Data-driven sparse sensor placement based on A-optimal design of experiment with ADMM

TL;DR

This paper introduces a novel sensor placement method using A-optimal design and ADMM, outperforming existing methods in accuracy and efficiency for large-scale data-driven sensor selection tasks.

Contribution

It presents a new sensor selection approach based on A-optimal design and ADMM, improving performance over greedy and convex relaxation methods.

Findings

Better A-optimality performance than existing methods

Shorter computation time than convex relaxation in large-scale problems

Effective in data-driven sensor selection with real sea surface temperature data

Abstract

The present study proposes a sensor selection method based on the proximal splitting algorithm and the A-optimal design of experiment using the alternating direction method of multipliers (ADMM) algorithm. The performance of the proposed method was evaluated with a random sensor problem and compared with the previously proposed methods such as the greedy method and the convex relaxation. The performance of the proposed method is better than an existing method in terms of the A-optimality criterion. In addition, the proposed method requires longer computational time than the greedy method but it is quite shorter than the convex relaxation in large-scale problems. The proposed method was applied to the data-driven sparse-sensor-selection problem. A data set adopted is the NOAA OISST V2 mean sea surface temperature set. At the number of sensors larger than that of the latent state variables, the proposed method showed similar and better performances compared with previously proposed methods in terms of the A-optimality criterion and reconstruction error.