Edge server placement on traffic lights for smart city services

Cities of the future will be smarter, with services being driven by citizens and adapted dynamically with a goal of ensuring higher efficiency and safety. Towards this vision, Internet-of-Things (IoT) collect data from the city to provide more information about the context and operate more smartly. However, processing the data generated by these sensors may not take place on the sensors due to their resource limitations. Hence, there is a need for edge computing infrastructure. As a densely deployed city infrastructure, street poles can be a promising candidate. In this study, we consider such a system and design an algorithm to decide where the edge servers should be placed to minimise costs without affecting latency and processing time of the edge servers. Using the data on traffic light locations for Enschede, we analyse the performance of our algorithm in comparison to a simple baseline, which places the edge servers on randomly-selected street poles. Our numerical analysis shows that total costs and number of edge servers deployed can decrease significantly compared to the baseline.