Adaptive Probabilistic RREQ Rebroadcasting Using Thompson Sampling for Mobile Ad Hoc Sensor Networks

Area of Science:

• Computer Science
• Network Engineering
• Wireless Communication

Background:

• Route discovery in Ad hoc On-demand Distance Vector (AODV) networks causes redundant broadcasts, increasing delay and energy waste.
• Fixed probabilistic rebroadcasting methods lack adaptability to varying network conditions.

Purpose of the Study:

• To propose Thompson-sampling probabilistic AODV (TSP-AODV), a lightweight, adaptive extension to AODV.
• To improve route discovery efficiency by reducing redundant rebroadcasts and energy consumption.

Main Methods:

• Implemented TSP-AODV, where nodes adapt forwarding probabilities using local duplicate-pressure context and delayed Route Reply (RREP) feedback.
• Utilized a reward function penalizing high forwarding probability in redundant contexts.
• Evaluated TSP-AODV against AODV, fixed probabilistic rebroadcasting, and other suppression methods via extensive simulations.

Main Results:

• TSP-AODV achieved the lowest normalized routing overhead and highest RREQ suppression ratio.
• No statistically significant Packet Delivery Ratio (PDR) difference was observed compared to fixed probabilistic rebroadcasting.
• End-to-end delay was significantly reduced compared to fixed probabilistic rebroadcasting.

Conclusions:

• TSP-AODV demonstrates effective context-dependent adaptation for efficient route discovery in MANETs.
• The proposed method offers a lightweight and adaptive solution without requiring additional control packets or topology exchange.