Simulating Realistic IoT Network Traffic Using Similarity-based DSE

In the research for novel network technologies and protocols, network simulators play a crucial role by offering a fast and safe evaluation environment. Here, it is crucial that the simulated traffic is similar to representative and realistic network traffic. Such traffic usually strongly depends on the set of connected devices and their application behaviors. These behaviors are often modeled by simulation parameters and inducing a vast design space-comprising all possible parameter settings that each may lead to different simulated traffic traces. To tackle the vast parameter design space, we propose a general methodology that efficiently explores this design space to find optimal parameter settings that result in simulated traffic traces that are most similar to a given trace. For an Internet of Things case study and two state-of-the-art similarity measures, we show that the proposed method can improve trace similarity by up to a factor of 184 and 4.92, respectively.