Abstract
Flying Ad Hoc Networks (FANETs) face significant challenges in data routing due to high-speed 3D mobility and frequent topology changes. This paper introduces the Course-Aware Opportunistic Routing Protocol (CORF), which optimizes data delivery by predicting optimal relay nodes based on flight trajectories and historical encounter patterns. By prioritizing relays moving toward the destination, CORF achieves a delivery rate of approximately 87% while maintaining low network overhead and latency compared to traditional flooding-based protocols.
Introduction
Unmanned Aerial Vehicles (UAVs) are increasingly deployed in collaborative Flying Ad Hoc Networks (FANETs) for critical applications such as disaster relief, search-and-rescue, and environmental monitoring. However, routing data in these environments is notoriously difficult. Unlike ground-based mobile networks, drones move rapidly in three dimensions, causing links to break and reform constantly. Traditional routing protocols often rely on "flooding"—broadcasting multiple copies of a message—which consumes excessive bandwidth and battery. Alternatively, distance-based protocols may fail when a nearby neighbor is moving away from the target. The core challenge lies in selecting the most effective "next-hop" neighbor in a highly dynamic 3D space.
The CORF Protocol
The proposed CORF protocol addresses these challenges by moving beyond simple distance metrics toward a sophisticated "Store-Carry-Forward" mechanism integrated with trajectory prediction.
Store-Carry-Forward Model
CORF utilizes a Store-Carry-Forward approach to handle intermittent connectivity. Instead of dropping packets when a direct path is unavailable, a node stores the data and carries it until it encounters a suitable relay or the destination itself.
Course-Awareness and Transfer Probability
The protocol's decision logic is governed by two primary factors:
- Course-Awareness: Drones exchange velocity vectors to determine relative movement. By calculating the angle between their flight paths, a source drone can identify if a neighbor is moving toward the destination.
- Transfer Probability: Based on the observation that drone missions often follow repetitive patterns, CORF calculates the likelihood of future encounters based on historical data.
When a drone needs to forward a message, it evaluates neighbors based on these criteria. If a neighbor is moving toward the destination and has a high historical meeting probability, it is selected as the relay. If no such candidate exists, the drone continues to carry the data, preventing unnecessary packet loss.
Performance Evaluation
The researchers evaluated CORF using the Opportunistic Network Environment (ONE) simulator, utilizing the INFOCOM05 real-world trace dataset. The simulation involved up to 600 UAVs and 30 Ground Stations (GS) within a 20km x 20km area.
Key results include:
- Delivery Success: CORF achieved a peak delivery ratio of 86.9%, significantly outperforming classical protocols like Epidemic and Prophet, which typically reached 70-80%.
- Efficiency: Unlike flooding-based methods, CORF’s overhead remained low and stable as node density increased, demonstrating superior scalability.
- Latency: By identifying efficient "shortcuts" through the network using trajectory data, CORF maintained lower average latency than baseline methods.
Conclusion and Future Work
CORF demonstrates that trajectory-aware routing is highly effective for the unique dynamics of FANETs. By treating Ground Stations as reliable anchors and prioritizing direction over mere proximity, the protocol ensures robust communication in 3D space. Future research will focus on adapting the protocol for GPS-denied environments, conducting detailed energy consumption analyses, and incorporating obstacle avoidance logic to handle complex physical terrains.
Summary updated