SBIR-STTR Award

Gamma ray induced failures are major threats to computer systems deployed in space. If not treated properly, radiation errors can seriously jeopardize a space or missile mission. In common practice, safety-critical computer equipment for space usually employs hardening combined with hardware redundancy to cope with radiation-induced errors. However, this technology is expensive, especially for payload developers. This project investigates an alternative, much less expensive architecture for the development of reliable payload controllers that relies on off-the-shelf computers and software, multi-programming, and on software implemented fault tolerance (SIFT) to achieve reliability. In the SIFT approach, memory is replicated, and multiple copies of the same program are executed and correct results of computations are decided by majority vote. The objective of the project is to design the architecture of a reliable payload controller, implement it, and to perform experiments that show its suitability to cope with gamma radiation. Moreover, the techniques developed in this project will be computerized in a payload controller shell that enables payload developers to employ these technologies in payloads. The proposal develops techniques and software to improve the performance, endurance, and survivability of space and missile payloads.

Keywords:
Endurance Endurance Survivability Survivability Fault-Tolerance Fault-Tolerance Self-Healing

Radiation induced errors can jeopardize space and missile missions. Safety-critical computers for space usually employ hardware redundancy with hardening to cope with radiation-induced errors. This technology is expensive, and its efficacy uncertain. Recently discovered gamma rays, while not as common as typical solar radiation, are more deadly, and can create subtle failures that are threats to space deployed computers. This proposes an alternative, less expensive software implemented fault tolerance (SIF) architecture for more fault tolerant payload controllers which can provide reliability with off-the-shelf computer. In SIFT, multiple multiprogrammed copies of programs are executed, and correct results decided by voting procedures. Phase I demonstrated the feasibility of SIFT to achieve space systems reliability, and provided plans about how to economically implement and empirically verify SIFT architecture in Phase II. Phase I designed the architecture of a reliable payload controller, and the means of implementing it. Phase II will implement the architecture and perform experiments that show its suitability to cope with radiation. Moreover, the techniques developed will be computerized in a controller shell that will enable developers to employ this technology in space payloads. The project will develop software techniques which improve performance, endurance, and survivability of space and missile payloads.|

Benefits:
Products resulting from this project have universal applicability to applications requiring high reliability, such as real-time process control. Two dual use examples are: 1) Payloads must be able to recover from radiation-induced failures. 2) Hazardous processes, such as oil refineries, must be able to recover from unexplainable failures by return to a stable configuration.|

Keywords:
Endurance Endurance Survivability Survivability Fault-Tolerance Fault-Tolerance Self-Healing