A BDI Agent-Based Asynchronous Scheduling Framework For Cloud Computing

Task scheduling is a critical challenge in cloud computing, where uncertainties and distributed nature make traditional centralized methods inefficient. This paper presents a decentralized Belief-Desire-Intention agent-based scheduling framework for cloud computing, which excels in distributed environments and handling uncertainties. To avoid communication stuck caused by real-world uncertainties, the framework employs an asynchronous communication protocol with a notify listener. The framework addresses scheduling and rescheduling stages, considering uncertain events that may disrupt task executions and stuck agents. Two algorithms are proposed for scheduling and rescheduling processes respectively, alongside a novel cycle recommendation algorithm to minimize information synchronization issues. The framework has been implemented in JADEX and tested in cloud computing environments. Experimental results demonstrate that the framework minimizes task makespan, balances resource utilization, and maximizes task success rate in resolving uncertain events.