Table 1 Simulations parameters
From: A deep reinforcement approach for computation offloading in MEC dynamic networks
The notation | Physical meaning | Value |
|---|---|---|
N | Mobile user | 20 |
K | Edge server | 4 |
\(f_k^{{\textrm{m}}ec}\) | Server capacity (CPU cycles) | \(\left[ 10,\,100\right]\)Â GHz |
\(f_m^{loc}\) | User computing capacity (CPU cycles) | \(\left[ 0.5,\,2\right]\)Â GHz |
D | Amount of user-generated task data | \(\left[ 1,\,5\right]\)Â Mbits |
B | System Bandwidth | 20Â MHz |
\({P_{\max }}\) | Maximum uplink transmit power | 0.2Â W |
\({\sigma ^2}\) | Background noise power | \(-100\)Â dBm |
\(h_i^k(t)\) | Channel gain | |
t | Time slot | 3Â ms |
\({\theta _i}(t)\) | The maximum tolerable delay | 3Â s |
C | Capacity of empirical replay buffer | 500 |
U | Capacity of the small batch of samples | 32 |
\(\varepsilon\) | Learning rate | 0.1 |
\(\phi\) | Reward | 0.9Â ms |