Research on the Difficulty of Mobile Node Deployment’s Self-Play in Wireless Ad Hoc Networks Based on Deep Reinforcement Learning

Table 1

The structure of deep reinforcement learning neural network under different grid sizes.

Structure	grid	grid	grid	grid
Input layer	(6,6,3)	(8,8,3)	(10,10,3)	(16,16,3)
Convolutional layer 1	(6,6,32)	(8,8,32)	(10,10,32)	(16,16,32)
Convolutional layer 2	(6,6,64)	(8,8,64)	(10,10,64)	(16,16,64)
Convolutional layer 3	(6,6,128)	(8,8,128)	(10,10,128)	(16,16,128)
Convolutional layer 4	(6,6,4)	(8,8,4)	(10,10,4)	(16,16,4)
Convolutional layer 5	(6,6,2)	(8,8,2)	(10,10,2)	(16,16,2)
Output layer	(36,2)	(64,2)	(100,2)	(256,2)