Statistical evaluation of data allocation optimization techniques with application to computer integrated manufacturing

摘要

A statistical approach is used to analyze the performance of data allocation optimization techniques in distributed computer systems. This approach is applied to a computer integrated manufacturing (CIM) system with hierarchical control constraints. The performance of some existing general purpose optimization techniques are compared with heuristics specifically developed for allocating data in a CIM system. Data allocation optimization techniques were actually implemented and performance was evaluated for a 250 node apparel CIM factory. The specific issues addressed include: (1) solution quality in terms of the minimum system wide input/output (I/O) task operating cost, (2) the probability of finding a better solution, (3) the performance of optimization heuristics in terms of their figure-of-merit, and (4) the impact of I/O task activation scenarios on the choice of an optimization technique for solving the CIM data allocation problem.