Comparison Of IoT Architectures Using A Smart City Benchmark

Several architectures namely Cloud, Fog, Mist, and Edge computing have been proposed for IoT systems, but efforts to practically correlate between them have proved hollow due to difficulties in creating live systems that replicate the frameworks. In this paper, a comparison of these architectures has been made. Simulation models have been developed to form an understanding of the variation in their performance for different values of configuration and workload parameters in a parallel multi-query processing environment. To derive realistic simulations for sensor-data, the Citybench smart city dataset has been used, which is extracted from measurements taken by traffic, temperature, and other sensors, along with a wide range of queries generated in a smart city application. By using seven highly contrasting single and six ratio based mixed Queries, it can be seen that Edge with proximity to query devices and sensors achieves the least latency in four of the single and five of the mixed query simulations with Mist being a close second in the same. However, Fog due to its distributed and high computational capability obtains minimum latency for two single query simulations with Cloud being best for compute-intensive cases.