Algorithm 1.

CNN-FL model based on local differential privacy

1 For Iteration t do

/* Service-Terminal: * /

2 ${\omega }_t=\frac{1}{Q}{\displaystyle \sum _{q=1}^Q{\omega }_t^k;}$

3 Send ωt to each participant;

/* participants: * /

4 for Participant q do

5 Do Localized differential privacy

6 ${\omega }_t^q={\omega }_t;$

7 For Local epoch e do

8 ${\omega }_t^q={\omega }_t^q-\eta \frac{\partial }{\partial X^q}Z\wp$

9 End

10 End

11 End