2020년 컴퓨터종합학술대회
Current Result Document :
| 한글제목(Korean Title) |
Update and Channel Aware User Selection with Bandwidth Allocation for Federated Learning in Wireless Networks |
| 영문제목(English Title) |
Update and Channel Aware User Selection with Bandwidth Allocation for Federated Learning in Wireless Networks |
| 저자(Author) |
Chit Wutyee Zaw
Choong Seon Hong
|
| 원문수록처(Citation) |
VOL 47 NO. 01 PP. 1013 ~ 1015 (2020. 07) |
한글내용
(Korean Abstract) |
|
영문내용
(English Abstract) |
Since the performance of the global model in Federated Learning (FL) is highly influenced by the participant users, the user selection has become a challenging approach. Due to the scarcity of wireless resources, it is impractical for all users to participate in FL. To improve the performance of the global model, the update of the local models is a critical parameter to take into account. In addition, the channel condition of the participant users should be considered to successfully decode the update of the local models. In this paper, we propose a user selection approach in which the selection of users is decided based on their updates and channel conditions. These parameters are assigned with a score which is controlled by the users. Moreover, the bandwidth allocation is determined based on the channel conditions where more bandwidth resources are allocated to the users with lower channel gains in order to minimize the time taken for one computation round. |
| 키워드(Keyword) |
|
| 파일첨부 |
PDF 다운로드
|
