The ubiquitous deployment of mobile and sensor devices is creating a new environment, known as the Internet of Things (IoT). In this new realm, wireless sensor nodes, smart devices along with information and communication systems together constitute the elements of new computing environment. For wide acceptance of IoT applications, guaranteeing their performance is important. However, performance analysis of IoT applications encounters a lot of challenges such as interaction among a number of different technologies, various usage patterns of smart devices, numerous possible transactions, unavailability of suitable testing platforms and so on. In this paper, we deal with performance analysis of a scalable IoT platform that attempts to take a holistic approach for enterprise level data management in the IoT domain as well as development of IoT applications. We predict the performance of specific APIs offered by the platform using queuing network modeling and also validate them through experimental analysis on two deployment platforms. Additionally, we are able to predict performance of a real-life energy monitoring application deployed on this platform in a production environment. Our analysis is done mostly based on data extracted from production environment and requires only limited performance tests.
S. Duttagupta, Kumar, M., Ranjan, R., and Nambiar, M., “Performance Prediction of IoT Application: An Experimental Analysis”, IoT'16 Proceedings of the 6th International Conference on the Internet of Things. ACM, New York, NY, USA, pp. 43-51, 2016.