Minimizing Age of Detection for a Markov Source over a Lossy Channel

Shivang Garde 1 , Jaya Prakash Champati 2 , Arpan Chattopadhyay 1 · 2025

Read Minimizing Age of Detection for a Markov Source over a Lossy Channel by Shivang Garde 1 , Jaya Prakash Champati 2 , Arpan Chattopadhyay 1 (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2503-02285.txt.

Introduction

Shivang Garde, Jaya Prakash Champati, and Arpan Chattopadhyay wrote a 2025 paper on how fast a monitor can detect changes in a system it watches from afar. A sensor samples a process whose state jumps by known rules, and sends updates over a lossy wireless link. The monitor wants to notice each jump quickly, but sampling too often wastes energy.

They define a freshness measure, the age of detection, for the time between a real change and its discovery. The monitor asks for samples on demand. The team frame the choice as a decision problem with a cap on average sampling, and solve it with standard algorithms. Their numbers map the trade-off among detection speed, sampling rate, and link reliability.

Catalog record

CREATORShivang Garde 1 , Jaya Prakash Champati 2 , Arpan Chattopadhyay 1
YEAR2025
COLLECTIONPAPERS
FORMATTXT
FILE SIZE25 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2503-02285

Browse the full library · About the Tower of Babel

Full text

The opening of the work, from the library's hosted text:

Minimizing Age of Detection for a Markov Source over a Lossy Channel Shivang Garde 1 , Jaya Prakash Champati 2 , Arpan Chattopadhyay 1 Abstract Monitoring a process/phenomenon of specific interest is prevalent in Cyber-Physical Systems (CPS), remote healthcare, smart buildings, intelligent transport, industry 4.0, etc. A key building block of the monitoring system is a sensor sampling the process and communicating the status updates to a monitor for detecting events of interest. Measuring the freshness of the status updates is essential for the timely detection of events, and it has received significant research interest in recent times. In this paper, we propose a new freshness metric, Age of Detection (AoD), for monitoring the state transitions of a Discrete Time Markov Chain (DTMC) source over a lossy wireless channel. We consider the pull model where the sensor samples DTMC state whenever the monitor requests a status update. We formulate a Constrained Markov Decision Problem (CMDP) for optimizing the AoD subject to a constraint on the average sampling frequency and solve it using the Lagrangian MDP formulation and Relative Value Iteration (RVI) algorithm. Our numerical results show interesting trade-offs between AoD, sampling frequency, and transmission success probability. Further, the AoD minimizing policy provides a lower estimation error than the Age of Information (AoI) minimizing policy, thus demonstrating the utility of AoD for monitoring DTMC sources. I Introduction Remote monitoring of an information source or a process of interest for detecting the occurrence of an event, e.g., the process exceeding a threshold, is crucial for a wide range of systems that include unmanned aircraft, self-driving cars, smart buildings, and Industry 4.0. For timely detection of the events, the monitor requires fresh status updates which in turn requires a sensor to sample the process as frequently as possible. However, sampling at such a high frequency could result in high energy consumption and a large bandwidth requirement for transmitting the samples to the monitor. On the other hand, sampling at a lower frequency causes staleness in event detection. Also, in practice, the communication channel is not ideal, i.e., the samples transmitted through the channel may not always be successfully received by the receiver. In this context, we address the following question: for a discrete-time remote monitoring system with a lossy communication channel, at every time slot, given the previously sampled state of the source, whether to sample or not to optimize both frequency of sampling and staleness. We address the above question for sources modeled by finite-state DTMC, which received considerable attention recently [ 1 , 2 , 3 , 4 , 5 ] . The events we seek to identify are the new state transitions in the DTMC. The motivation for studying the DTMC source is that it can serve as an abstract model to describe the states of a plant or a process in industries. For example, we may classify the states of a plant into “good” and “bad” and study a 2-state DTMC by deriving the transition probabilities from the history of observations made on the plant. In this paper, we consider that the DTMC states are unknown to the sensor between sampling instants. This is relevant in monitoring applications such as machine fault detection [ 6 ] , where an IoT sensor wakes up periodically and samples the machine state (vibrations, rotation speed, etc.) and does not know the state of the machine in between the sampling instants. Quantifying the staleness of status updates using the Age of Information (AoI) metric was extensively studied for a wide variety of queuing and wireless networking systems in the past decade; see for example [ 7 , 8 , 9 , 10 , 11 , 12 , 13 , 14 , 15 , 16 , 17 ] . AoI is defined as the time elapsed since the generation of the freshest update available at the destination. AoI and any cost function of AoI, e.g., see [ 18 ] , only capture the timing of the information update and do not account for the semantics or the dynamics of the information source. Thus, there has been significant research interest in studying new freshness metrics that account for variations or states of the source. Some early studies proposed freshness metrics based on mutual information in [ 19 ] and conditional entropy in [ 20 ] . More recently, an entropy-based freshness metric was also studied in [ 21 ] , where the authors consider the problem of selecting a subset of Markov sources for transmitting over unreliable shared channels. The problem is solved using the restless MAB framework. Unlike their metric, we define a new freshness metric, Age of Detection (AoD), as a per-slot cost in an MDP and solve a CMDP to minimize AoD subject to a sampling frequency constraint. Another semantics-based freshness metric Age of Incorrect Information (AoII) proposed in [ 4 ] received significant research attention. It is defined as a product of two functions: one is an increasing function of time until the next update is received, and the other is a general penalty function for mismatch in the states at the source and the receiver. Different instantiations of AoII were studied in [ 4 , 1 , 2 , 3 ] for estimating the states of a DTMC source. However, these works consider a parameterized symmetric DTMC for analytical tractability, where the transition probability matrix has the following special structure. For any state, the self-transition probability is p_{t} , and the transition probability to any other state is p_{R} . Consequently, these values have to satisfy p_{R}+(N-1)p_{t}=1 , where N is the number of states. In contrast to the above works, in our previous work [ 5 ] , we studied the N -state ergodic DTMC without further assumptions on the structure of the state transition probability matrix and considered that the sensor only knows the state of the DTMC when it samples. We proposed the age penalty metric, defined as the time elapsed since the first DTMC state change occurred since the last sampling instant. We analyzed the age penalty metric for the scenario where the communication between the sensor and the monitor is either negligible or takes exactly one slot. The age penalty metric was independently studied under Age of Outdated Information (Ao 2 I) in [ 22 ] . However, [ 5 ] and [ 22 ] assumed a deterministic communication delay of one slot for any status update transmission. In this paper, we relax this assumption by considering a lossy wireless channel. The random delays induced by the losses cannot be accounted for in the age penalty metric, which motivates the AoD definition. In a slot, if the receiver sends no request for a new update or if there is a transmission failure, AoD induces a cost that is a function of the time elapsed since the generation time of the freshest update at the receiver and the time elapsed since the latest request sent. A salient feature of AoD is that it is tractable, and we show that it reduces to the age penalty metric under the assumption that the communication channel has no losses. Our main contributions are summarized below: • We propose AoD

Read the complete text