Abstract
While small, low-cost satellites continue to increase in capability and popularity, their reliability remains a problem. Traditional techniques for increasing system reliability are well known to satellite developers. They include the use of radiation-hardened and screened components, extensive cold redundancy and thorough test campaigns. However, the implementation of these techniques on small, low-cost satellites is often limited due to intrinsic mass, volume and budgetary restrictions. Aiming for graceful degradation, therefore, may be a more promising route. Inspired by the robustness of single-celled and multi-cellular biological organisms, bio-inspired computing systems, multi-agent systems, and modular spacecraft concepts, this work presents the design, implementation and analysis of an artificial, cell-based system architecture. Named the Satellite Stem Cell Architecture, the proposed system aims to replace a significant portion of a typical satellite’s bus avionics with a set of initially identical, mass produced, artificial cells. Analogous to their biological counterparts, the artificial cells can differentiate during runtime to take on a variety of tasks, thanks to a set of artificial proteins. Each cell reconfigures its own proteins within the context of a system-wide, distributed task management strategy. In this way, essential tasks can be maintained, even as system cells fail. The Satellite Stem Cell Architecture differs from existing bio-inspired computing systems by extending the concept to include reconfigurable interfaces to real-world sensors and actuators, and by its inclusion of a set of middleware which turns each cell into a multi-agent platform. Furthermore, an emphasis is placed on practical applicability, with power consumption, volume and production cost driving the implementation. A detailed description of the artificial cell hardware, and multi-agent middleware, is given. Additionally, two CubeSat-scale, practical implementations of the architecture are described. While one, which forms the payload interface computer of the SMESAT CubeSat, demonstrates only a subset of the proposed multicellular features, the other is a full testbed based on two artificial cells of four proteins each. To compare the reliability of the proposed architecture to traditional forms of redundancy, an analytical reliability equation is derived for predicting the lifetimes of multicellular systems. It is shown that determining the optimal configuration of proteins per cell and cells per system is complex, as different configurations are optimal during different phases of the mission lifetime. Nevertheless, a set of trends in system behaviour are discovered, which will prove useful to system designers. Using a purpose-developed, multicellular simulation environment, the results of the analytical work are verified, and further problems relating to peripheral interfaces and cross-strapping are investigated. Finally, using measured characteristics of the implemented testbed and the derived analytical lifetime predictions, the Artificial Stem Cell Architecture is compared against traditional CubeSat and microsatellite avionics suites. The results show that the proposed architecture gives increasing reliability and performance benefits with increased scale, and that while its power consumption overheads make it prohibitive for implementation on CubeSats, it is well-suited to microsatellites.