Self-Sustaining Multi-Sensor LoRa-Based Activity Monitoring for Community Workout Parks

Victor Luder, Michele MagnoSas 2025 · DOI

TL;DR

This paper introduces a novel, self-sustaining LoRaWAN sensor designed to monitor activity in public street workout parks. By leveraging energy harvesting and an adaptive, low-power algorithm, the sensor achieves year-round operation, providing valuable data for urban planning and demonstrating the potential of IoT for community infrastructure.

What Problem Does This Paper Address?

The increasing deployment of Internet of Things (IoT) devices in urban environments, particularly for smart city applications, presents a significant challenge: maintaining these sensors. Traditional sensors often require frequent battery replacements or external power sources, leading to high maintenance costs and operational complexities. This is especially true for public infrastructure like community workout parks, where continuous monitoring is desired to understand usage patterns and justify public funding. The paper addresses the need for a robust, low-power, and self-sustaining sensor solution that can autonomously monitor activity in such public spaces without requiring constant human intervention.

Methods

The core of the system is a custom-designed, low-power IoT sensor node. Activity detection is primarily achieved using a 3-axis Inertial Measurement Unit (IMU) that captures micro-vibrations from user interactions with the workout equipment. An energy-aware firmware algorithm samples data at a low frequency (5 Hz) and uses a ring buffer to calculate variance, identifying activity based on oscillations. To further enhance accuracy and detect presence, a time-of-flight (ToF) sensor is integrated, activated only when the IMU detects potential activity. This cascaded approach minimizes power consumption.

Crucially, the sensor is designed for self-sustainability. It incorporates an energy harvesting circuit that utilizes solar cells to recharge a 330 mAh lithium-ion battery. The system is optimized for extremely low power consumption, with a standby mode drawing as little as 0.712 mW. Data is transmitted wirelessly using LoRaWAN, a low-power, wide-area network technology suitable for long-range communication in urban environments. A user-friendly web interface is also developed to visualize and access the collected data.

Datasets

The paper does not discuss datasets used for training. However, it introduces a new dataset collected from a week-long field test in Zurich, where three sensors were deployed at different street workout parks. This dataset captures activity duration, timing, and presence information, and is made publicly available through an online dashboard.

Key Insights and Ideas

  • Self-sustainability is achievable for urban IoT: The paper demonstrates that with careful hardware design (energy harvesting) and intelligent firmware (adaptive sampling, low-power modes), IoT sensors can operate autonomously year-round, even in challenging conditions like winter with limited sunlight.
  • Vibration analysis for activity detection is effective and low-power: Using an IMU to detect vibrations propagating through metal structures proves to be a reliable and energy-efficient method for identifying activity in workout parks.
  • Cascaded sensing reduces power consumption: Activating more power-hungry sensors (like ToF) only when initial detection (from IMU) occurs significantly reduces overall energy usage.
  • Data from public spaces can inform urban planning: The collected data reveals clear patterns in park usage (peak times, duration, location), offering valuable insights for city planners to justify public funding and optimize community resources.

Key Results

  • The sensor achieved an average measured error of only 2.8 seconds for activity duration detection.
  • The system consumes a mere 1.147 mW in normal operation and as low as 0.712 mW in standby mode.
  • The estimated battery runtime is 46.75 days on a single charge, with the energy harvesting system enabling year-round operation.
  • During a field test, 100% of detected activities were successfully transmitted via LoRaWAN.
  • The ToF sensor correctly detected 100% of activities and their locations within the park.
  • A 24-hour field test revealed peak usage in the evenings, with activity times up to 700 seconds, suggesting simultaneous use by multiple individuals.

Limitations and Future Work

  • Limited environmental sensing: While the current sensor focuses on activity, future work could extend it to collect environmental data (temperature, humidity) for a more comprehensive understanding of park usage context.
  • Enhanced people-counting: The paper suggests that a more elaborated people-counting approach, building upon the existing ToF sensor, could provide even greater value.
  • Broader application scope: The authors envision extending this approach to other public infrastructures and community spaces, demonstrating its versatility.

Summary updated

BibTeX
@inproceedings{luder2025self,
  author = {Victor Luder and Michele Magno},
  title = {{Self-Sustaining Multi-Sensor LoRa-Based Activity Monitoring for Community Workout Parks}},
  booktitle = {Sas},
  year = {2025},
  doi = {10.1109/sas65169.2025.11105143},
}
Figure