Optimizing Proximity Strategy for Federated Learning Node Selection in the Space-Air-Ground Information Network for Smart Cities

TY - JOUR

T1 - Optimizing Proximity Strategy for Federated Learning Node Selection in the Space-Air-Ground Information Network for Smart Cities

AU - Wang, Weidong

AU - Li, Ping

AU - Li, Siqi

AU - Zhang, Jihao

AU - Zhou, Zijiao

AU - Wu, Dapeng Oliver

AU - Kim, Duk Kyung

AU - Zhang, Guangwei

AU - Gong, Peng

N1 - Publisher Copyright: IEEE

PY - 2024

Y1 - 2024

N2 - As the Internet of Things (IoT) technology and Artificial Intelligence (AI) technology continue to evolve, many envisaged concepts regarding smart cities are gradually becoming a reality. However, the proliferation of numerous IoT devices in smart cities has led to several challenges. The existing 5G networks are incapable of meeting the requirements of these devices in terms of channel capacity and network coverage. Additionally, traditional cloud-based centralized machine learning methods fail to ensure the privacy of user data. At this juncture, Space-Air-Ground Information Network, along with Federated Learning (FL), are perceived as viable solutions to address these issues. This paper focuses on addressing FL challenges in smart cities using the Space-Air-Ground Information Network. Here, data distribution heterogeneity leads to increased federated training time and higher energy costs. The paper begins by analyzing the reasons for the non-independent and non-identically distributed (NON-IID) data collected by devices in this scenario. Subsequently, from the perspective of device selection, the paper proposes a node selection model based on near-edge strategy optimization, termed "Low Node Selection in Federated Learning" (LCNSFL). Finally, the LCNSFL algorithm is compared with federated av-eraging algorithms based on random selection strategies and the FedProx algorithm. Experimental results demonstrate that the federated learning model aided by the LCNSFL algorithm achieves the target accuracy with fewer communication rounds, considerably reducing the required training time and energy costs compared to the other two algorithms.

AB - As the Internet of Things (IoT) technology and Artificial Intelligence (AI) technology continue to evolve, many envisaged concepts regarding smart cities are gradually becoming a reality. However, the proliferation of numerous IoT devices in smart cities has led to several challenges. The existing 5G networks are incapable of meeting the requirements of these devices in terms of channel capacity and network coverage. Additionally, traditional cloud-based centralized machine learning methods fail to ensure the privacy of user data. At this juncture, Space-Air-Ground Information Network, along with Federated Learning (FL), are perceived as viable solutions to address these issues. This paper focuses on addressing FL challenges in smart cities using the Space-Air-Ground Information Network. Here, data distribution heterogeneity leads to increased federated training time and higher energy costs. The paper begins by analyzing the reasons for the non-independent and non-identically distributed (NON-IID) data collected by devices in this scenario. Subsequently, from the perspective of device selection, the paper proposes a node selection model based on near-edge strategy optimization, termed "Low Node Selection in Federated Learning" (LCNSFL). Finally, the LCNSFL algorithm is compared with federated av-eraging algorithms based on random selection strategies and the FedProx algorithm. Experimental results demonstrate that the federated learning model aided by the LCNSFL algorithm achieves the target accuracy with fewer communication rounds, considerably reducing the required training time and energy costs compared to the other two algorithms.

KW - Computational modeling

KW - Costs

KW - Data models

KW - Federated learning

KW - Internet of Things

KW - Smart cities

KW - Training

KW - federated learning

KW - node selection model

KW - non-IID data

KW - smart cities

KW - space-air-ground information network

UR - https://www.scopus.com/pages/publications/85196703888

U2 - 10.1109/JIOT.2024.3416943

DO - 10.1109/JIOT.2024.3416943

M3 - Article

AN - SCOPUS:85196703888

SN - 2327-4662

SP - 1

JO - IEEE Internet of Things Journal

JF - IEEE Internet of Things Journal

ER -