Abstract:This paper presents communication-constrained distributed conformal risk control (CD-CRC) framework, a novel decision-making framework for sensor networks under communication constraints. Targeting multi-label classification problems, such as segmentation, CD-CRC dynamically adjusts local and global thresholds used to identify significant labels with the goal of ensuring a target false negative rate (FNR), while adhering to communication capacity limits. CD-CRC builds on online exponentiated gradient descent to estimate the relative quality of the observations of different sensors, and on online conformal risk control (CRC) as a mechanism to control local and global thresholds. CD-CRC is proved to offer deterministic worst-case performance guarantees in terms of FNR and communication overhead, while the regret performance in terms of false positive rate (FPR) is characterized as a function of the key hyperparameters. Simulation results highlight the effectiveness of CD-CRC, particularly in communication resource-constrained environments, making it a valuable tool for enhancing the performance and reliability of distributed sensor networks.

What problem does this paper attempt to address?

The problem that this paper attempts to solve is how to perform multi - label classification tasks (such as segmentation) in a distributed sensor network under the condition of limited communication resources, while ensuring that the specific false negative rate (FNR) requirements are met and minimizing the false positive rate (FPR). Specifically, the paper proposes a communication - constrained distributed consensus risk control (CD - CRC) framework, aiming at: 1. **Dynamically adjust local and global thresholds**: In order to identify important labels, ensure that the target false negative rate (FNR) is reached, while complying with the communication capacity limit. 2. **Provide deterministic worst - case performance guarantees**: Provide deterministic worst - case performance guarantees in terms of FNR and communication overhead, while analyzing the performance of FPR as a function of key hyper - parameters. ### Problem Background In a distributed sensor network, multiple sensors observe different features of the environment and transmit this information to a central server for decision - making. For example, in a binary segmentation task, each sensor generates a local binary mask based on partial and noisy observations, and the central server combines these local predictions to generate a global binary mask to estimate the spatial coverage of the object of interest. ### Limitations of Existing Methods Existing sensor fusion methods usually rely on statistical models, such as weighted averaging, Bayesian inference, or Kalman filtering, etc. These methods assume statistical models of source signals and sensor observations when optimizing the overall prediction performance. However, these methods cannot provide formal performance guarantees without making strong modeling assumptions. ### Innovations of the Paper To solve the above problems, the paper introduces the CD - CRC framework, which innovates in the following aspects: - **Consider communication constraints**: Different from the existing distributed consensus prediction (D - CP), CD - CRC considers the communication constraints between sensors and the central server, and meets the constraints of FNR and communication capacity by dynamically adjusting multiple local thresholds and global thresholds. - **Online optimization strategy**: Use the online exponential gradient descent algorithm to estimate the relative quality of different sensor observations, and use the online consensus risk control mechanism to adjust local and global thresholds. - **Performance guarantee**: Provide deterministic, assumption - free performance guarantees regarding FNR and communication load, and establish an upper bound for FPR. ### Model and Problem Definition The paper describes in detail the model and problem definition of the system, including: - **System setup**: Consists of K sensors and a central server, and the goal is to perform distributed multi - label prediction under reliability and communication constraints. - **Design problem**: Update local thresholds, combination weights, and global thresholds through an online optimization algorithm to minimize FPR while meeting the constraints of FNR and communication overhead. ### Main Contributions The main contributions of the paper include: - Propose the distributed consensus risk control (D - CRC) protocol applicable to scenarios without communication constraints, and provide theoretical analysis. - Introduce the CD - CRC protocol, which can dynamically update local and global thresholds and combination weights under communication constraints to meet the requirements of FNR and communication capacity while minimizing FPR. - Provide performance analysis of CD - CRC, ensure deterministic, assumption - free performance guarantees of FNR and communication load, and establish an upper bound for FPR. - Demonstrate the advantages of CD - CRC in a communication - resource - limited environment through simulation results. In summary, this paper aims to solve the multi - label classification tasks in a distributed sensor network under the condition of limited communication resources by introducing the CD - CRC framework, ensure long - term reliability and provide performance guarantees.

Conformal Distributed Remote Inference in Sensor Networks Under Reliability and Communication Constraints

A Reliable Collaborative Spectrum Sensing Scheme Based on the ROCQ Reputation Model for Cognitive Radio Networks

Reliable Sub-Nyquist Spectrum Sensing via Conformal Risk Control

Distributed Censored Regression over Networks

Distributed Censored Regression Over Networks.

Cross-Validation Conformal Risk Control

Dynamic Channel Assignment for Wireless Sensor Networks: A Regret Matching Based Approach

Using random forest for reliable classification and cost-sensitive learning for medical diagnosis

Stability Analysis of Networked Control Systems with Round-Robin Scheduling and Packet Dropouts

Sub-Spectrum Sensing Based On Reputation In Crns

Communication-Efficient Secure Distributed Estimation With Noisy Measurement Against FDI Attack

Covert and Reliable Semantic Communication Against Cross-Layer Privacy Inference over Wireless Edge Networks

Federated Inference with Reliable Uncertainty Quantification over Wireless Channels via Conformal Prediction

Decentralized RLS with Data-Adaptive Censoring for Regressions over Large-Scale Networks

Decentralized Detection in Ad Hoc Sensor Networks with Low Data Rate Inter Sensor Communication.

Conformal Risk Control for Ordinal Classification

Scalable consensus filtering for uncertain systems over sensor networks with Round‐Robin protocol

Selecting informative conformal prediction sets with false coverage rate control

Non-Exchangeable Conformal Risk Control

Robust Conformal Prediction under Distribution Shift via Physics-Informed Structural Causal Model