The formation of a smart city is a dynamic process that involves the implementation of systemic steps that transform the city into a comfortable environment for living. Smart cities are evolving on the basis of a flexible telecommunications architecture for IoT devices. Existing sustainability technologies require a large amount of computing power to process IoT data. For effective detection and localization of dysfunctions of complex socio-technical systems of smart cities, it is proposed to use an approach based on a parametric representation of objects of interest. In order to eliminate the influence of the variability of the Internet of Things on the classification accuracy, it is proposed to use a combination of optimality principles, taking into account the parameters of energy consumption, processor and memory usage.