A New Differential Privacy Crowdsensing Scheme Based on the Multilevel Interactive Game

Table 1

System parameters used in the simulation experiments.

Sensing Tasks	Total MCS cycles	-LDP value	Service Duration
	4.5 K sensing cycles/s	0.1	3,000 sec (50 min)
	6.5 K sensing cycles/s	0.15	2,700 sec (45 min)
	9 K sensing cycles/s	0.2	2,880 sec (48 min)
	12 K sensing cycles/s	0.25	2,760 sec (46 min)
Parameter	Value	Description
	4	the number of DP controllers
n	100	the number of mobile devices
	1.25, 2	the minimum and maximum privacy levels
	1, 1.5, 2, 2.5, 3	the MCS service price per bit
	200, 600 cycles/s	the minimum and maximum sensing data amounts
	1.2, 2	the minimum and maximum proportional sharing factors
	0.1	the learning rate for DP controllers
	0.9	the recency parameter to select a DP controller
	0.5	the parameter to avoid a premature convergence
	4	the weight parameter to select a DP controller
,	1.5, 1	the estimation parameters to calculate the DP cost
,	1.1, 0.7	the estimation parameters to calculate the service cost
	1.5	the learning rate for mobile devices
,	0.1, 0.1	the learning rate parameters for the MCS servers
	0.5	the control factor for the MCS server