Quality Enhanced Multimedia Content Delivery for Mobile Cloud with Deep Reinforcement Learning

Table 5

Adaptation algorithm parameters.

Algorithm	Parameters	Values
FESTIVE	Target buffer	15 s
	Buffer Randomness	0.25 s
		10
MPC	Buffer Size	30 s
		2
QDASH		0.1
	Policy	Softmax
A3C		0.99
	Policy	Ε greedy
	Batch Size	128
	Actor learning rate	10-4
	Critic learning rate	10-3
ADQ	Hidden neurons Nh	256
	Learning rate	10-3
	Policy	Greedy Policy
	Lambda	0.9