Edge Computing for Smart Manufacturing: A Case Study at Hyundai Motor Company

Thesis Abstract

This study investigates the integration of edge computing within Hyundai Motor Company's smart manufacturing ecosystem to address latency, bandwidth, and data sovereignty challenges in production and supply chain operations. The research addresses the problem of timely decision-making and operational resilience in highly automated manufacturing lines, where centralized cloud processing introduces latency that can hinder real-time control, predictive maintenance, and dynamic scheduling. The aim is to evaluate how an edge-centric architecture can enhance real-time analytics, reduce network dependency, and improve overall equipment effectiveness (OEE) in a mass-production context. Specific objectives include (1) identifying critical bottlenecks in data flows from shop-floor sensors to edge nodes and central cloud, (2) designing an edge deployment blueprint tailored to Hyundai’s assembly and powertrain facilities, (3) evaluating the impact of edge computing on latency, throughput, and OEE compared with cloud-centric baselines, (4) examining data governance, security, and reliability implications of edge integration, and (5) deriving a scalable governance and orchestration model for heterogeneous edge devices. A mixed-methods research design combines quantitative experiments and qualitative insights. The population comprises edge-enabled production cells and associated IT/OT staff at Hyundai’s Ulsan and Georgia facilities. A purposive sample of 6 automated lines and 3 ancillary supply-chain hubs will be observed, involving approximately 48 shop-floor operators, 12 process engineers, and 8 data scientists. Data collection instruments include (i) high-frequency telemetry logs (sampled at 200 Hz) from programmable logic controllers (PLCs) and industrial IoT sensors, (ii) standardized OEE and cycle-time metrics collected over a 12-week pilot, (iii) latency and bandwidth measurements across edge-to-cloud paths, (iv) structured surveys assessing perceived responsiveness, and (v) semi-structured interviews with practitioners to capture governance and security considerations. Data analysis employs regression analysis to quantify latency reductions and OEE gains attributable to edge deployment, ANOVA to compare performance across lines and hubs, and time-series analysis to detect improvements in throughput. Thematic analysis of interview transcripts will uncover organizational and governance factors, cross-validated with the Conceptual Model of Edge-Enabled Smart Manufacturing. A multi-criteria decision analysis (MCDA) will be applied to weigh performance, reliability, security, and cost outcomes, informing the proposed deployment blueprint. Key expected findings include (i) measurable latency reductions of 40–65% for predictive maintenance alerts and real-time control loops, (ii) increases in OEE by 6–12 percentage points in edge-enabled lines, (iii) improved data processing latency and reduced cloud dependency by 50–70%, and (iv) enhanced data governance, with localized data processing improving privacy and incident response times. The results are anticipated to demonstrate that edge computing enables more resilient scheduling, faster fault diagnosis, and better support for AI-enabled defect detection models deployed at the edge. The study will also uncover organizational enablers and barriers, such as the alignment of OT and IT teams, governance standards, and the need for robust edge orchestration platforms. The contribution to knowledge includes a novel, context-specific blueprint for edge-enabled smart manufacturing in an automotive mass-production setting, a comparative empirical evaluation of edge versus cloud-centric architectures in Hyundai’s operational environment, and an integrated governance framework balancing performance, security, and cost. It advances theory by extending the Technology-Organization-Environment (TOE) and Resource-Based View (RBV) perspectives to edge-centric manufacturing, illustrating how architectural choices influence capabilities such as real-time analytics, autonomous decision-making, and supply-chain visibility. Practical implications offer a scalable, repeatable model for other automotive manufacturers seeking to implement edge computing, including guidance on infrastructure topology, data stratification, latency budgeting, and KPI-driven evaluation. The study concludes that strategic edge deployment, coupled with rigorous data governance and an orchestrated resource management strategy, yields substantial efficiency gains and resilience in Hyundai’s smart manufacturing environment. Recommendations include expanding edge-enabled pilots across additional lines, investing in edge-native AI inference optimization, establishing standardized data schemas and security protocols, and developing a centralized edge analytics center of excellence to sustain continuous improvement and knowledge transfer across facilities.