Federated learning is a promising strategy for indoor localization that can reduce the labor cost of constructing a fingerprint dataset in a distributed training manner without privacy disclosure. However, the traffic generated during the whole training process of federated learning is a burden on the up-and-down link, which leads to huge energy consumption for mobile devices. Moreover, the non-independent and identically distributed (Non-IID) problem impairs the global localization performance during the federated learning. This paper proposes a communication-efficient FedAvg method for federated indoor localization which is improved by the layerwise asynchronous aggregation strategy and layerwise swapping training strategy. Energy efficiency can be improved by performing asynchronous aggregation between the model layers to reduce the traffic cost in the training process. Moreover, the impact of the Non-IID problem on the localization performance can be mitigated by performing swapping training on the deep layers. Extensive experimental results show that the proposed methods reduce communication traffic and improve energy efficiency significantly while mitigating the impact of the Non-IID problem on the precision of localization.
Jinjie LIANG
Guangdong University of Technology
Zhenyu LIU
Guangdong University of Technology
Zhiheng ZHOU
South China University of Technology
Yan XU
Guangdong University of Technology
The copyright of the original papers published on this site belongs to IEICE. Unauthorized use of the original or translated papers is prohibited. See IEICE Provisions on Copyright for details.
Copy
Jinjie LIANG, Zhenyu LIU, Zhiheng ZHOU, Yan XU, "Communication-Efficient Federated Indoor Localization with Layerwise Swapping Training-FedAvg" in IEICE TRANSACTIONS on Fundamentals,
vol. E105-A, no. 11, pp. 1493-1502, November 2022, doi: 10.1587/transfun.2022EAP1004.
Abstract: Federated learning is a promising strategy for indoor localization that can reduce the labor cost of constructing a fingerprint dataset in a distributed training manner without privacy disclosure. However, the traffic generated during the whole training process of federated learning is a burden on the up-and-down link, which leads to huge energy consumption for mobile devices. Moreover, the non-independent and identically distributed (Non-IID) problem impairs the global localization performance during the federated learning. This paper proposes a communication-efficient FedAvg method for federated indoor localization which is improved by the layerwise asynchronous aggregation strategy and layerwise swapping training strategy. Energy efficiency can be improved by performing asynchronous aggregation between the model layers to reduce the traffic cost in the training process. Moreover, the impact of the Non-IID problem on the localization performance can be mitigated by performing swapping training on the deep layers. Extensive experimental results show that the proposed methods reduce communication traffic and improve energy efficiency significantly while mitigating the impact of the Non-IID problem on the precision of localization.
URL: https://global.ieice.org/en_transactions/fundamentals/10.1587/transfun.2022EAP1004/_p
Copy
@ARTICLE{e105-a_11_1493,
author={Jinjie LIANG, Zhenyu LIU, Zhiheng ZHOU, Yan XU, },
journal={IEICE TRANSACTIONS on Fundamentals},
title={Communication-Efficient Federated Indoor Localization with Layerwise Swapping Training-FedAvg},
year={2022},
volume={E105-A},
number={11},
pages={1493-1502},
abstract={Federated learning is a promising strategy for indoor localization that can reduce the labor cost of constructing a fingerprint dataset in a distributed training manner without privacy disclosure. However, the traffic generated during the whole training process of federated learning is a burden on the up-and-down link, which leads to huge energy consumption for mobile devices. Moreover, the non-independent and identically distributed (Non-IID) problem impairs the global localization performance during the federated learning. This paper proposes a communication-efficient FedAvg method for federated indoor localization which is improved by the layerwise asynchronous aggregation strategy and layerwise swapping training strategy. Energy efficiency can be improved by performing asynchronous aggregation between the model layers to reduce the traffic cost in the training process. Moreover, the impact of the Non-IID problem on the localization performance can be mitigated by performing swapping training on the deep layers. Extensive experimental results show that the proposed methods reduce communication traffic and improve energy efficiency significantly while mitigating the impact of the Non-IID problem on the precision of localization.},
keywords={},
doi={10.1587/transfun.2022EAP1004},
ISSN={1745-1337},
month={November},}
Copy
TY - JOUR
TI - Communication-Efficient Federated Indoor Localization with Layerwise Swapping Training-FedAvg
T2 - IEICE TRANSACTIONS on Fundamentals
SP - 1493
EP - 1502
AU - Jinjie LIANG
AU - Zhenyu LIU
AU - Zhiheng ZHOU
AU - Yan XU
PY - 2022
DO - 10.1587/transfun.2022EAP1004
JO - IEICE TRANSACTIONS on Fundamentals
SN - 1745-1337
VL - E105-A
IS - 11
JA - IEICE TRANSACTIONS on Fundamentals
Y1 - November 2022
AB - Federated learning is a promising strategy for indoor localization that can reduce the labor cost of constructing a fingerprint dataset in a distributed training manner without privacy disclosure. However, the traffic generated during the whole training process of federated learning is a burden on the up-and-down link, which leads to huge energy consumption for mobile devices. Moreover, the non-independent and identically distributed (Non-IID) problem impairs the global localization performance during the federated learning. This paper proposes a communication-efficient FedAvg method for federated indoor localization which is improved by the layerwise asynchronous aggregation strategy and layerwise swapping training strategy. Energy efficiency can be improved by performing asynchronous aggregation between the model layers to reduce the traffic cost in the training process. Moreover, the impact of the Non-IID problem on the localization performance can be mitigated by performing swapping training on the deep layers. Extensive experimental results show that the proposed methods reduce communication traffic and improve energy efficiency significantly while mitigating the impact of the Non-IID problem on the precision of localization.
ER -