In recent years, Fog computing has become a hot topic in research areas because it facilitates overcoming resource constraints by offloading computationally intensive tasks to the remote cloud servers. Offloading computation tasks in fog computing is crucial for enabling mobile devices to support resource-intensive applications and minimize long-term system energy consumption.
   Traditional computation offloading possess high complexity, and it makes it ineffective for offloading decision in the fog computing model. Thus, Deep Reinforcement Learning based offloading empowers effectively learning the optimal offloading decision, minimizing long-term system energy consumption. Moreover, Multi-agent deep reinforcement learning learns dynamically interacting with their environment, and it can provide efficient resource management for fog computing in a distributed aspect.