TY - GEN
T1 - A Task Placement System for Face Recognition Applications in Edge Computing
AU - Satake, Hayata
AU - Tani, Ryotaro
AU - Shigeno, Hiroshi
N1 - Publisher Copyright:
© 2020 IEEE.
Copyright:
Copyright 2020 Elsevier B.V., All rights reserved.
PY - 2020/1
Y1 - 2020/1
N2 - Mobile devices have certain restrictions such as processing power and network bandwidth. Edge computing can reduce load and delay by migrating the parts of application processing from mobile devices to edge servers. Edge computing is a method that extends cloud computing to the edge of network. Conventional schemes of face recognition based on edge computing can reduce the amount of network transmission data between the edge servers and the cloud servers by preprocessing data on the edge servers. However, the schemes execute each task on a certain resource, thus application response time may increase according to the conditions of computational resources and network. Hence, a task placement system considering the conditions of computational resources and network is necessary. This paper proposes a task placement system for face recognition applications in edge computing. The proposed system calculates estimated response time of possible task placements based on the task placement decision formula. The formula parameterizes power and CPU utilization of computational resources and delay and bandwidth of network. The task placement that minimizes estimated response time is determined. We implemented a prototype of the proposed task placement system and evaluated its performance through experiment.
AB - Mobile devices have certain restrictions such as processing power and network bandwidth. Edge computing can reduce load and delay by migrating the parts of application processing from mobile devices to edge servers. Edge computing is a method that extends cloud computing to the edge of network. Conventional schemes of face recognition based on edge computing can reduce the amount of network transmission data between the edge servers and the cloud servers by preprocessing data on the edge servers. However, the schemes execute each task on a certain resource, thus application response time may increase according to the conditions of computational resources and network. Hence, a task placement system considering the conditions of computational resources and network is necessary. This paper proposes a task placement system for face recognition applications in edge computing. The proposed system calculates estimated response time of possible task placements based on the task placement decision formula. The formula parameterizes power and CPU utilization of computational resources and delay and bandwidth of network. The task placement that minimizes estimated response time is determined. We implemented a prototype of the proposed task placement system and evaluated its performance through experiment.
KW - Distributed Processing
KW - Edge Computing
KW - Face Recognition
KW - Task Placement
UR - http://www.scopus.com/inward/record.url?scp=85085488255&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=85085488255&partnerID=8YFLogxK
U2 - 10.1109/CCNC46108.2020.9045194
DO - 10.1109/CCNC46108.2020.9045194
M3 - Conference contribution
AN - SCOPUS:85085488255
T3 - 2020 IEEE 17th Annual Consumer Communications and Networking Conference, CCNC 2020
BT - 2020 IEEE 17th Annual Consumer Communications and Networking Conference, CCNC 2020
PB - Institute of Electrical and Electronics Engineers Inc.
T2 - 17th IEEE Annual Consumer Communications and Networking Conference, CCNC 2020
Y2 - 10 January 2020 through 13 January 2020
ER -