TY - GEN
T1 - Software Runs Everything Off-World
T2 - AIAA Aviation Forum and ASCEND, 2024
AU - Linger, Richard
AU - McHugh, John
AU - Mili, Ali
AU - Pleszkoch, Mark
AU - Ghardallou, Wided
AU - McGaughey, Jack
N1 - Publisher Copyright:
© 2024, American Institute of Aeronautics and Astronautics Inc, AIAA. All rights reserved.
PY - 2024
Y1 - 2024
N2 - It is no exaggeration to say that software drives everything off-world, and no space mission can succeed without it. As earth-bound societies accelerate space exploration, it is critical that mission software be correct and secure. Because of the high stakes, considerable time and energy is expended in space software verification and security analysis. But faults and compromises in operational software persist despite best efforts. These risks are being magnified by the growth in autonomous and AI-generated software. Current verification methods will always be useful, but given the high consequences of failure, a need exists for more comprehensive and efficient verification techniques. The emerging technology of software behavior computation holds promise to fill this gap. The mathematics-based behavior computation process, known as Function Extraction (FX), produces the as-built specification of a program, whether human- or AI-generated. FX provides domain-to-range coverage of all behavior and subsumes all test cases that could be executed. Computed behavior is a new software engineering artifact that can enable a new approach to functional verification and security analysis. The computations reveal how variable values are computed in programs, not just their final values as in testing. This paper describes behavior computation technology as implemented in an FX prototype, and illustrates the process for 1) controlling complexity for human understanding, 2) scaling computations across program hierarchies, 3) verifying behavior of an imagined space habitat environmental controller, and 4) computing the behavior of AI-generated code prior to use.
AB - It is no exaggeration to say that software drives everything off-world, and no space mission can succeed without it. As earth-bound societies accelerate space exploration, it is critical that mission software be correct and secure. Because of the high stakes, considerable time and energy is expended in space software verification and security analysis. But faults and compromises in operational software persist despite best efforts. These risks are being magnified by the growth in autonomous and AI-generated software. Current verification methods will always be useful, but given the high consequences of failure, a need exists for more comprehensive and efficient verification techniques. The emerging technology of software behavior computation holds promise to fill this gap. The mathematics-based behavior computation process, known as Function Extraction (FX), produces the as-built specification of a program, whether human- or AI-generated. FX provides domain-to-range coverage of all behavior and subsumes all test cases that could be executed. Computed behavior is a new software engineering artifact that can enable a new approach to functional verification and security analysis. The computations reveal how variable values are computed in programs, not just their final values as in testing. This paper describes behavior computation technology as implemented in an FX prototype, and illustrates the process for 1) controlling complexity for human understanding, 2) scaling computations across program hierarchies, 3) verifying behavior of an imagined space habitat environmental controller, and 4) computing the behavior of AI-generated code prior to use.
UR - http://www.scopus.com/inward/record.url?scp=85203682412&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=85203682412&partnerID=8YFLogxK
U2 - 10.2514/6.2024-4850
DO - 10.2514/6.2024-4850
M3 - Conference contribution
AN - SCOPUS:85203682412
SN - 9781624107160
T3 - AIAA Aviation Forum and ASCEND, 2024
BT - AIAA Aviation Forum and ASCEND, 2024
PB - American Institute of Aeronautics and Astronautics Inc, AIAA
Y2 - 29 July 2024 through 2 August 2024
ER -