Edge-Cloud Computing for Internet of Things Data Analytics

连接设备的数量正以前所未有的速度增长，导致物联网（IoT）数据的爆炸式增长，这些数据正在设备、存储和处理位置之间移动。来自传感器的数据通常被传输到云端进行存储和分析，然后交付给各种软件应用程序。例如，来自智能家居的数据被传输到数千英里外的数据中心，然后再传输回来，在家中的设备上显示。由于数据中心流量已经以zb为单位进行测量，试图将所有物联网数据传输到云端将使通信网络紧张并增加成本。此外，在安全关键应用程序中常见的实时系统不能容忍传输到云所造成的延迟。

The number of connected devices is growing at an unprecedented rate, creating an explosion of Internet of Things (IoT) data that are moving among devices, storage, and processing locations. Data from sensors are typically transferred to the cloud for storage and analysis and then delivered to various software applications. For example, data from a smart home are transferred to a data centre possibly thousands of miles away and then back again for display on devices within the home. With data centre traffic already measured in zettabytes, attempting to transfer all IoT data to the cloud will strain communication networks and increase cost. Moreover, real-time systems, which are common in safety critical applications, cannot tolerate latencies caused by transfer to the cloud. To address these IoT challenges, the proposed research program will combine edge and cloud computing. Whereas cloud computing transfers data to the computing location, edge computing brings computation closer to the network edge where data are generated. Here, the term edge encompasses measuring devices (e.g., sensors, smart phones) and fog nodes, the resources close to data sources (e.g., gateways, local servers). Edge-cloud approaches will decrease network traffic and latencies, improve user experience, enable off-line operation, and reduce exposure to cloud-related security risks. The long term objective of this research program is to advance the field of IoT data analytics by extending it to edge computing. The short-term objectives include: 1. Enabling Edge-Cloud IoT Data Analytics for Smart Cities will initiate work on edge-cloud analytics by adopting edge computing for specific smart city use cases such as energy management and sustainable buildings. 2. Designing Resource Management Models to allocate and manage computation across diverse nodes of the edge-cloud environment to minimize response time and/or network traffic. 3. Enabling IoT Interoperability to integrate data from diverse IoT devices and non-IoT data in support of novel data analytics methods and new software applications through edge computing. 4. Designing IoT Edge-Cloud Data Management Methods that will distribute data storage over edge and cloud nodes to reduce network traffic and response time by storing or replicating data close to where they are produced (edge) or consumed (edge or cloud). The proposed research will enable new IoT applications and services that are currently not possible due to latencies or costs associated with data transfer to the cloud. Software applications will benefit through improved quality of service, ability to offload computation to nearby devices, and increased location awareness. Companies using IoT analytics will benefit through process optimization, insights discovery, and improved decision-making. This research program will contribute to Canadian leadership in this rapidly growing field.