Authors:
Daniil A. Zorin
and
Valeriy A. Kostenko
Affiliation:
Lomonosov Moscow State University, Russian Federation
Keyword(s):
Systems Design, Job Shop Scheduling, Scheduling Algorithms, Reliability, Multiprocessor Systems, Optimization Problems, Real-Time Systems.
Related
Ontology
Subjects/Areas/Topics:
Applications
;
Artificial Intelligence
;
Knowledge Discovery and Information Retrieval
;
Knowledge-Based Systems
;
Methodologies and Technologies
;
Operational Research
;
Optimization
;
Pattern Recognition
;
Resource Allocation
;
Scheduling
;
Software Engineering
;
Symbolic Systems
Abstract:
This paper describes a method of job shop scheduling and co-designing a multiprocessor system with the minimal number of processors. The program is represented with a direct acyclic graph, and there is a fixed real-time deadline as well as a restriction on the reliability of the system. The system is supposed to tolerate both hardware and software faults. A simulated annealing algorithm is proposed for the problem, and it is evaluated both experimentally and theoretically in terms of asymptotic convergence. The algorithm is also applied to a practical problem of scheduling in radiolocation systems.