10th International Congress on Information and Communication Technology in concurrent with ICT Excellence Awards (ICICT 2025) will be held at London, United Kingdom | February 18 - 21 2025.
Authors - Sebastian Kiunke, Lars Wischhof Abstract - Direct communication between mobile devices enables a wide range of distributed sensing applications, e.g. environmental monitoring or the generation of decentralized pedestrian density maps. In many cases, data generated by these distributed sensing applications is evaluated by artificial intelligence (AI), for example, to predict critical situations in the future and react before these occur. However, training AI models that can be used for this task requires a large amount of data. Detailed discrete event simulations of relevant situations, such as crowded areas, are time-consuming, particularly in the case of many communicating sensors. In this paper, the application layer within the network simulation is subset by two less complex communication models to reduce the simulation time. To evaluate if the simplified modelling leads to significantly different simulation results, a typical example application is evaluated. It creates a local pedestrian density map based on direct communication. Initial results confirm that the detailed 4G/5G communication model can be replaced by a simpler model to reduce the required simulation time by up to 56% in specific scenarios. However, one of the two evaluated example scenarios also leads to a significant underestimation of the mean square error of the sensed data values.
