Privacy-Enhanced Data Fusion for Federated Learning Empowered Internet of Things

Algorithm 2

Global data fusion with differential privacy protection.

	Input: Local model parameter , privacy budget , sensitivity , and Gaussian noise to be added satisfying
	Output: Global model parameter to be released
(1)	for each FL-based data fusion center do
(2)	Calculate the weighted average model by
(3)	Add Gaussian noise by , where
(4)	end for