Improving vehicular safety message delivery through the implementation of a cognitive vehicular network

Ali J. Ghandour, Kassem Fawaz, Hassan Artail, Marco Di Felice, Luciano BononiAd Hoc Networks 2013 · DOI

TL;DR

This paper tackles the issue of control channel (CCH) congestion in vehicular networks, which hinders the reliable delivery of critical safety messages. They propose a novel cognitive vehicular network (CVANET) architecture that dynamically extends the CCH bandwidth by opportunistically utilizing vacant spectrum in the neighboring 5.8 GHz ISM band, significantly improving safety message delivery performance.

What Problem Does This Paper Address?

Vehicular communication systems, particularly those using the Dedicated Short Range Communication (DSRC) standard, are crucial for safety applications like collision avoidance. However, the allocated 10 MHz control channel (CCH) in the DSRC spectrum is proving insufficient for the growing demand in congested urban environments. This congestion leads to increased message delays and a higher risk of messages not being delivered in time, which can have severe safety implications. Existing solutions often involve retransmissions, which can exacerbate congestion, or dynamic parameter tuning, which may not be sufficient. The paper addresses this fundamental problem of spectrum scarcity on the CCH by proposing a more dynamic and intelligent approach.

Methods

The core of the proposed solution is a Cognitive Radio Network (CRN) based Vehicular Ad hoc NETwork (CVANET) architecture. Key components and methods include:

  • Cooperative Spectrum Sensing: Vehicles equipped with Software Defined Radio (SDR) transceivers sense the 5.8 GHz ISM band during their Service Channel (SCH) intervals to identify available spectrum holes.
  • Road Side Units (RSUs) as Spectrum Coordinators: RSUs act as central points for collecting spectrum sensing data from vehicles. They do not perform sensing themselves but aggregate information and manage spectrum allocation.
  • Novel Contention Metric: A new metric, $C_r(t)$, is introduced to quantify the contention level at a given location. This metric linearly combines the historical contention level with the current sensed contention, providing a more robust estimation. It considers factors like access delay and the proportion of un-transmitted safety packets.
  • Fuzzy-Logic Based Spectrum Allocation: A Fuzzy Logic System (FLS) is employed by the RSUs to dynamically determine the amount of additional bandwidth needed to extend the CCH. The FLS takes the sensed contention level (represented by inverse channel utilization and the number of un-transmitted packets) as input and outputs the required additional bandwidth.
  • Non-Contiguous OFDM (NC-OFDM): The paper leverages NC-OFDM techniques to efficiently utilize the newly acquired ISM spectrum, allowing for the co-existence of primary and secondary users.

Datasets

Not discussed in the paper. The evaluation is based on simulations.

Key Insights and Ideas

  • Opportunistic Spectrum Access is Key: The paper highlights that leveraging the underutilized 5.8 GHz ISM band is a practical and cost-effective way to augment the limited DSRC CCH without needing to protect licensed users.
  • RSUs as Intelligent Hubs: Shifting the spectrum management intelligence to RSUs simplifies vehicle-side processing and allows for a more coordinated approach to spectrum allocation across multiple vehicles.
  • Dynamic Contention Measurement: The proposed $C_r(t)$ metric is a significant improvement over static metrics, as it accounts for both immediate and historical contention, making the system more adaptive to fluctuating network conditions.
  • Fuzzy Logic for Intelligent Allocation: Using fuzzy logic allows for a more nuanced and human-like decision-making process for spectrum allocation, effectively translating linguistic terms of contention into concrete bandwidth adjustments.

Key Results

The simulations demonstrate the effectiveness of the CVANET-ON mode compared to the legacy CVANET-OFF (standard 802.11p/1609.4) mode:

  • Reduced Message Delay: CVANET-ON significantly reduces the end-to-end delay of safety messages, especially in congested scenarios. For instance, with 100 vehicles and 800-byte payloads, the delay is reduced from approximately 70ms to under 20ms.
  • Lower Packet Un-Transmitted Risk Index (PURI): The PURI is substantially decreased, indicating a much lower probability of safety messages failing to be transmitted within the CCH interval. For 100 vehicles and 800-byte payloads, PURI drops from around 0.35 to below 0.1.
  • Improved Per-Vehicle Throughput: CVANET-ON shows a marked increase in per-vehicle throughput, as fewer messages are lost due to collisions or timeouts.
  • Higher Packet Delivery Ratio (PDR): The PDR is consistently higher with CVANET-ON, especially under bursty arrival conditions, showcasing the system's ability to adapt to sudden increases in network load.

Limitations and Future Work

  • Simulation-Based: The results are based on simulations using NS2 and Matlab. While comprehensive, real-world testing is needed for full validation.
  • Focus on CCH: The paper primarily focuses on improving CCH performance and does not delve deeply into the impact on SCH performance.
  • Future Work: The authors suggest future research directions including:
  • Studying and comparing various network metrics for CCH contention.
  • Conducting experimental analysis of the spectrum sensing/allocation framework using small-scale vehicular testbeds.
  • Further investigation into the propagation issues on the 5.8 GHz ISM band over urban scenarios.

Summary updated

BibTeX
@article{ghandour2013improving,
  author = {Ali J. Ghandour and Kassem Fawaz and Hassan Artail and Marco Di Felice and Luciano Bononi},
  title = {{Improving vehicular safety message delivery through the implementation of a cognitive vehicular network}},
  journal = {Ad Hoc Networks},
  year = {2013},
  doi = {10.1016/j.adhoc.2013.06.005},
}
Figure