Context-aware resource allocation for IoRT-aware business processes based on decentralized multi-agent reinforcement learning

In Industry 4.0 (I4.0), IoRT-aware Business Processes aim to automate classic Business Processes (BP) by integrating IoT and robotics capacities. However, executing these processes inside the enterprise may be costly due to the required software and hardware components. To overcome this deficiency, the Business Process Outsourcing (BPO) strategy can be used to execute an IoRT-aware BP using external environments such as Fog and Cloud. In these environments, heterogeneous resources may have different specifications, which makes allocating Fog and Cloud resources challenging. Therefore, this work addresses the resource allocation (RA) issue for outsourcing an IoRT-aware BP. Toward this objective, we propose an optimal context-aware RA approach based on the decentralized multi-agent reinforcement learning (MARL) technique. The effectiveness and feasibility of the proposed context-aware RA-based decentralized MARL are demonstrated through a set of experiments. The preliminary experimental evaluation of the proposed approach demonstrates a high precision value and an encouraging recall value regarding other RA approaches.