Position mapping using BLE mesh networks

Sammanfattning

Device location is a practical problem in indoor wireless deployments. Without knowing where devices are positioned in a building, technicians struggle to locate failing equipment, understand coverage issues, and perform maintenance on devices. To facilitate technicians, they often get elevated access to the intranetwork, which not only gives them access to everything, but they still fail to know where devices are located physically in the building. To fully solve this problem, an indoor positioning system is helpful, as it not only displays the devices’ locations in a building but also displays the statuses of the devices. This thesis explores the viability of creating such a system using Bluetooth 6.0 with mesh technology and utilizes three different distance measuring techniques called Phase-Based Ranging (PBR), Round-Trip Time (RTT), and Inverse Fast Fourier Transform (IFFT). For the intended room-level installer support use case, meter-level accuracy is sufficient to identify the correct area rather than exact sub-meter coordinates. Across both Line of Sight (LOS) and Non-Line of Sight (NLOS) conditions, Inverse Fast Fourier Transform (IFFT) performs best, with a mean error of 1.40 m in LOS and 2.40 m in NLOS. In contrast, Phase-Based Ranging (PBR) performs worst with a mean error above 5.00 m for both conditions, while Round-Trip Time (RTT) often performs worse than 1.70 m. The degradation from LOS to NLOS is approximately 40% for PBR, 71% for IFFT, and 102% for RTT. Based on these ranging results, IFFT is chosen for the indoor positioning system. The position map is then reconstructed using two algorithms,Multidimensional Scaling (MDS) and Semidefinite Programming (SDP). The error from IFFT is low enough for reconstruction attempts. Although the IFFT ranging accuracy is higher, reconstruction introduces additional errors in our experiments, SDP performs better with a mean topology error of 3.37 m, compared to 4.07 m for MDS. This increase in error relative to the IFFT input precision is influenced by multiple factors, especially the network topology and the number of devices. Likewise, the performance of each reconstruction algorithm depends heavily on the deployment environment, the number of devices, and node connectivity. It is therefore difficult to draw a general conclusion about which method is superior across all scenarios.