행사개요: "EdgeCPS 환경에서의 데이터 오프로딩(data offloading) 최신 기술 동향 및 연구 성과 소개"

- 강연자: DGIST 정보통신융합전공 김백규 교수

- 강연 내용: 지능형 엣지 사이버 물리 환경의 거대 사물이 수집한 데이터의 효율적인 오프로딩을 위한 최신 연구 동향 소개

- 일시: 2021년 7월 15일 (목) 오전 11시

- 장소: 경북대학교 IT융복합관 324호

- 초청: 컴퓨터학부 서영균 교수

- 주최: BK21 지능융합 소프트웨어 교육연구단

- 주관: ETRI, 경북대학교 초연결융합기술연구소

Title: Research Trend of Computation and Data Offloading in Automotive Edge Computing

Abstract

Modern IoT (Internet of Things) systems perform their missions often sending, processing or receiving

big data. While a number of Quality of Service (QoS) criteria, such as timeliness or service availability,

need to be guaranteed, it is challenging since compute resources and data sources are not always

physically co-located due to IoT hardware limitations or security/privacy concerns.

We present a research trend of the edge computing, and its application on the connected car

(automotive) domain. An edge device (e.g., a car) may offload data or computation to remote servers,

either cloud edge or enterprise cloud (e.g., Amazon Web Service or Microsoft Azure), in order to

overcome inherent hardware limitations such as limited storage or computation powers or energy

consumption.

However, guaranteeing QoS requirements (e.g., timeliness) is non-trivial due to the inherent

architectural complexity of the edge computing. We introduce three fundamental research problems,

(1) workload characterization, (2) resource management, (3) data/computation offloading scheduling

that are related to guaranteeing QoS requirements in the edge computing. We discuss how these

approaches may be similarly applicable beyond the automotive domain.