TL;DR
This paper introduces a low-power, privacy-preserving LoRa-based sensor system for collecting urban data on public space utilization, specifically focusing on chair occupancy. The system successfully demonstrated its feasibility and energy efficiency over a two-month field trial, providing valuable insights for urban planning.
What Problem Does This Paper Address?
Urban planners increasingly rely on data to optimize public spaces, but traditional data collection methods like surveys or video analysis are often labor-intensive, costly, and raise privacy concerns. While existing smart city solutions can track general foot traffic, they struggle to capture nuanced usage of specific public infrastructure like benches or chairs. This paper addresses the need for a cost-effective, energy-efficient, and privacy-preserving solution to gather detailed data on how public seating is utilized, enabling better urban design and resource allocation.
Methods
The core of the system is a custom-designed, waterproof sensor node that leverages LoRaWAN for low-power, long-range communication. Each node is equipped with:
- Inertial Measurement Unit (IMU): Detects sitting and standing movements to infer chair occupancy.
- Microphone: Measures ambient noise levels, providing context about the environment (e.g., festivals, construction).
- Temperature and Humidity Sensor: Gathers environmental data, useful for understanding microclimates and user comfort.
- GNSS Module: Provides location data for each sensor node, aiding in setup and device tracking.
The sensor nodes process data onboard, classifying chair occupancy and then transmitting aggregated environmental and occupancy data via LoRaWAN to a central server. The system's firmware is built on the Zephyr RTOS and employs power-gating techniques to optimize energy consumption. The communication protocol is designed to minimize airtime, with data transmissions occurring every four sampling intervals.
Datasets
The paper describes an in-field experimental evaluation conducted over two months on two public squares in a city. The data collected includes:
- Temperature and humidity readings.
- Ambient noise levels.
- Chair occupancy (derived from IMU data).
- GNSS positions.
This data was collected from 16 sensor nodes and is made publicly available as open data. The paper also references using existing city reference sensors for comparison of temperature data.
Key Insights and Ideas
- Privacy-Preserving Occupancy Sensing: By focusing on chair occupancy rather than individual identification, the system offers a privacy-friendly approach to understanding public space usage.
- Energy Efficiency is Paramount: The design prioritizes low power consumption, achieving a battery life of up to two months with all subsystems active, crucial for long-term, large-scale deployments.
- Contextual Data for Urban Planning: Combining occupancy data with environmental factors (noise, temperature, humidity) provides a richer understanding of how and why public spaces are used.
- Robustness in Real-World Conditions: The system was tested in actual urban environments, demonstrating its ability to withstand weather and the challenges of vandalism (with a simple, yet effective, mounting method).
- LoRaWAN's Suitability: The choice of LoRaWAN proved effective for its low power consumption and long-range capabilities, suitable for city-wide deployments.
Key Results
- Battery Life: The sensor node achieves a battery lifetime of 2 months with all subsystems enabled, consuming approximately 33.65 mWh per day. Deactivating GNSS can extend this to up to 5 months.
- Power Consumption Breakdown: The GNSS module is the largest energy consumer (approx. 67% of total), followed by the noise and sitting detection task (9.24 mWh/day). LoRaWAN transmission is the lowest consumer (1.91 mWh/day).
- Deployment Success: Out of 16 deployed nodes, 11 remained operational for the entire two-month test period, with failures attributed to theft or vandalism.
- Data Insights: The collected data revealed distinct usage patterns between a central, busy square (M) and a suburban, transit-oriented square (V), correlating usage with weather conditions and time of day. For instance, warmer days saw higher usage in the central square, while the suburban square showed consistent, practical usage patterns.
Limitations and Future Work
- Vandalism/Theft: The paper acknowledges that 5 out of 16 nodes were lost due to theft or vandalism, highlighting a practical challenge for widespread deployment in public spaces.
- GNSS Power Consumption: The GNSS module is identified as the most significant power drain, suggesting that further optimization or alternative localization methods could improve energy efficiency.
- Future Work: The authors suggest that the collected data can be used to inform future urban planning decisions, and the system's architecture could be extended to incorporate additional sensors for a more comprehensive understanding of urban environments. The publicly available data also serves as a benchmark for future research.
Summary updated