Edge Optimization: Operating Systems & Software on the Edge
- Reference number
- FFL21-0091
- Project leader
- Ali-Eldin Hassan, Ahmed
- Start and end dates
- 220801-271231
- Amount granted
- 14 300 000 SEK
- Administrative organization
- Chalmers University of Technology
- Research area
- Information, Communication and Systems Technology
Summary
We aim to take a clean-slate approach to designing distributed edge cloud systems to enable latency-critical real-time applications to utilize edge resources for offloading computations and processing while maintaining strict latency guarantees. We take a holistic approach spanning the three layers in the system; 1) starting from the edge cloud architecture, how resources should be provisioned, where they should be provisioned, and what type of resources should be there on each edge; 2) In parallel, we will study how to design, and manage self-adaptive edge application software to make use of edge capabilities, including what programming model should developer use, tools for debloating, deflating, and customization can be used to make applications more robust, and lean; 3)To tie these two layers---the architecture and hardware resources, and the software developed, we will build an edge operating system capable of managing the distributed resources and applications, while guaranteeing the performance of the applications. Towards this end, the PI with a team of four newly recruited PhD students will solve the following four research questions: 1- What is the right architecture to build a distributed edge cloud? 2- How to design adaptive edge applications? 3-How to manage the workload dynamics of real-time edge cloud applications? 4- How to manage and process massive data of edge applications in real-time?
Popular science description
Over the past decade, cloud computing has emerged as a popular paradigm for running a variety of distributed systems applications ranging from web applications to Al and high performance computing workloads. In recent years, edge computing has emerged as a complement to cloud computing, particularly for running latency-sensitive workloads. Edge computing involves using computational and storage resources deployed at the edge of the network to run applications that can benefit from low network latency. Emerging edge applications include mobile augmented reality, Internet of Things (IoT) analytics, autonomous vehicles, and edge Al using machine learning models. While a relatively mature idea, first time proposed in the early 2000s, edge computing has not provided the digital revolution it promised. Applications still lack behind what was expected. For example, Elon Musk the CEO of Tesla repetitively claimed that they are close to level-5 autonomous vehicles. Many other applications that were thought to be within a few years of reach have proven elusive so far. This is simply due to many of the technical challenges in the management of edge computing infrastructures, their architecture, and challenges related to the software architecture that will make use of these systems. In this proposal, we aim to take a clean-slate approach to edge cloud system and software design, starting from the basic architecture of the edge, to the operating systems that manages its resources, all the way to the application that will run on top. The project has the potential to enable many key applications for tomorrow's society, making our roads safer, our healthcare better, and our factories more productive. The project is done in collaboration with Swedish industry and international world leading researchers, hiring three PhD new students to work closely with the PI of the project.