Research Article
Optimization of LoRa SF Allocation Based on Deep Reinforcement Learning
Table 3
DQN algorithm parameters.
| Parameter | Value |
| Greedy policy | 0.9 | Batch size | 256 | Reward discount | 0.9 | Target update frequency | 100 | Memory capacity | 5000 | Actions | 18 | States | 7 |
|
|