Table 3.

Simulation hyperparameters.

S. No	Hyperparameter	Value
1.	Learning rate (α)	0.001
2.	Discount factor (γ)	0.99
3.	Batch size	64
4.	Replay buffer size	100,000
5.	Target network update frequency	10,000 steps
6.	Exploration rate (ε)	Initial: 1.0
7.		Final: 0.01
8.		Decay: 0.995/episode
9.	Number of training episodes	10,000
10.	GNN layers	3 layers
11.	Hidden units per layer	128 units
12.	Dropout rate	0.5
13.	Optimizer	Adam
14.	Learning rate for GNN	0.0005
15.	Max episodes per simulation	500