M.Sc. Aleksandr Zavodovski defends his doctoral thesis Open Infrastructure for Edge Computing on Friday the 23rd of October 2020 at 10 o'clock in the University of Helsinki Exactum building, Auditorium CK112 (Pietari Kalmin katu 5, basement). His opponent is Associate Professor Mario di Francesco (Aalto University, Finland) and custos Professor Jussi Kangasharju (University of Helsinki). The defence will be held in English. It is possibe to follow the defence as a live stream at https://helsinki.zoom.us/j/7079805848?pwd=QzZSS3dGeGVyZUJRQzZ0NnNhbExvQT09.
The thesis of Aleksandr Zavodovski is a part of research done in the Department of Computer Science and in the Collaborative Networking research group at the University of Helsinki. His supervisor has been Professor Jussi Kangasharju (University of Helsinki).
Open Infrastructure for Edge Computing
Edge computing, bringing the computation closer to end-users and data producers, has now firmly gained the status of enabling technology for the new kinds of emerging applications, such as Virtual/Augmented Reality and IoT. The motivation backing this rapidly developing computing paradigm is mainly two-fold. On the one hand, the goal is to minimize the latency that end-users experience, not only improving the quality of service but empowering new kinds of applications, which would not even be possible given higher delays. On the other, edge computing aims to save core networking bandwidth from being overwhelmed by myriads of IoT devices, sending their data to the cloud. After analyzing and aggregating IoT streams at edge servers, much less networking capacity will be required to persist remaining information in distant cloud datacenters.
Having a solid motivation and experiencing continuous interest from both academia and industry, edge computing is still in its nascency. To leave adolescence and take its place on a par with the cloud computing paradigm, finally forming a versatile edge-cloud environment, the newcomer needs to overcome a number of challenges. First of all, the computing infrastructure to deploy edge applications and services is very limited at the moment. Indeed, there are initiatives supported by the telecommunication industry, like Multi-access Edge Computing. Also, cloud providers plan to establish their facilities near the edge of the network. However, we believe that even more efforts will be required to make edge servers generally available. Second, to emerge and function efficiently, the ecosystem of edge computing needs practices, standards, and governance mechanisms of its own kind. The specificity originates from the highly dispersed nature of the edge, implying high heterogeneity of resources and diverse administrative control over the computing facilities. Finally, the third challenge is the dynamicity of the edge computing environment due to, e.g., varying demand, migrating clients, etc.
In this thesis, we outline underlying principles of what we call Open Infrastructure for Edge (OpenIE), identify its key features, and provide solutions for them. Intended to tackle the challenges we mentioned above, OpenIE defines a set of common practices and loosely coupled technologies creating a unified environment out of highly heterogeneous and administratively partitioned edge computing resources. Particularly, we design a protocol capable of discovering edge providers on a global scale. Further, we propose a framework of Ingelligent Containers (ICONs), capable of autonomous decision making and forming a service overlay on a large-scale edge-cloud setting. As edge providers need to be economically incentivized, we devise a truthful double auction mechanism where edge providers can meet application owners or administrators in need of deploying an edge service. Due to truthfulness, in our auction, it is the best strategy for all participants to bid one's privately known valuation (or cost), thus making complex market behavior strategies obsolete. We analyze the potential of distributed ledgers to serve for OpenIE decentralized agreement and transaction handling and show how our auction can be implemented with the help of distributed ledgers. With the key building blocks of OpenIE, mentioned above, we hope to make an entrance for anyone interested in service provisioning at the edge as easy as possible. We hope that with the emergence of independent edge providers, edge computing will finally become pervasive.
Availability of the dissertation
An electronic version of the doctoral dissertation is available on the e-thesis site of the University of Helsinki at http://urn.fi/URN:ISBN:978-951-51-6651-7.
Printed copies will be available on request from Aleksandr Zavodovski: aleksandr.zavodovski@helsinki.fi.