Average Age-of-Information with a Backup Information Source

TL;DR

This paper studies how to optimally schedule data updates from primary and backup IoT sources to minimize average age of information, considering energy constraints and source trade-offs, using a Markov decision process approach.

Contribution

It introduces a novel MDP-based framework for AoI minimization with dual sources and energy harvesting, highlighting the benefits of backup sources in IoT systems.

Findings

Optimal policies depend on energy levels and source trade-offs.

Having a backup source improves AoI performance.

Identifies solution structures for scheduling policies.

Abstract

Data collected and transmitted by Internet of things (IoT) devices are typically used for control and monitoring purposes; and hence, their timely delivery is of utmost importance for the underlying applications. However, IoT devices operate with very limited energy sources, severely reducing their ability for timely collection and processing of status updates. IoT systems make up for these limitations by employing multiple low-power low-complexity devices that can monitor the same signal, possibly with different quality observations and different energy costs, to create diversity against the limitations of individual nodes. We investigate policies to minimize the average age of information (AoI) in a monitoring system that collects data from two sources of information denoted as primary and backup sources, respectively. We assume that each source offers a different trade-off between the AoI and the energy cost. The monitoring node is equipped with a finite size battery and harvests ambient energy. For this setup, we formulate the scheduling of status updates from the two sources as a Markov decision process (MDP), and obtain a policy that decides on the optimal action to take (i.e., which source to query or remain idle) depending on the current energy level and AoI. The performance of the obtained policy is compared with an aggressive policy for different system parameters. We identify few types of optimal solution structures and discuss the benefits of having a backup source of information in the system.