1: \begin{abstract}
2: In this paper, the problem of federated learning (FL) through digital communication between clients and a parameter server (PS) over a multiple access channel (MAC), also subject to differential privacy (DP) constraints, is studied.
3: %
4: More precisely, we consider the setting in which clients in a centralized network are prompted to train a machine learning model using their local datasets.
5: %
6: The information exchange between the clients and the PS takes places over a MAC channel and must also preserve the DP of the local datasets.
7: %simultaneous
8: %communications with a PS with the aim of better utilizing the computational resources in the network.
9: %
10: %We also consider the additional constraint in which
11: %The communication between the users and the PS takes place over a MAC
12: %
13: Accordingly, the objective of the clients is to minimize the training loss subject to (i) rate constraints for reliable communication over the MAC and (ii) DP constraint over the local datasets.
14: %
15: %while also satisfying the privacy and transmission rate constraints over the MAC, the digital distributed stochastic gradient descent (D-DSGD) is performed by each client.
16: %
17: For this optimization scenario, we proposed a novel consensus scheme in which digital distributed stochastic gradient descent (D-DSGD) is performed by each client.
18: %
19: To preserve DP, a digital artificial noise is also added by the users to the locally quantized gradients.
20: %
21: The performance of the scheme is evaluated in terms of the convergence rate and DP level for a given MAC capacity.
22: %
23: The performance is optimized over the choice of the quantization levels and the artificial noise parameters.
24: %
25: %to achieve efficiency in terms of convergence are investigated, subject to privacy constraint and capacity limit of the MAC channel.
26: %
27: Numerical evaluations are presented to validate the performance of the proposed scheme.
28: \end{abstract}
29: