Off-campus WSU users: To download campus access dissertations, please use the following link to log into our proxy server with your WSU access ID and password, then click the "Off-campus Download" button below.

Non-WSU users: Please talk to your librarian about requesting this dissertation through interlibrary loan.

Access Type

WSU Access

Date of Award

January 2018

Degree Type

Dissertation

Degree Name

Ph.D.

Department

Computer Science

First Advisor

Weisong Shi

Abstract

Stream processing is a critical technique to process huge amount of data in real-time manner.

Cloud computing has been used for stream processing due to its unlimited computation

resources. At the same time, we are entering the era of Internet of Everything (IoE). The emerging

edge computing benefits low-latency applications by leveraging computation resources at

the proximity of data sources. Billions of sensors and actuators are being deployed worldwide

and huge amount of data generated by things are immersed in our daily life. It has become

essential for organizations to be able to stream and analyze data, and provide low-latency analytics

on streaming data. However, cloud computing is inefficient to process all data in a centralized

environment in terms of the network bandwidth cost and response latency. Although

edge computing offloads computation from the cloud to the edge of the Internet, there is not

a data sharing and processing framework that efficiently utilizes computation resources in the

cloud and the edge. Furthermore, the heterogeneity of edge devices brings more difficulty to the development of collaborative cloud-edge applications.

To explore and attack the challenges of stream processing system in collaborative cloudedge

environment, in this dissertation we design and develop a series of systems to support

stream processing applications in hybrid cloud-edge analytics. Specifically, we develop an

hierarchical and hybrid outlier detection model for multivariate time series streams that automatically

selects the best model for different time series. We optimize one of the stream

processing system (i.e., Spark Streaming) to reduce the end-to-end latency. To facilitate the

development of collaborative cloud-edge applications, we propose and implement a new computing

framework, Firework that allows stakeholders to share and process data by leveraging

both the cloud and the edge. A vision-based cloud-edge application is implemented to demonstrate

the capabilities of Firework. By combining all these studies, we provide comprehensive

system support for stream processing in collaborative cloud-edge environment.

Off-campus Download

Share

COinS