Minimizing the Age of Incorrect Information for Real-time Tracking of Markov Remote Sources

TL;DR

This paper introduces a new scheduling policy based on Whittle's index to minimize the Mean Age of Incorrect Information in real-time monitoring of Markov sources, addressing the exploration-exploitation challenge.

Contribution

It models the AoII minimization as a POMDP and develops a novel Whittle's index-based scheduling scheme using belief states, a first in AoI research.

Findings

The proposed policy outperforms classical AoI-based scheduling.

The problem is proven to be indexable with a threshold structure.

Numerical results demonstrate improved information freshness.

Abstract

The age of Incorrect Information (AoII) has been introduced recently to address the shortcomings of the standard Age of information metric (AoI) in real-time monitoring applications. In this paper, we consider the problem of monitoring the states of remote sources that evolve according to a Markovian Process. A central scheduler selects at each time slot which sources should send their updates in such a way to minimize the Mean Age of Incorrect Information (MAoII). The difficulty of the problem lies in the fact that the scheduler cannot know the states of the sources before receiving the updates and it has then to optimally balance the exploitation-exploration trade-off. We show that the problem can be modeled as a partially Observable Markov Decision Process Problem framework. We develop a new scheduling scheme based on Whittle's index policy. The scheduling decision is made by updating a belief value of the states of the sources, which is to the best of our knowledge has not been considered before in the Age of Information area. To that extent, we proceed by using the Lagrangian Relaxation Approach, and prove that the dual problem has an optimal threshold policy. Building on that, we shown that the problem is indexable and compute the expressions of the Whittle's indices. Finally, we provide some numerical results to highlight the performance of our derived policy compared to the classical AoI metric.