Artificial Intelligence for Smart Cities: Comparing Latency in Edge and Cloud Computing

A smart city collects and uses data to streamline and improve multiple facets of city life. This approach is becoming an important strategy to keep cities a desirable place to live in while keeping them attractive and beneficial for business. The applications which constitute a smart city range from traffic to waste management, but are all dependent on data acquisition and data processing. This work focuses on the latency difference between edge and cloud computing for smart cities, which is illustrated by a real-life example: a dash cam in a car to monitor traffic in real time. Two scenarios are compared: the first one uses a single board computer on the edge to process the data and to realise inference, whereas in the second scenario, the analysis is done in the cloud, but the result is still returned to the edge. The results show the benefits and disadvantages of edge and cloud computing for a smart city environment for which latency is an important parameter, in particular when time-sensitive applications are considered such as on-board capturing of traffic situations.