ABSTRACT
In current times, smart cities have become the conventional methodologies for urbanization. The well-being of city inhabitants is drawing the increased attention of practitioners, researchers, and policymakers. The social, economic, and environmental sustainability, advanced considerations, and high quality of life through digital inclusion are examined as significant constituents in the design of smart cities. The smart city paradigm incorporates physical devices and information communication technology to the IoT network to enhance the good quality of organized services to the citizens. The network traffic of a smart city operates in real-world time via IoT objects. These IoT objects are being associated with sensors/actuators, which is mounting exponentially and initiating new cybersecurity challenges. The objective is to make an essential cyber-physical system to alleviate IoT-related security threats and attacks that exploit security vulnerabilities. So, the objective of this chapter is to propose an intelligent machine learning (ML)-based framework approach for distinguishing and classifying anomalies from normal behavior based on the type of attack. This chapter estimates the complete experimentation performance and evaluations of ML algorithms for recognition of categorical attacks (data probing, DoS attack, malicious control, malicious operation, scan, spying, and wrong setup) found in the DS2OS data set. This chapter estimates the overall experimentation performance and evaluations for the recognition of various categorical attacks found in the cyber-physical system. The experimental results of the simulation model report that the gradient boosting algorithm performs well in categorizing the attacks.
