FRESHDATA: Redesigning Networking for Information Freshness

Project Coordinator: Prof. Dr. Elif Uysal
Project Type: TUBITAK 2247-B Scientific and Technological Research Projects Funding Program
Project Duration: 24 months
Project Start Date: July 1, 2021
Funded Personnel: 2 PhD Students (Full-Time), 1 Postdoc, 1 Engineer

Project Summary: The Internet will soon be dominated by nodes using Machine-Type Communications (MTC) (e.g., industrial control, autonomous vehicles, social network apps). The key performance metric for MTC is sufficiently timely data, i.e., freshness of status updates. However, current networks have been optimized for maximizing throughput for a moderate number of high rate connections. So, how to re-architect networks to provide fresh information to an explosive number of status-update flows? This is a major challenge as classical network theory does not even have tools to address information freshness.

Age of Information (AoI) is a recently popularized metric for information freshness. The PI contributed to the original solution of controlling generalized AoI penalty under random delay or energy constraints. Our work revealed that, contrary to conventional wisdom, an age-optimal policy neither transmits at the highest possible rate nor requires the smallest end- to-end delay. We also showed how freshness-aware sampling of a stochastic process can improve accuracy of remote monitoring several-fold compared to uniform sampling. These results motivate a paradigm shift in re-engineering networks for MTC, based on AoI and more advanced freshness metrics to replace traditional performance indicators such as throughput or delay.

Propelled by these game-changing revelations, FRESHDATA will develop the theory of freshness-optimal network design through clean-slate formulations that innovate across network layers. Fundamental trade-offs between freshness, energy and reliability will be characterized. Revolutionary sampling, transmission and service policies for information flows will be devised. We also take on the challenge of demonstrating the impact of our new technologies on real-world IoT implementations. We envision the impact to reach beyond the communication networks research area, and influence the interface with control, robotics and data analytics, the main end users of MTC.

Keywords: Age of information for status updates, semantics of information, networked control systems, cross-layer wireless resource allocation, transmission scheduling, usage-limited estimation and control, finite blocklength, energy harvesting networks