A Realistic Decision Making for Task Allocation in Heterogeneous Multi-agent Systems

Task allocation is one of the keys to maximize organizational benefits by handling as many tasks as possible. Many computational multi-agent systems use agent's capability for task allocation. When a task arrives at the queue to be delivered a task allocator will determine which takes the task by finding the best-capable agent. In real world situation, each agent should not only consider the new task with their capability, but also tasks that they are currently handling before sending their capability to the task allocator. This research study proposes a CPU-scheduling based algorithm to allow agents to perform rational decision making when they think that they can handle the new task while taking care of its current tasks. The result shows that applying algorithm provide a significant improvement of their performance.