TY - GEN
T1 - Random Orthogonalization for Federated Learning in Massive MIMO Systems
AU - Wei, Xizixiang
AU - Shen, Cong
AU - Yang, Jing
AU - Poor, H. Vincent
N1 - Funding Information:
XW and CS were partially supported by the US National Science Foundation (NSF) under ECCS-2033671. JY was supported in part by the NSF under CNS-1956276 and CNS-2114542. HVP was supported in part by the NSF under CCF-1908308.
Publisher Copyright:
© 2022 IEEE.
PY - 2022
Y1 - 2022
N2 - We propose a novel uplink communication method, coined random orthogonalization, for federated learning (FL) in a massive multiple-input and multiple-output (MIMO) wireless system. The key novelty of random orthogonalization comes from the tight coupling of FL model aggregation and two unique characteristics of massive MIMO - channel hardening and favorable propagation. As a result, random orthogonalization can achieve natural over-the-air model aggregation without requiring transmitter side channel state information, while significantly reducing the channel estimation overhead at the receiver. Theoretical analyses with respect to both communication and machine learning performances are carried out. In particular, an explicit relationship among the convergence rate, the number of clients and the number of antennas is established. Experimental results validate the effectiveness and efficiency of random orthogonalization for FL in massive MIMO.
AB - We propose a novel uplink communication method, coined random orthogonalization, for federated learning (FL) in a massive multiple-input and multiple-output (MIMO) wireless system. The key novelty of random orthogonalization comes from the tight coupling of FL model aggregation and two unique characteristics of massive MIMO - channel hardening and favorable propagation. As a result, random orthogonalization can achieve natural over-the-air model aggregation without requiring transmitter side channel state information, while significantly reducing the channel estimation overhead at the receiver. Theoretical analyses with respect to both communication and machine learning performances are carried out. In particular, an explicit relationship among the convergence rate, the number of clients and the number of antennas is established. Experimental results validate the effectiveness and efficiency of random orthogonalization for FL in massive MIMO.
KW - Convergence Analysis
KW - Federated Learning
KW - Massive MIMO
UR - http://www.scopus.com/inward/record.url?scp=85137270041&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=85137270041&partnerID=8YFLogxK
U2 - 10.1109/ICC45855.2022.9838988
DO - 10.1109/ICC45855.2022.9838988
M3 - Conference contribution
AN - SCOPUS:85137270041
T3 - IEEE International Conference on Communications
SP - 3382
EP - 3387
BT - ICC 2022 - IEEE International Conference on Communications
PB - Institute of Electrical and Electronics Engineers Inc.
T2 - 2022 IEEE International Conference on Communications, ICC 2022
Y2 - 16 May 2022 through 20 May 2022
ER -