Indoor Positioning System

Study Case INDOOR POSITIONING SYSTEM Michael Kurniawan (2018123456) University of Cincinnati, 2600 Clifton Ave, Cincinnati, OH 45221, United States kurniaml@mail.uc.edu Abstract Independent living for the elderly can often be a major challenge. Vulnerable health conditions can be dangerous if there is no adequate supervision. Therefore, a system capable of tracking the position of the elderly in real-time indoors is needed. One of the new technologies used for tracking is Ultra-Wideband (UWB), which utilizes radio waves to calculate distance. With radio signals, positions can be determined with precision even between rooms, as the waves can penetrate the air without significant obstacles. This allows the tracking process to be carried out wirelessly. Keywords: indoor tracking, ultra-wideband, DW1000, IoT, ESP32, POSITIONING SYSTEM INTRODUCTION The United States currently has a significant elderly population. In 2021, approximately 16% of the total population will be individuals aged 65 and above. This increase in the number of elderly people marks a demographic shift towards an aging society. As a result, the elderly increasingly need health assistance and monitoring of their daily activities. Not all elderly people have personal caregivers who accompany them at all times. Many families in the US are busy with work, making it difficult to provide full supervision to elderly family members. In fact, elderly people with fragile health conditions require special attention. Therefore, a solution in the form of a remote monitoring system is needed to help families monitor the activities of the elderly. Elderly monitoring systems provide information about the location and daily activities of the elderly. Several healthcare technology companies in the US, such as Philips Lifeline and Apple Health integrations, have developed similar systems that MICHAEL KURNIAWAN 1 Study Case include location tracking and step monitoring. The sensor technologies that can be used in these systems include UWB, Bluetooth, Wi-Fi, and RFID. One of the advantages of a position tracking system is its integration with telepresence robots. If an elderly person is detected in an emergency situation, the robot can be directed to their location. The caregiver controlling the robot can interact directly with the elderly person to ensure their safety. SYSTEM DESCRIPTION Figure 1. Device Architecture Figure 1 shows the framework of the indoor position detection system. The system architecture consists of three main entities: 1. ESP32 UWB sensor that functions as a position tracking device and sends data to the server via MQTT broker. 2. Server that receives, processes data, and provides API services. 3. A smartphone application (iOS/Android) that displays the user's position. This application connects to the server via REST API. MICHAEL KURNIAWAN 2 Study Case Figure 2. System Architecture ESP32 UWB is divided into two parts: anchors (placed at several points in the room) and tags (carried by the elderly being monitored). Anchors and tags exchange signals to estimate distance. The distance calculation data is sent to the server via MQTT. The server then processes the data to obtain position coordinates and sends the results to the smartphone application. Figure 3. ESP32 UWB Anchor Flowchart MICHAEL KURNIAWAN 3 Study Case Figure 4. ESP32 UWB Anchor Flowchart (antenna calibration) MICHAEL KURNIAWAN 4 Study Case Figure 5. ESP32 Tag Flowchart The anchor functions to measure distance with the tag. If calibration is requested, the anchor will recalculate the distance to the nearest tag. The tag only responds to anchor requests and sends information on the time the signal was received and sent. The anchor calculates the distance based on the time difference. Distance Measurement Formula with Time of Flight (ToF): where: • d = distance (meters) • c = speed of light (3 × 10^8 m/s) • Δt = signal travel time difference MICHAEL KURNIAWAN 5 Study Case Figure 6. Server Trilateration Flowchart The server that receives the distance data will process it using a Kalman Filter to reduce noise. After that, trilateration calculations are performed to obtain the tag location coordinates. 2D Trilateration Formula (with three anchors): where: • (x, y) = tag position • (xi, yi) = position of anchor i • di = distance between anchor i and tag MICHAEL KURNIAWAN 6 Study Case Figure 7. REST API Register Flowchart The REST API allows users to register an account so that their identity is stored on the server before using the monitoring service. Figure 8. REST API Login Flowchart The authentication REST API accepts a username and password, then generates a token as the user's digital identity. This token must be sent when accessing other services. MICHAEL KURNIAWAN 7 Study Case Figure 9. REST API History Flowchart This REST API stores trilateration results in a database for historical analysis, such as the daily activity patterns of elderly people. Figure 10. REST API Validation Flowchart Used to check token validity so that users can stay connected without having to log in again. MICHAEL KURNIAWAN 8 Study Case Figure 11. REST API Logout Flowchart Service to delete user tokens, for example when users want to change accounts or log out of the server. Figure 12. Database ERD The database server consists of four tables: 1. Anchor_coordinates table – stores the anchor position points in the room. 2. Location table – stores the trilateration results of the tag position. 