An Incentive Based Dynamic Ride-Sharing System for Smart Cities

Abu Saleh Md Bakibillah, Yi Feng Paw, Md Abdus Samad Kamal, Susilawati Susilawati, Chee Pin TanSmart Cities 2021 · OpenReview

TL;DR

This paper introduces an incentive-based dynamic ride-sharing system designed to alleviate urban traffic congestion by encouraging private vehicle owners to share rides. Through microscopic traffic simulations and a lab-scale prototype, the system demonstrates significant reductions in fuel consumption, CO2 and CO emissions, and average waiting times, while increasing average vehicle speeds.

What Problem Does This Paper Address?

Urban areas worldwide are grappling with severe traffic congestion, leading to increased travel times, higher fuel consumption, and significant environmental pollution. The primary driver of this problem is the high number of private vehicles, often with low occupancy rates (typically 1.1-1.8 passengers per car in Europe and the US). While ride-sharing is a known solution, existing services often prioritize job creation over congestion mitigation, and their effectiveness can be limited. This paper addresses the need for a dynamic ride-sharing system that actively incentivizes users to share rides, thereby reducing the number of vehicles on the road and improving overall traffic flow and environmental sustainability.

Methods

The core of this research is a proposed incentive-based dynamic ride-sharing system. This system operates on a mobile-cloud architecture, facilitating real-time ride matching for travelers with similar routes and schedules. The key innovation lies in its point-reward incentive mechanism, which rewards both drivers and passengers. These points are calculated based on factors like base ride points, travel time, distance, traffic conditions (using a traffic scaling factor $F_{ts}$), and the number of passengers (using a passenger scaling factor $F_{ps}$). This approach aims to balance economic benefits for participants while preventing drivers from treating it purely as a commercial venture, thus focusing on congestion reduction.

To evaluate the system's effectiveness, the authors employed two primary methods:

  1. Microscopic Traffic Simulation: They used the Simulation of Urban Mobility (SUMO) software to model a realistic traffic network in Sunway City, Malaysia. This allowed them to simulate traffic flow under various ride-sharing adoption scenarios and compare the performance against a baseline scenario without ride-sharing.
  2. Lab-Scale Prototype: A prototype using Internet of Things (IoT) concepts was developed. This involved small-scale autonomous vehicles controlled by Arduino Mega 2560, communicating wirelessly via NodeMCUs and MQTT. AprilTags were used for localization within the lab environment.

Datasets

The paper does not explicitly mention or use pre-existing datasets for training. Instead, it utilizes a simulated traffic network of Sunway City, Malaysia, created within the SUMO environment. This simulated environment is designed to mimic real-world traffic conditions, including moderate to congested flows and signalized intersections. The authors also developed a lab-scale prototype which, while not using real-world traffic data, serves as a validation of the system's operational logic in a controlled environment.

Key Insights and Ideas

  • Incentives Matter, But How: The point-based incentive system is crucial. By offering rewards that can be redeemed for products or services, it encourages participation. However, the design aims to prevent it from becoming a purely commercial taxi service, which could paradoxically increase vehicle numbers.
  • Small Participation, Big Impact: A surprisingly small percentage of ride-sharing adoption (as low as 2-10% riders) can lead to significant improvements in traffic performance. This suggests that even moderate adoption can yield substantial benefits.
  • Balancing Driver and Rider Benefits: The incentive structure is designed to provide mutual benefits, ensuring drivers are compensated for sharing their vehicles while also encouraging riders to utilize the service. The scaling factors for traffic and passengers are key to this balance.
  • IoT for Real-World Validation: The development of a physical lab-scale prototype demonstrates the practical feasibility of the system and its components, bridging the gap between simulation and potential real-world implementation.

Key Results

The simulation results, comparing four cases (Case 1: no ride-sharing; Cases 2-4: increasing ride-sharing user participation), highlight the system's effectiveness:

  • Fuel Consumption: Total fuel consumption was reduced by up to 16.40% in Case 4 (10% riders out of 20% users) compared to the traditional system.
  • Emissions:
    • CO2 emissions were reduced by up to 15.16% in Case 4.
  • CO emissions were reduced by up to 19.13% in Case 4.
  • Average Speed: The average speed of cars increased by up to 11.74% in Case 4.
  • Average Waiting Time: While improvements were observed, the paper notes that the difference in average waiting time between the best ride-sharing case and the traditional system was not significant, attributed to consistent traffic signal timing. However, overall waiting times were reduced.
  • Cost Savings: An example calculation shows a potential monthly transportation cost saving of approximately $9.07 for a single user.

Limitations and Future Work

  • Simulation vs. Reality: The authors acknowledge that the performance improvement observed is based on a simulation in a simple, realistic scenario. They suggest that results may differ in more complex road-traffic scenarios and that further validation through large-scale implementation/simulation is desired.
  • User Adoption Threshold: The paper notes that for the system to be truly effective, a certain percentage of users must join. The exact threshold for successful ride request fulfillment is not definitively established.
  • Privacy and Security: Issues related to user privacy and data security are identified as areas that need to be addressed to motivate and attract more users, but these were outside the scope of this particular study.
  • Optimization of Incentives: The values for the incentive parameters ($P_b, P_t, P_d$) were determined hypothetically. Future work could involve optimizing these values using substantial ride-sharing data.
  • Integration with Advanced Systems: The authors suggest future integration with connected and automated vehicles, and advanced traffic signal control systems, to further enhance the discovery of a congestion-free smart city.

Summary updated

BibTeX
@inproceedings{bakibillah2021incentive,
  author = {Abu Saleh Md Bakibillah and Yi Feng Paw and Md Abdus Samad Kamal and Susilawati Susilawati and Chee Pin Tan},
  title = {{An Incentive Based Dynamic Ride-Sharing System for Smart Cities}},
  booktitle = {Smart Cities},
  year = {2021},
  url = {https://openreview.net/forum?id=sza9Ah6lF7},
}
Figure