The online reconfiguration of a distributed on-board computer: The time and network behaviour of a dependable scheduling algorithm

Abstract

On-board Computers (OBCs) are at the centre of space-faring systems. With the increasing demand for cost-effective computing power in space, using high-performance commercial-off-the-shelf (COTS) components for OBCs has gained significant traction. COTS components, however, do not provide the necessary fault tolerance mechanisms. The ScOSA (Scalable On-board computing for Space Avionics) architecture uses COTS components in a distributed system to provide more computing performance and dependability. The effects of node failures are mitigated by removing the failed node from the system through reconfiguration. A reconfiguration is performed by using a set of predetermined configurations, which hinders system scalability due to exponentially increasing memory consumption depending on the number of nodes. This paper continues the work on the ScOSA online reconfiguration algorithm as a solution to this scalability problem. The online reconfiguration algorithm, which has been integrated into a scheduler, makes task scheduling decisions at run-time, eliminating the need for predetermined configurations. The six-phase scheduling mechanism uses the real-time state of the system and is a step towards higher dependability in distributed on-board computing. New test scenarios have been introduced to provide insight into the temporal and network behaviour of online reconfiguration. By evaluating in terms of time, network traffic and memory usage, it is shown that online reconfiguration is not only capable of dynamically generating configurations but also providing a solution to the scalability problem for systems with varying numbers of both nodes and tasks.