Real-time Tracking System with Partially Coupled Sources

Saeid Sadeghi Vilni, Risto Wichman · 2025

Read Real-time Tracking System with Partially Coupled Sources by Saeid Sadeghi Vilni, Risto Wichman (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-2503-21552.txt.

Introduction

Saeid Sadeghi Vilni and Risto Wichman study a tracking system where one watcher follows several partly linked sources over a faulty wireless link. The watcher, or sink, may ask just one source for an update at each moment. Because the sources are coupled, news about one source partly reveals the others. The goal is to keep tracking error low without paying for too many transmissions.

The watcher never sees the full state, so the team frames the problem as a partially observable decision process and converts it into a belief-state form. A relative value iteration algorithm then yields a request policy. In simulations, that policy beat a baseline rule on the combined cost of distortion and transmissions. The authors present the setup as a model for cheap, timely monitoring of linked sensors.

Catalog record

CREATORSaeid Sadeghi Vilni, Risto Wichman
YEAR2025
COLLECTIONPAPERS
FORMATTXT
FILE SIZE27 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2503-21552

Browse the full library · About the Tower of Babel

Full text

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

Real-time Tracking System with Partially Coupled Sources Saeid Sadeghi Vilni, Risto Wichman Abstract We consider a pull-based real-time tracking system consisting of multiple partially coupled sources and a sink. The sink monitors the sources in real-time and can request one source for an update at each time instant. The sources send updates over an unreliable wireless channel. The sources are partially coupled, and updates about one source can provide partial knowledge about other sources. We study the problem of minimizing the sum of an average distortion function and a transmission cost. Since the controller is at the sink side, the controller (sink) has only partial knowledge about the source states, and thus, we model the problem as a partially observable Markov decision process (POMDP) and then cast it as a belief-MDP problem. Using the relative value iteration algorithm, we solve the problem and propose a control policy. Simulation results show the proposed policy’s effectiveness and superiority compared to a baseline policy. Index Terms: Real-time tracking, pull-based system, coupled sources, partially observable Markov decision process. I Introduction Timely status updates are critical for real-time applications such as industrial automation, intelligent transportation, and smart grids. In these systems, a sensor observes a physical process such as industrial machine operations, vehicle movements, or power grid parameters and transmits a status update to an intended destination, such as a remote controller or a decision-making entity [ 1 ] . The effectiveness of real-time tracking depends not only on the update rate but also on the value and relevance of the received information. To capture this goal-oriented aspect, i.e., real-time tracking, distortion measure is a commonly used metric that captures the discrepancy between the sources’ states and their estimate at the destination [ 2 ] . Recent studies on real-time tracking systems often assume independent sources. However, in many practical systems, the physical processes being monitored are partially coupled due to shared environments, network constraints, or system dependencies [ 3 , 4 ] . This coupling introduces opportunities for more efficient update policies, as selecting one source for transmission can provide partial information about other sources, potentially reducing the overall estimation error and redundant transmissions. This motivates us to study a real-time tracking system with partially coupled sources. In this work, we consider a pull-based status update system where the sink selectively requests a single source to send an update. Upon receiving the command, the designated source transmits an update to the sink over an unreliable wireless channel (see Fig. 1 ). The underlying physical processes of the sources are partially coupled. More precisely, the partially coupled model is derived by combining fully coupled and independent models via a coupling factor. The considered system model may represent a status update system in a smart grid, where sensors monitor critical parameters like voltage, current, and temperature at different grid locations. These sensors are often coupled due to shared environmental factors and network inter-dependencies [ 5 ] . For instance, in power grids, voltage fluctuations at one bus can directly impact neighboring buses, leading to intrinsic coupling among voltage sensors. The data is transmitted to a central control system, which tracks the grid’s status in real-time [ 5 ] . We address a real-time tracking problem, aiming to minimize the expected time-average sum of a distortion function and a transmission cost 1 1 1 The transmission cost represents the usage of radio resources. . The solution to the problem defines a control policy that determines whether to stay idle or request an update and, if updating, selects the source to transmit. Since the state of the sources is only partially observable at the sink side, we model our problem as a partially observable Markov decision process (POMDP), and subsequently, we cast it as a belief-MDP problem. After truncating the belief-state space of the belief-MDP, we implement the relative value iteration algorithm (RVIA) to derive a control policy for solving the problem. The numerical results show the effectiveness of the proposed policy compared to a baseline policy. Works [ 3 , 4 , 6 , 7 ] investigate status update systems with correlated sources, focusing on age of information (AoI), while works [ 8 , 9 , 10 ] investigate estimation error in such systems. The work [ 8 ] derived closed-form expressions of average AoI and state estimation error in a multi-sensor system with correlated sources under an M/M/1/1 queue model. The work [ 9 ] derived upper and lower bounds on estimation error using weighted sum AoI and proposed near-optimal AoI-based scheduling for correlated sources. Unlike these works, we consider a pull-based status update system where the scheduler (sink) only has partial knowledge about the source states. In contrast to [ 8 , 9 ] , we characterize the correlation through a coupling factor, which allows for a structured analysis of its impact. The work in [ 10 ] studied a pull-based status update system with two correlated sources. Unlike our model, where the destination has only partial knowledge about the other sources, the correlation in [ 10 ] is captured through probabilistic observation. Specifically, the destination may decode the status of one source upon receiving an update from the other. Due to this fundamental difference, the policies derived in [ 10 ] are not applicable to our system. Moreover, we solve an optimization control problem to minimize estimation error and transmission cost jointly, whereas [ 8 ] characterized a closed-form expression of the average estimation error and [ 9 ] does not consider transmission cost. II System Model and Problem Formulation II-A System Model We consider a real-time tracking system consisting of K sources with corresponding transmitters and a sink, as shown in Fig. 1 . The sources are partially coupled, and each source is modeled as a binary time-varying discrete process. The sink is interested in the real-time tracking of the sources and sends a request to a single source for an update at each time instant. Each source, upon receiving a command, sends an update. The time is discrete with unit time slots {t\in\{1,2,...\}} . Figure 1: The considered system model. II-A 1 Source Model To characterize the coupling between sources, we define independent sources and (fully) coupled sources, and then define partially coupled sources [ 11 ] . Independent Source Let {X_{k}(t)\in\{0,1\}} denote the state of source k at slot t . Each source, at each slot, either remains in the same state with a probability p or transits to any other state with a probability q=1-p . Let \textbf{X}(t)\in\mathcal{X} denote a vector containing all source states at slot t , i.e., \textbf{X}(t)=[X_{1}(t),\dots,X_{K}(t)] , where \mathcal{X} is the space of all possible vectors of X 2 2 2 The space \mathcal{X} is given as \mathcal{X}=\{[j_{1},\dots,j_{K}]\mid j_{1},\dots,j_{K}\in\{0,1\}\} . Let P^{I}(\textbf{X},\textbf{X}^{\prime}) denote the state transition probability of independent sources defined as the probability of moving from source states X to source states \textbf{X}^{\prime} . Since the sources are independent P^{I}\big(\textbf{X},\textbf{X}^{\prime}) is given as \displaystyle P^{I}\big(\textbf{X},\textbf{X}^{\prime})=\Pi_{k=1}^{k=K}\mathrm{Pr}(X_{k}^{\prime}\mid X_{k}). (1) Coupled Sources In the (fully) coupled model, the values of the sources are the same and move to the same state. If they start from different states, they become the same in the next slot. Let P^{C}(\textbf{X},\textbf{X}^{\prime})

Read the complete text