Optimal design of a distributed supersystem
David F. Palmer, James P. Ignizio, Catherine M. Murphy · afips · 1899
The design of distributed systems, in particular supersystems, requires an extraordinary number of decisions. Designers must define numbers and types of computing components, interconnection networks, software modules, and the allocation of software to hardware.This paper describes a formal approach to determining complex design decisions. The approach differs from most formal decision techniques as follows: (1) it generates multiple optimal design solutions; (2) it deals with multidimensional criteria; and (3) it has been implemented by an efficient (polynomial growth) algorithm. The approach uses fuzzy clustering, generalized goal programming, and a refined vectormax technique.