Edge Computing-Aided Coded Vertical Federated Linear Regression

Dai, Mingjun; Zheng, Ziying; Hong, Zhaoyan; Zhang, Shengli; Wang, Hui

doi:10.1109/tccn.2022.3174615

Cited by 6 publications

(1 citation statement)

References 25 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…For example, unmanned aerial vehicles (UAVs) belong to such an IoT network, which requires strictly low computing or learning duration. To meet such strict computation duration/delay requirement, edge computing/edge learning/federated learning 2 that employs IoT terminals within the network may be brought into play, which partitions the original computation task into several parallel small computation tasks.…”

Section: Introductionmentioning

confidence: 99%

Low overhead vector codes with combination property and zigzag decoding for edge‐aided computing in UAV network

Dai,

Huang,

Wang

et al. 2024

Computational Intelligence

Self Cite

View full text Add to dashboard Cite

Codes that possess combination property (CP) and zigzag decoding (ZD) simultaneously (CP‐ZD) has broad application into edge aided distributed systems, including distributed storage, coded distributed computing (CDC), and CDC‐structured distributed training. Existing CP‐ZD code designs are based on scalar code, where one node stores exactly one encoded packet. The drawback is that the induced overhead is high. In order to significantly reduce the overhead, vector CP‐ZD codes are designed, where vector means the number of stored encoded packets in one node is extended from one to multiple. More specifically, in detailed code construction, cyclic shift is proposed, and the shifts are carefully designed for cases that each node stores two, three, and four packets, respectively. Comparisons show that the overhead is reduced significantly.

show abstract

Section: Introductionmentioning

confidence: 99%