Reliable and Secure Coded Edge Computing

Edge computing is a new paradigm that relies on the strategic deployment of edge servers in order to move computational power from the cloud closer to edge devices that generate most of the data in Internet of Things (IoT) scenarios.

However, outsourcing computations to edge servers in a distributed fashion poses significant challenges in terms of reliability (because servers may fail or straggle) and security (because third parties may eavesdrop or maliciously alter the computations).

We will address both problems using tools from coding theory, where the general principle is to distribute carefully planned redundant computations across multiple edge servers.

The project will explore how this can be done efficiently with the overall goal to enable reliable and secure edge computing.

Our research plan is divided into three main tasks, where the specific objectives are to (i) develop new coding schemes that enhance reliability in multi-user edge-computing scenarios, (ii) develop lightweight schemes that allow for low-complexity, low-overhead data security, and (iii) provide a theoretical framework for the analysis of the proposed schemes in terms of their overall latency and complexity.

The project will be carried out by a PhD student and the applicant over the duration of 4-years, and it covers both well-established computational tasks (e.g., large-scale matrix-vector multiplications) and also explores new applications like distributed principal component analysis.