Collaborative Caching in Edge Computing via Federated Learning and Deep Reinforcement Learning
Algorithm 2
The edge cache algorithm based on Dueling-DDQN.
1: Input: The capacity of the experience replay poll , train starts , size of minibatch , discount factor ,-greedy exploration , learning rate , number of episodes , target network update period
2: Initialization:
3: For ep do
4: Input initial state space .
5: Fordo
6: Choose an action via -greedy policy
7: Execute action and get the next status and reward , judge whether is a terminal state.
8: Put the sample into the experience replay pool.
9: Ifthen
10: Randomly sample training samples with a minibatch size K from the experience replay pool .