Machine learning and data analytics are popular and powerful data processing methods to extract knowledge from raw sensor data in IoT. With the fast growing sensor data, data compression near the data source is a key method to minimize energy, transmission and storage cost. However, to perform machine learning and data analytics, the data is often required to be decompressed first, which often causes long delay and is even not practical for the emerging Edge computing based IoT applications. As many IoT applications cannot simply reply on central cloud-based computing infrastructure, due to time criticality, constrained long-distance connectivity and others, Edge computing is a promising paradigm to deploy the storage and computing resource close the end devices, i.e., extend the computing to the edge of the network. One of the challenges is that edge servers do not have large memory and storage as central cloud computing, it calls for efficient methods to allow distributed machine learning and data analytics directly in the compressed data domain, i.e., avoiding data decompression. In the literature, compressed learning have been applied for high-resolution images, video processing and text analysis.
This PhD project will focus on the design and development of compressed learning and analytics by leveraging sensor data compression techniques, e.g., compressed sensing, for IoT. The project will study the relation between compression ratio, processing time, and accuracy of the outcomes, as well as the cost in terms of energy, transmission and storage. In particular, the applicant will bridge the fundamental theory with practice in implementation. The primary goal is to develop novel compressed learning and analytics techniques in edge computing for time-critical and resource constrained IoT applications. The applicant is expected to deploy such designs in practical systems to demonstrate the results.
Our group operates at the intersection of theory and implementation of communication and computation, distributed systems, and Cloud/Edge technologies. For this reason, we encourage applicants with a strong theoretical profile or a strong implementation/programming profile (or ideally both) to apply. Our group also counts with a new and well-funded laboratory to deploy large storage servers pods, computing servers, ARM micro-servers for energy efficient edge operations, Jetson TX2 board, and a number of end-user devices.
This project is expected to result in high impact publications in journals, conference participation, and patent filings. Our goal is to design and develop compressed learning and analytics at edge cloud for emerging IoT applications.
Qualifications and specific competences:
We are looking for highly motivated and independent students willing to take the challenge to do a successful 3-year PhD programme in Aarhus University. The ideal candidate will have the following profile (but not all items are required for a successful application):
• Relevant Master’s degree (e.g., Computer Engineering, Software Engineering, Electrical Engineering), although exceptional candidates from related disciplines (e.g., Applied Mathematics) will also be considered.
• Excellent undergraduate and master degree grades are required.
• Background on signal processing, machine learning, data analytics, and computing and storage system is highly desired, but candidates from other disciplines will be considered based on their merits and potential.
• Background on linear algebra, mathematics and probability is desired.
• Background on programming, particularly python, C++, and/or Java, or experience with embedded system is desired.
• Good English verbal and written skills are required.
Place of Employment and Place of Work:
The place of employment is Aarhus University, and the place of work is the Department of Engineering, Turing Building (5341), Åbogade 34, 8200 Aarhus N, Denmark.
Aarhus University was founded in 1928. It has 40,000 students; about 1,800 PhD students – of which one in four has a foreign nationality – and close to 900 postdoctoral scholars together with 11,500 employees. Aarhus University has been establishing itself as a university for cutting-edge research, and has been regularly included in the top-100 Universities of the most important university ranking lists.
Applicants seeking further information are invited to contact:
Assoc. Prof. Qi Zhang (email: [email protected]
Assoc. Prof. Daniel Enrique Lucani Rötter (email [email protected]
The detailed application procedures refers to the announcement at Aarhus University webpage: http://phd.scitech.au.dk/for-applicants/apply-here/november-2019/compressed-learning-and-analytics-at-edge-cloud/