A high dependability computer attainable by a tracking redundancy scheme

A high dependability computer configured with a new scheme is proposed by combining and sublating spatial diversity and time diversity which have already been proven. This paper starts with fault analysis of the on-board computer tested on the satellite "Hiten" followed by a study of the fault distribution over components constituting typical computer systems to optimize the reliability to cost ratio. Based upon these studies, the paper proposes a scheme, which we named "tracking redundancy". The paper introduces its generic definition and develops its details, introducing examples to be applied at different levels of granularity. The paper next evaluates the improvement in the system reliability expected for the scheme. The paper also studies the various influences of the scheme on the performance and cost together with their partial solutions. The paper gives suggestions for extension and degeneration of the scheme.