FDDL is a cooperative cache approach.
This repository contains cache environment and interaction between nodes of FDDL.
The implementation of Federated Learning can be referenced to FedAvg and adaptive-federated-learning.
The implementation of Deep Reinforcement Leanring can be referenced to ElegantRL.
The dataset is built in the following form:
-
Prepare the original dataset as:
<timestamp, contentID, contentSize>
-
Extract features as:
<timestamp, contentID, contentSize, frequency, timestamp between two same content, interval between two same content, frequency in different time spans>