The rapid advancements in Internet of Things (IoT) have led to a surge in in-home health monitoring systems, driven by the goal of creating smart homes that support older adults to age in place. Among the various types of monitoring data, users’ daily indoor trajectories are invaluable for assessing health status, as they can provide insights into an individual’s ambulation abilities and behavioral patterns – critical indicators not available from hospital visits alone.
Evolving Sensor Network Landscape
Traditionally, device-free schemes for indoor tracking have been preferred for their user-friendly nature, as they do not require users to carry or wear any devices. However, existing solutions have faced challenges, such as the need for intensive manual calibration of sensor placements, which is feasible only at small scales and by researchers. Scaling these deployments to tens or hundreds of real homes would incur prohibitive manual efforts, making them infeasible for layman users.
Recent advancements in RF-based indoor tracking have shown promising results, leveraging techniques such as WiFi, FMCW, and UWB. These solutions, however, still rely on well-calibrated sensor placements, requiring hours of intensive setup and specialized expertise. Some single-site configurations may offer a more plug-and-play approach, but their performance can degrade significantly when the subject is far from the sensor due to fixed angle resolution.
Introducing SCALING: A Plug-and-Play Solution
To address these challenges, we present SCALING (Self-Calibrating Indoor Tracking), a device-free indoor trajectory monitoring system that a layman user can easily set up by walking a one-minute loop trajectory after placing sensor nodes on walls. The key enabler of SCALING is a novel self-calibrating algorithm that estimates sensor locations using only distance measurements to the person walking in the monitoring space, eliminating the need for intensive manual calibration efforts.
The SCALING framework consists of three main components:
-
Self-Calibration: SCALING formulates the self-calibration problem in a bipartite graph, leveraging the rigidity of the graph to ensure the uniqueness of the geometrical topology and enable convergence of the iterative optimization process.
-
Distance Measurements: SCALING utilizes low-cost COTS UWB radars in distributed nodes to measure distances to the human subject, addressing challenges from multi-path effects and complex human body reflections.
-
Localization: With the estimated sensor locations from the self-calibration process, SCALING applies multilateration for localization and tracking of the human subject’s daily indoor trajectories.
Evaluating SCALING’s Performance
We evaluated SCALING’s performance through extensive experiments, including simulations and real-world testbeds in both dense (3m x 6m) and sparse (45m x 85m) home configurations. The results demonstrate that SCALING achieves satisfactory accuracy, with an 80-percentile error of 53 cm in estimating sensor locations and 405 cm in tracking subjects.
Notably, when comparing SCALING’s tracking accuracy to the classical multilateration approach with known sensor placements, the performance degradation is only 1% – a negligible cost to save hours of intensive calibration efforts with specialized expertise.
Optimizing Sensor Placements for Scalability
To develop practical guidelines for large-scale deployments, we conducted numerical analysis on the impact of sensor placements using Monte Carlo experiments. The results suggest that configurations with widely distributed sensor placements, either evenly or randomly surrounding the monitoring area, consistently provide the best performance (around 10 cm accuracy), outperforming co-linear arrays by a significant margin.
These findings serve as valuable insights for practical sensor network deployments, enabling scalable and accurate indoor trajectory monitoring systems that can be easily set up by layman users without the need for intensive calibration efforts.
Unlocking the Potential of Daily Indoor Trajectories
With the ease of setup and scalability enabled by SCALING, we envision deploying these systems in tens or hundreds of real homes, generating long-term datasets of users’ daily indoor trajectories. This rich data can provide unprecedented insights into individuals’ health status, going beyond just occupancy monitoring to capture nuanced behavioral patterns, such as ambulation ability and engagement in functional spaces.
By leveraging these informative daily trajectories, researchers and healthcare providers can better assess the onset or progression of conditions like Alzheimer’s dementia, leading to earlier interventions and improved quality of life for aging populations.
Conclusion
In this article, we have presented SCALING, a plug-and-play device-free indoor trajectory monitoring system that addresses the challenges of intensive manual calibration efforts and limited scalability faced by previous solutions. Through a novel self-calibrating algorithm and optimized sensor placements, SCALING enables large-scale deployments manageable by layman users, paving the way for new data-driven insights into individual health and well-being.
As the IoT landscape continues to evolve, solutions like SCALING will play a crucial role in unlocking the potential of sensor networks to support aging populations and improve quality of life. By prioritizing ease of use and scalability, we can empower a wider range of users to benefit from the transformative power of smart home technologies.
Visit sensor-networks.org to explore more innovative sensor network solutions and their applications.