Big Data Handling Approaches in Smart Cities: Techniques, Algorithms, and Architectures

Abstract

The increase of urban populations is challenging transportation systems, healthcare, energy, public safety, environmental systems, and city services. Smart cities include information and communication technology combined with the Internet of Things, and the use of data for operational decisions. This paper focuses on the algorithms and architectural frameworks for big data associated with smart cities. The papers reviewed show the importance of artificial intelligence, machine learning, and deep learning in addition to data mining, clustering, classification, and optimization techniques for predictive modeling, and detection of anomalies, as well as resource allocation and the automation of decision-making. Real-time data associated with urban areas of a large volume and velocity is handled by the combination or use of Hadoop, Apache Spark, MapReduce, cloud, fog, and edge computing. The literature shows the use of hybrid cloud-edge systems in smart city applications, such as intelligent transportation, emergency response, and monitoring of the environment to enhance system scalability, minimize latency, and make local decisions. Despite the advantages of these systems, issues of data heterogeneity, lack of interoperability, legacy system integration, privacy and security issues, algorithm bias, and insufficient data governance remain. To advance smart city applications, new analytical technologies and robust data governance and sharing frameworks, as well as standardized data-sharing and secure systems, need to be incorporated. The analysis of the literature points out the need for research into privacy-preserving analytics, blockchain-based decentralized data management, explainable artificial intelligence, ethical surveillance, and interoperability.