Parallel machine scheduling problems with partial information: Distributed decision models and algorithms

In this paper Lagrangian-based distributed algorithms for scheduling jobs on unrelated parallel machines are presented. In the algorithms, the scheduling process is the result of a cooperation process among several decision makers (DMs). DMs have a local knowledge of the system, and the possibility to decide which type of information to exchange each other. Our focus is to investigate the performance of different algorithms based on different knowledge degrees of the parallel machine system. The implementation issues and the effectiveness of the algorithms are analysed via simulation. Extensive experimental results are reported, allowing to evaluate the trade–off between knowledge degree and system performance.