Bio-inspired reconfigurable architecture for reliable systems

Living organisms are the most complex systems ever known to man. Failures in them are common, yet because self-diagnostic activity and self-healing continues incessantly, overall functional reliability of the system is extremely high. Nature achieves this, not by solving the reliability problem at the system level, but rather on the level of its 'primitive' constituents, the cells. If we apply the fundamental characteristics of the living organisms to the silicon world of electronic systems, the results are cell-array based hardware systems with biological-like fault tolerant properties. Such bio-cell array architectures are embryonic systems with reconfigurable characteristics similar to those found in cellular organisms, where reconfiguration mechanism occurs during embryonic development. In electronics however, it occurs when the individual cell's functionality, and that of the entire. system at 'birth' is defined, or when a fault in its operation is detected Our long-term goal is the development of a bio-inspired fault-tolerant hardware system. However, this paper focuses on our first steps in using a genetic algorithm (GA) to find suitable DNA for our reconfigurable embryonic array architecture. We first describe how we envisage the GA operating within the framework of our reconfigurable system. We then go on to present the results of a preliminary experiment designed to test these ideas, focusing on the configuration settings of a single cell in what will eventually become an array of cells.