Aggregated Traffic Models for Real-World Data in the Internet of Things

Traffic models play a key role in the analysis, design and simulation of communication networks. The availability of accurate models is essential to investigate the impact of traffic patterns created by the introduction of new services such as those forecasted for the Internet of Things (IoT). The Poisson model has historically been a popular aggregated traffic model and has been extensively used by the IoT research community. However, the Poisson model implicitly assumes an infinite number of traffic sources, which may not be a valid assumption in various plausible application scenarios. The practical conditions under which the Poisson model is valid in the context of IoT have not been fully investigated, in particular under a finite (and possibly reduced) number of traffic sources with random inter-arrival times. In this context, this letter derives exact mathematical models for the packet inter-arrival times of aggregated IoT data traffic based on the superposition of a finite number of traffic sources, each of which is modelled based on real-world experimental data from typical IoT sensors (temperature, light and motion). The obtained exact models are used to explore the validity of the Poisson model, showing that it can be extremely inaccurate when a reduced number of traffic sources is considered. Finally, an illustrative example is presented to show the importance of having accurate and realistic models such as those presented in this letter.