total avionics system cost. To reduce cost during life-cycle evolution, great emphasis is placed on capturing software investment when re-hosting code on new state-of-the-art target systems. To complicate matters, quite often, the fielded system stays in use in parallel with the new system for some period of time. Simultaneous support of different target baselines leads to issues in supporting both systems and migrating similar changes between each baseline. This paper describes the methods used to develop a common abstraction layer in support of the Air Force Research Laboratory (AFRL) Information Directorate's Multi-Platform Open System Transition Technology Demonstration program (MOST-TD) program. The specific issues addressed were: Ada83 to Ada95 conversion, dealing with compiler specific extensions, vendor provided libraries and the strongly-typed data types associated with them. A special emphasis was placed on minimizing or eliminating the computational burden that might be introduced by inserting an additional layer of software on each system. Through careful use of these techniques the author has added an abstraction layer to a fielded avionics operational flight program, while incurring no additional computational burden and actually obtaining a slight overall computational savings. The abstraction layer has already been delivered back into the original platform and has successfully completed flight test. The real savings realized from change isolation using software abstraction is obtained in the maintenance costs of supporting code bases for multiple target systems as well as host-based development