On the Generation of Variable Structure Distributed Architectures
Abstract
A methodology to model and generate variable structure distributed intelligence systems is presented. First, the objects and the functional entities that belong to the system are defined, as well as generic interactions between them. A mathematical framework is developed to represent these interactions. This framework is embedded in Colored Petri Net theory, which is used as the basic technique to generate variable structures. The set of variable structures that satisfy both general constraints and user-defined requirements is analyzed with results from Lattice theory. A class of solutions to the design problem is characterized. This class corresponds to the variable structures whose variability corresponds exactly to the requirements of the user. This class of solution is decomposed into subsets of structures with the same input links. Each subset is delimited by minimal and maximal elements. There is a layer of partially ordered subsets between one minimal element and one maximal element. Each layer is a lattice. The methodology is applied to two illustrative examples, the coordination of tasks in a submarine and the coordination of tasks among air traffic controllers at an airport. Finally, policy considerations of the research on distributed intelligence systems are presented.
Document Details
- Document Type
- Technical Report
- Publication Date
- May 01, 1989
- Accession Number
- ADA208953
Entities
People
- Jacques J. Demael
Organizations
- Massachusetts Institute of Technology