FSF: Applying Machine Learning Techniques to Data Forwarding in Socially Selfish Opportunistic Networks

Area of Science:

• Computer Science
• Networking
• Distributed Systems

Background:

• Opportunistic networks offer communication solutions where traditional infrastructure is unavailable or overloaded.
• Efficient data delivery in these networks remains a challenge, balancing delivery efficiency, delay, and cost.
• Existing routing protocols often rely on network state or node mobility, with social behavior emerging as a promising alternative.

Purpose of the Study:

• To propose a novel routing protocol, FriendShip and Acquaintanceship Forwarding (FSF), for opportunistic networks.
• To enhance data delivery efficiency by incorporating social ties, node selfishness, and device resource levels into routing decisions.

Main Methods:

• Developed the FriendShip and Acquaintanceship Forwarding (FSF) protocol, which classifies social ties and assesses node selfishness and resource levels.
• Utilized logistic functions to model node selfishness and evaluated resource constraints of potential relay nodes.
• Conducted trace-driven simulation experiments using the ONE simulator.

Main Results:

• The FSF algorithm significantly outperforms existing schemes by increasing the delivery ratio by up to 20%.
• FSF introduces a lower number of forwarding events compared to previous methods.
• Buffer management algorithms are critical for network performance, especially in scenarios involving selfish nodes.

Conclusions:

• Accounting for node selfishness and social relationships in routing decisions is crucial for improving opportunistic network performance.
• The FSF protocol offers a more efficient and effective approach to data delivery in challenging network environments.
• Further research into buffer management strategies can further optimize opportunistic network performance.