Deep Reinforcement Learning for Scheduling and Offloading in UAV-Assisted Mobile Edge Networks

Algorithm 2

DRUA algorithm solves the problem of UAV scheduling and offloading about UAV-assisted ESs.

Input: 

Output: ,

1: Using random parameter and to initialize the two DNNs, respectively;

2: Set the number of iterations M and training interval δ;

3: fort =1,2,...,Mdo

4: Generate relaxed action and ;

5:  Get and from Algorithm1;

6: Update memory by adding and ;

7:  ift mod =0 then

8:  Extract a batch of samples and from memory;

9: Using these samples training two DNNs, respectively, and update and by Adam;

10:  end if

15: end for