A Whittle Index Approach to Minimizing Age of Multi-Packet Information in IoT Network

Mianlong Chen, Kui Wu, Linqi Song

Introduction

In the Internet of Things (IoT), the utility of data is often tied to its "freshness." The Age of Information (AoI) metric has emerged as a standard for quantifying this freshness, measuring the time elapsed since the most recent successful update was generated. While much of the existing literature assumes that a status update consists of a single packet, real-world applications—such as high-resolution sensor telemetry or video frames from autonomous systems—often require multiple packets to constitute a single meaningful update. This paper addresses the scheduling challenges inherent in these multi-packet environments.

The Multi-Packet Scheduling Challenge

When updates are multi-packet, the destination's information is only refreshed once the entire set of packets is received. This introduces significant complexity into network scheduling. The base station must track the progress of partially received updates across numerous sources to decide which packet to transmit next. Traditional "greedy" algorithms, which prioritize the source with the highest current age, often fail because they do not account for the "work-in-progress" required to complete a multi-packet transmission.

System architecture for multi-packet status updates.

Methodology: The MWIP Approach

The authors frame this scheduling problem as a Restless Multi-Armed Bandit (RMAB) optimization. To make the problem tractable, they propose the Multi-packet Whittle Index Policy (MWIP).

Optimality and Non-Preemption

A key theoretical contribution is Lemma 1, which establishes that in an optimal schedule, once the base station begins transmitting packets for a specific update, it should continue until that update is complete. Interrupting a source mid-update (preemption) generally increases the total system age. This "don't stop" rule significantly simplifies the search space for an optimal solution.

Whittle Index Derivation

By decoupling the $N$-source problem into $N$ independent sub-problems, the authors derive a "Whittle Index." This index represents the marginal reward for serving a specific source based on an approximation of the AoI. The authors prove that the problem is "indexable," a necessary condition for the Whittle Index approach to be valid.

Evolution of Age of Information over time.

Handling Lossy Networks

In environments with high packet loss, the authors incorporate Erasure Coding. By encoding $r$ original packets into $n$ packets, the receiver only needs to successfully receive any $r$ out of $n$ packets to refresh the information. This reduces the overhead of individual packet acknowledgments (ACKs) and improves overall freshness in lossy environments.

Key Results and Performance

The proposed MWIP was evaluated through extensive numerical simulations, comparing it against brute-force optimal solutions and standard baselines like the Greedy Policy (GP).

  • Near-Optimal Performance: MWIP achieved an average AoI very close to the theoretical minimum in small-scale scenarios where a brute-force "Opt" solution could be calculated.
  • Computational Efficiency: While a brute-force optimal search took over 314,572 ms for a small network, MWIP computed the schedule in just 0.3 ms, demonstrating its suitability for real-time IoT applications.
  • Robustness: MWIP consistently outperformed both GP and a Naïve Whittle Index Policy (NWIP), which fails to account for the multi-packet nature of updates.
  • Coding Gains: The use of Erasure Codes (EC Trans) significantly lowered the average AoI in lossy network conditions compared to standard transmission protocols with individual ACKs.

Conclusion and Future Directions

The MWIP provides a high-performance, low-complexity solution for managing data freshness in multi-packet IoT networks. By leveraging the RMAB framework and the Whittle Index, the authors have developed a policy that is both theoretically grounded and practically efficient. Future research may explore the value of "partial updates," where receiving a subset of packets provides some utility, and address fairness concerns to prevent the starvation of sources with large packet sizes or poor channel conditions.

Summary updated

Abstract

Age of information (AoI) captures the freshness of information and has been used broadly as an important performance metric in big data analytics in the Internet of Things (IoT). We consider a general scenario where a meaningful piece of information consists of multiple packets and the information is not complete until all related packets have been correctly received. Minimizing AoI in this general scenario is challenging in both scheduling algorithm design and theoretical analysis, because we need to track the history of received packets before a complete piece of information can be updated. We first analyse the necessary condition for optimal scheduling, based on which we present an optimal scheduling method. The optimal solution, however, has high time complexity. To address the problem, we investigate the problem with a special type of learning, i.e., learning in restless multi-armed bandits (RMAB), and propose a Whittle index-based scheduling method. We also propose a new transmission strategy based on erasure codes to improve the performance of scheduling policies in lossy networks. Performance evaluation results demonstrate that our solution outperforms other baseline policies such as greedy policy and naïve Whittle index policy in both lossless and lossy networks.
BibTeX
@article{chen2021whittle,
  author = {Mianlong Chen and Kui Wu and Linqi Song},
  title = {{A Whittle Index Approach to Minimizing Age of Multi-Packet Information in IoT Network}},
  journal = {IEEE Access},
  year = {2021},
  doi = {10.1109/access.2021.3059966},
}
Figure