On the conceptual design of multi-agent system for load balancing using multi-class queueing networks

In manufacturing and service systems, processing or attention is not complete until the customer has circulated within the process of sequentially at different stages with waiting lines phenomena. The complexity of the analysis required for modeling such systems in terms of key performance factors, as well as the ever-increasing need to control complex dynamic systems under uncertainty, it has led to the formulation of hybrid theoretical - practical optimization methods or to allow for the approximation of functions through simple structures with resilience to situations of features multi-variable and even non-linearity. We review the general techniques and methods for the analysis of queueing network systems, exploring its alternative modeling using the paradigm called multi-agent intelligent systems (MAS) and soft-computing techniques. A methodology is proposed based on MAS, artificial intelligence and operations research techniques to achieve a load balancing approach in multi-class queueing networks systems. The proposed methodology considers the most important factors identified by various authors to sustain the relevance of the schema integration techniques suggested.