Minimizing Age of Information Under General Models for IoT Data Collection

Abstract

A critical component in IoT infrastructure and applications is data collection at network edge. Recently, a new metric, called age of information (AoI), has become popular to quantify the freshness of information collected at network edge. AoI research is still in infancy and most efforts assume overly simplified models in their investigation, making their results far from useful when addressing practical problems in IoT applications. In this article, we close this gap by considering more general models for AoI research that are more relevant in the real world. Specifically, we consider general and heterogeneous sampling behaviors among source nodes, varying sample size, and a transmission model with multiple transmission units in each time slot. Based on these generalizations, we develop new theoretical results (in terms of fundamental properties and performance bounds) and a new near-optimal low-complexity scheduling algorithm to minimize AoI. Our results make a major advance to existing AoI research and help bridge the gap between theory and practice.