Purpose This study aims to bridge this gap by examining electronic-nose research from a system-level perspective rather than treating sensor materials, signal degradation, data processing and learning models as isolated components. Although significant advances have been achieved in electronic-nose systems, their transition from laboratory research to reliable real-world deployment remains limited. Design/methodology/approach A systematic review of studies published between 2021 and 2025 is conducted and reorganized into a unified, deployment-oriented processing pipeline. The framework integrates sensor material design, drift formation and compensation, data representation, learning paradigms and implementation constraints, with explicit emphasis on cross-component interactions across the sensing-to-artificial intelligence (AI) continuum. Findings The analysis synthesizes recent advances in electronic nose technology and AI while identifying key unresolved challenges, including long-term sensor drift, limited data set representativeness, evaluation bias, weak model interpretability, poor cross-environment generalization and computational constraints in low-power deployment. These factors collectively constrain scalability, robustness and practical applicability. Originality/value Unlike existing surveys that primarily summarize individual subsystems, this review introduces an integrated system-level analytical framework that reveals structural bottlenecks across the full processing chain. It provides interdisciplinary research directions to support scalable, reliable and deployment-ready artificial olfaction systems.
Ha et al. (Thu,) studied this question.