[Paper Review] Data Gathering for Sensing Applications in Vehicular Networks

Mohammad Nozari Zarmehri, Ana Aguiar|arXiv (Cornell University)|Sep 4, 2012

Vehicular Ad Hoc Networks (VANETs)21 references4 citations

TL;DR

This paper proposes Back off-based Per-hop Forwarding (BPF), a receiver-oriented, geographic broadcast protocol for urban vehicular networks that uses destination location in packet headers to prioritize forwarding based on proximity, reducing redundancy while achieving higher packet delivery rates than existing protocols under high-density urban conditions.

ABSTRACT

We propose to use Vehicular ad hoc networks (VANET) as the infrastructure for an urban cyber-physical system for gathering up-to-date data about a city, like traffic conditions or environmental parameters. In this context, it is critical to design a data collection protocol that enables retrieving the data from the vehicles in almost real-time in an efficient way for urban scenarios. We propose Back off-based Per-hop Forwarding (BPF), a broadcast-based receiver-oriented protocol that uses the destination location information to select the forwarding order among the nodes receiving the packet. BFP does not require nodes to exchange periodic messages with their neighbors communicating their locations to keep a low management message overhead. It uses geographic information about the final destination node in the header of each data packet to route it in a hop-by-hop basis. It takes advantage of redundant forwarding to increase packet delivery to a destination, what is more critical in an urban scenario than in a highway, where the road topology does not represent a challenge for forwarding. We evaluate the performance of the BPF protocol using ns-3 and a Manhattan grid topology and compare it with well-known broadcast suppression techniques. Our results show that BPF achieves significantly higher packet delivery rates at a reduced redundancy cost.

Motivation & Objective

• Address the challenge of real-time urban data collection using vehicular networks as a cyber-physical system infrastructure.
• Design a low-overhead, scalable data collection protocol for high-density urban scenarios where all vehicles can be data sources.
• Mitigate broadcast storms in urban VANETs without relying on periodic neighbor exchange or complex topology-based routing.
• Achieve high packet delivery rates with reduced redundancy and low end-to-end delay in dynamic urban environments.

Proposed method

• BPF uses geographic information (current sender and final destination) embedded in each data packet header for per-hop forwarding decisions.
• Nodes receiving a packet calculate a back-off timer based on their distance to the destination, with closer nodes receiving shorter back-off times to prioritize forwarding.
• Suppression is applied to limit the number of forwarders, reducing redundant transmissions while maintaining delivery reliability.
• The protocol operates in a distributed, receiver-driven manner without requiring periodic neighbor list exchanges, minimizing management overhead.
• Forwarding decisions are made on a per-packet basis using only the geographic header and local back-off logic, avoiding global topology knowledge.
• The protocol is evaluated using ns-3 with a Manhattan grid topology under varying node densities and source node counts.

Experimental results

Research questions

• RQ1 How can a broadcast-based data collection protocol efficiently deliver data in high-density urban VANETs with all vehicles as potential sources?
• RQ2 To what extent can geographic proximity-based back-off scheduling reduce redundancy while maintaining high delivery rates compared to traditional suppression techniques?
• RQ3 How does the protocol perform under increasing network load when all nodes generate data, and how does it compare to existing broadcast suppression methods?
• RQ4 What is the impact of path length and sink congestion on data gathering performance in urban VANETs?

Key findings

• BPF achieves significantly higher packet delivery rates (PDR%) than slotted-0.5, weighted-p, and PRAODV under all tested node densities and source counts.
• Despite higher network load, BPF maintains higher PDR% even when 100% of nodes are sources, outperforming other protocols in dense urban scenarios.
• BPF achieves the same number of hops as other protocols, indicating comparable routing efficiency, but with reduced redundancy.
• The protocol produces the same average number of replicas per uniquely received packet as other protocols, but with higher PDR%, proving greater efficiency.
• Without using WAIT_TIME, BPF still achieves the same redundancy level as other protocols but with higher delivery rates, demonstrating superior performance per unit of redundancy.
• In high-density scenarios (9.6 nodes/km), PDR% decreases across all protocols due to congestion, but BPF consistently maintains the highest delivery efficacy.