A Digital Communication Twin for Performance Prediction and Management of Bluetooth Mesh Networks

Abstract

Bluetooth Mesh technology can be used to realize heterogeneous IoT networks, consisting of a smart lighting backbone augmented with sensor-based applications. It offers many configuration options to adhere to the diverse application requirements and limit the overhead. Finding an optimal configuration of the network is a complex issue which leads to numerous performance-related questions. In this paper, Digital Twin technology is used to provide continuous and diverse insights into the network's behavior, under monitored or artificial network conditions. The twin is constantly connected to the physical network and combines selective simulation, graph algorithms and a data driven link model into a single toolbox. We evaluated its usage for smart lighting applications on a real-life testbed. The results indicate that the twin can produce end-to-end (E2E) latencies, E2E path distributions and a packet delivery ratio comparable to experiment outcomes in the physical network and in compliance with application requirements. Statistical validation of the similarity between measured and predicted E2E latency distributions indicates a Mean Absolute Error < 10 % and a positive discrete Kolmogorov-Smirnov test for all results.