3. User table – stores user identity data. 4. User_token table – stores authentication tokens for the REST API. MICHAEL KURNIAWAN 9 Study Case Workflow: trilateration data is stored in the location table, then sent to the user application via REST API. This information allows family members or caregivers to remotely monitor the movement history of the elderly. EXPERIMENTS AND ANALYSIS Device Power Consumption Measurement In power consumption testing, when the device was operating in energy-saving mode, a current of 0.19 A was recorded at a voltage of 5 V. Meanwhile, in high-frequency mode, the current increased to 0.25 A with a constant voltage of 5 V. The maximum current value was taken as a reference. Based on this data, power consumption was calculated using the formula: where: • P = power (Watts) • V = voltage (Volts) • I = current (Amperes) Thus: • Energy-saving mode: P=5×0.19=0.95 • High-frequency mode: P=5×0.25=1.25 When tested using a battery as the energy source, the device can only turn on for ±100 seconds. The total energy used is calculated using the equation: where t = time (seconds). In energy-saving mode: MICHAEL KURNIAWAN 10 Study Case Figure 13. Device Power Consumption Measurement Results Accuracy Testing with Equipment in the Room Figures 14 and 15 show the measurement results when there is equipment around the room and when the room is relatively clean. The measurement results are closer to the actual distance in conditions without much equipment (Figure 15). Interference can occur due to signal reflection from metal objects such as hanging lamps, causing large errors at certain distances. For example, at a distance of 4 meters, the measurement results fluctuate due to signal reflection. Figure 14. Measurement Graph with Equipment in the Room MICHAEL KURNIAWAN 11 Study Case Figure 15. Measurement Graph Without Equipment Around Next, testing was conducted when the antenna was near a wall (Figure 16) compared to a position without obstacles (Figure 17). In Figure 16, the actual distance was 6 meters, but the device calculated approximately 8 meters, resulting in an error of ±2 meters. In Figure 17, the results were more accurate, ranging from 6 to 6.1 meters. MICHAEL KURNIAWAN 12 Study Case Figure 16. Measurement with Antenna Blocked by Wall Figure 17. Measurement with Antenna Without Obstacles From these results, it can be concluded that the antenna should not be positioned directly facing a wall, as physical obstacles can cause measurement distortion. Device Range Measurements Table 1. Device Range Measurements Table 1 shows that antenna orientation affects communication range. The most ideal condition is when the anchor and tag antennas are facing each other, with a range of up to 101.03 meters before the connection is lost. Conversely, when the antenna is pointed upward, the measurement results become unstable because the signal is reflected off the ceiling before being received by the tag. The presence of objects such as lights in that position can also interfere with accuracy. Trilateration Tests Indoors and Outdoors The results of the indoor trilateration experiment are shown in Table 2 with the MICHAEL KURNIAWAN 13 Study Case configuration as shown in Figure 18. Anchors A1, A2, and A3 were placed in specific positions to measure the tag's coordinates. The results show that if the anchors and tags are in the same room, the calculations can be performed correctly. However, when the anchors and tags are separated by a wall, the tracking process fails. Nevertheless, the device is able to overcome obstacles in the form of equipment, although there is still an error of ±0.5 meters due to signal interference. Table 2. Indoor Experiment MICHAEL KURNIAWAN 14 Study Case Figure 18. Experiment Schematic from Table 2 Table 3 and Figure 19 show the measurement results in an open area. The results are more accurate than those in a room because there is minimal reflection from equipment. Almost all points were detected well, except for the position with coordinates x=−5, y=5, which experienced an error on the z-axis of about 0.2 meters due to the tag's position facing away from the anchor. Table 3. Measurement Results in Open Spaces MICHAEL KURNIAWAN 15 Study Case Figure 19. Experiment Scheme from Table 3 Overall, the experimental results show that the system can work with fairly good accuracy in open spaces. However, in closed spaces with a lot of equipment or wall obstacles, the accuracy is still affected. CONCLUSION Based on a series of experiments, this UWB-based device is capable of wireless position tracking with fairly high accuracy in open spaces. However, its performance decreases significantly in enclosed environments with many walls or metal equipment. Key conclusions: 1. Antenna orientation greatly affects measurement results. Facing each other is the most ideal condition. 2. Wall obstacles cause large errors (up to 2 meters), while obstacles from equipment only cause small errors (around 0.5 meters). 3. The power consumption of the DW1000 module is relatively high (around 0.95–1.25 W). Standard battery power sources are not sufficient for long-term operation; a power bank or direct power supply is required. 4. ESP32-based anchor devices are more stable when using a permanent power supply, but their placement is limited because they must be close to a power outlet. With antenna orientation optimization and adequate power supply, this system has the potential to become an indoor positioning system-based monitoring solution for the elderly in the United States. MICHAEL KURNIAWAN 16 Study Case REFERENCE MICHAEL KURNIAWAN 17 Study Case MICHAEL KURNIAWAN 18