iParking: An Intelligent Indoor Location-Based Smartphone Parking Service

Jingbin Liu, Ruizhi Chen, Yuwei Chen, Ling Pei, Liang Chen2012 · DOI

TL;DR

This paper introduces "iParking," an intelligent smartphone-based service that leverages indoor positioning to streamline the parking experience. By fusing WLAN signals and smartphone sensor data, iParking offers accurate indoor navigation to parking spots and provides location-aware services, addressing the common frustrations of finding parking and navigating complex indoor environments.

What Problem Does This Paper Address?

Finding parking, especially in large indoor spaces like shopping malls or multi-story garages, is a persistent urban challenge. Traditional GPS systems fail indoors, leaving drivers without navigation assistance. This leads to wasted time, increased fuel consumption, higher emissions, and frustration. Existing parking systems often lack proactive reservation capabilities or accurate indoor guidance. This paper tackles the need for a seamless, intelligent parking solution that bridges the gap between on-road navigation and precise indoor location awareness, enhancing both user experience and parking facility efficiency.

Methods

The core of the iParking system is its Hybrid Indoor Positioning Engine (HIPE). This engine fuses data from multiple sources:

  • WLAN RSSI (Received Signal Strength Indicator): The system learns the statistical dependency between WLAN signal strength and geographical locations. A Weibull function is used to model the probability density function of RSSI observables, which is more efficient and accurate than traditional histogram-based methods.
  • Smartphone Sensors: This includes accelerometer and compass data for Dead Reckoning (DR). DR is used to propagate position estimates during the intervals between WLAN scans, significantly improving the availability of positioning updates. The system employs different DR algorithms for "vehicle mode" (before parking) and "pedestrian mode" (after parking).
  • Hidden Markov Models (HMM): HMMs are used to fuse the different types of measurements (WLAN RSSI and motion dynamics information from sensors) to estimate indoor locations.
  • RSSI Integrity Monitoring: A novel aspect is the monitoring of RSSI observables to identify and exclude outliers, thereby improving positioning accuracy and reliability. This involves checking the "degree of pattern matching" between observed RSSIs and the learned knowledge database.

The client application is a lightweight smartphone app that connects to a distributed architecture including parking facilities and service providers via the internet.

Datasets

Not explicitly detailed as separate datasets. The system relies on:

  • Learned Location-RSSI Dependency: A database is built by collecting RSSI measurements from multiple WLAN access points at predetermined reference points (RPs) within the parking facility.
  • Real-world Parking Environment: The iParking prototype was demonstrated and evaluated in a real parking facility at the Iso-Omena shopping mall in Espoo, Finland.

Key Insights and Ideas

  • Ubiquitous Smartphone Integration: Leveraging the widespread adoption of smartphones makes an intelligent parking solution cost-effective and accessible.
  • Fusion for Robustness: Combining WLAN signals with smartphone sensor data (especially for dead reckoning) is crucial for achieving both accuracy and high availability of indoor positioning, overcoming the limitations of individual sensor types.
  • Integrity Monitoring is Key: Simply using raw sensor data isn't enough. Actively monitoring the integrity of RSSI signals and excluding unreliable data significantly improves positioning accuracy.
  • Dead Reckoning for Availability: DR is essential for providing continuous positioning updates between less frequent WLAN scans, ensuring a smoother user experience.
  • Beyond Just Finding a Spot: iParking aims to provide a holistic experience, including navigation to the spot, guidance to points of interest after parking, and even helping users find their car later.

Key Results

  • Positioning Accuracy: The final DR-integrated HIPE solution achieved positioning errors of less than the width of a parking space (2.9 m) in most epochs.
  • Availability: The DR-integrated solution provides position updates at a rate of 1 second, a significant improvement over the 8-10 second update rate of the grid-based filter and MLE solutions alone.
  • Error Reduction: Compared to the Maximum Likelihood Estimation (MLE) algorithm, the grid-based filter showed significantly reduced RMSE (32.7%), AE (47.7%), and R95 (50%). The DR-integrated solution further improved upon this.
  • Navigation Success: In field demonstrations, users were guided to the correct or an adjacent parking space in over 85% of trials. In a few exceptions, navigation errors were up to three parking spaces.

Limitations and Future Work

  • Financial Aspect Not Covered: The paper explicitly states that the financial aspect of the service was not covered as it was not considered a technological issue.
  • API Restrictions: The development of the HIPE solution is restricted by the API availability of the smartphone platform (e.g., Symbian^3 OS).
  • Future Development:
    • More Functionalities: Expanding the service to support operational requirements of parking operators and offering more intelligent services for pre- and post-parking activities.
  • Wider Deployment: Deploying the service at more parking facilities to reach a mainstream market.

Summary updated

BibTeX
@misc{liu2012iparking,
  author = {Jingbin Liu and Ruizhi Chen and Yuwei Chen and Ling Pei and Liang Chen},
  title = {{iParking: An Intelligent Indoor Location-Based Smartphone Parking Service}},
  year = {2012},
  doi = {10.3390/s121114612},
}
Figure