Rocket assembly line testbed as a service (TaaS): a comparison of data acquisition strategies
M.R. McCormick, Fadi El Kalach, Mojtaba A. Farahani, Ramy Harik, Thorsten Wuest · Manufacturing Letters · 2025
Manufacturing data scarcity poses a significant challenge for the development of emerging technologies including Artificial Intelligence (AI), Machine Learning (ML), and digital twins, as well as development of the technical expertise necessary to wield them. When data is available, subtle biases in data acquisition strategies may covertly impact the performance of downstream emerging technologies. To illuminate these biases, their potential impact on analytics, address data scarcity, and improve technical expertise, this study presents three new manufacturing datasets totaling 38 h of runtime in Testbed as a Service (TaaS) format and performs a detailed comparison of the data acquisition strategies used to acquire them including a quantitative evaluation of potential impact on downstream analytics. As a key finding, this study posits that an evaluated strategy offers superior capability for generating actionable insights from high frequency phenomena at the cost of requiring sensor fusion and clock synchronization, driven by tradeoffs in software design and corresponding network latency. As a result of technical analysis and practical guidance, this study equips industry practitioners and academic researchers with insights that improve the selection or implementation of an appropriate data acquisition strategy that ensures performant downstream analytics to support their use case.