IoT data quality: a review of cross-layer challenges, infrastructure impact, and blockchain integration opportunities

Abstract

This systematic literature review presents a unified, cross-layer analysis of the factors affecting data quality in Internet of Things (IoT) ecosystems-spanning architectural, analytical, strategic, and brokerage components. Unlike previous studies that treat these dimensions separately, it provides an integrated perspective with a focus on the Publish–Subscribe (PS) communication model-a widely used yet underexplored IoT data exchange mechanism. From an initial 1,027 records, 367 studies are included after screening and eligibility assessment. Particular attention is given to identifying systemic weaknesses in PS brokerage architectures, especially concerning IoT data quality management. The review emphasises understanding IoT architecture and data for quality frameworks. It begins with a detailed analysis of IoT architecture, identifying potential bottlenecks affecting data quality across the perception, network, middleware, and application layers, followed by a structured evaluation of data quality assessment strategies-including statistical, probabilistic, neural network, and deep learning methods. It also assesses analytical approaches that encompass both real-time and historical data processing, applied to stationary and non-stationary IoT data. The review then advances to an in-depth examination of the PS paradigm, identifying systemic vulnerabilities in current centralised brokerage systems that hinder accurate data quality evaluation. To address these challenges, it explores blockchain integration, highlighting its potential to enhance data analytics and decentralisation in IoT data handling. Finally, it synthesises emerging efforts to develop advanced quality assessment models within brokerage environments, aimed at mitigating broker visibility and analytical limitations, as well as addressing data degradation and concept drift during dynamic data processing and verification.