Large system decomposition and simulation methodology using axiomatic analysis

Abstract

Abstract-Large, established systems are essentially impossible to replace and must be improved incrementally to incorporate disaster tolerance. The question of which components need to be upgraded or have added redundancy requires the system to be wholly analyzed or simulated. Unfortunately, extremely large systems are not amenable to comprehensive simulation studies due to the large computational complexity requirements. This research presents a new method that results in a dramatic decrease in simulation time from that required by an entire system simulation to the sum of the total time of simulations for decomposed subsystems. Axiomatic analysis principles of information and independence are used as part of the methodologies to determine the best decomposition boundaries between subsystems. This allows the smaller subsystems to have maximum independence so that the entire system response can be reconstructed from a linear combination of the individual subsystem simulation responses.