Enhancing cybersecurity in smart and cognitive cities: A systematic mapping of AI-based techniques

The rapid development of smart and cognitive cities has led to significant advancements in urban technology, but also introduced new cybersecurity challenges. This issue is aggravated in cognitive cities, where citizens act not only as service recipients but also as active data generators, heightening the need for human-centric security frameworks. This review examines the application of AI-based techniques to enhance cybersecurity, with a particular focus on human-centric concerns in smart and cognitive urban environments. We conducted a systematic mapping study and identified 173 studies on AI-based threat detection techniques, where qualitative data was collected and analyzed. These studies were analyzed and categorized according to the Cyber Security Body of Knowledge (CyBOK) framework. The findings reveal that while network security is the most extensively studied area in the CyBOK, critical domains such as human factors remain underexplored. We observed that most AI-based techniques concentrate on the detection phase, often using supervised learning, while only a minority incorporate identification, protection, or response phases. AI-driven techniques are often combined with approaches such as federated learning and blockchain, which are pivotal for safeguarding citizen data; however, challenges persist in balancing privacy-preserving methods and detection performance. This review provides valuable insights into AI-driven cybersecurity techniques. It provides a novel CyBOK-based mapping of threats, while also identifying opportunities for future research, including the development of real-world datasets tailored to cognitive cities and the refinement of human-centric solutions. These contributions offer a foundation for researchers, practitioners, and policymakers to enhance the security of smart and cognitive cities.