Learning-Based QoS Control Algorithms for Next Generation Internet of Things

Table 1

System parameters used in the simulation experiments.

Traffic class	Message application	Bandwidth requirement	Connection duration; average/sec
I	Delay-critical emergency applications	32 Kbps	30 sec (0.5 min)
II	Event-related applications	32 Kbps	120 sec (2 min)
		64 Kbps	180 sec (3 min)
III	General applications	128 Kbps	120 sec (2 min)
		256 Kbps	180 sec (3 min)
IV	Multimedia applications	384 Kbps	300 sec (5 min)
		512 Kbps	120 sec (2 min)
Parameter	Value	Description
	1.2	The player’s willingness to pay for his perceived service worth
	−0.2	The control parameter between throughput and delay
	1.1	The estimation parameters of the cost function
	0.3	A probability that the user keeps staying at the selected strategy
	1	Predefined minimum bound for stable status
	1	The price for resource unit in the cost function
m	3	The number of strategies for QoS schedulers
	1	A parameter to control the learning size
	0.8	A discount factor for the respective probability distribution
	1	A control parameter on the Boltzmann distribution