Fuzzy Reinforcement Learning in Opportunistic Routing for Wireless Sensor Networks

In recent times, routing in wireless sensor networks (WSNs) has emerged as one of the key research challenges because of the dynamic characteristics and constrained resources of these networks. Opportunistic Routing (OR) has surfaced as an effective model that utilizes the broadcast features of wireless communication to improve network efficiency. The fundamental concept of OR is to choose a suitable candidate subset for each node, whereby, upon receiving a packet, only the best candidate sends it on, while the others discard it, thus enhancing reliability and minimizing redundancy. This paper aims to identify the optimal group of candidates in opportunistic routing. This document suggests a new hybrid routing method, FRLOR (Fuzzy Reinforcement Learning-based Opportunistic Routing), that combines Fuzzy Logic (FL) with Reinforcement Learning (RL) to enable smart, dynamic, and adaptive candidate selection in opportunistic routing. The fuzzy inference system assesses three fundamental input factors—geographical distance, neighbor node density, and link probability—to identify an initial candidate set. The RL element subsequently enhances this collection by persistently learning from network feedback and optimizing policies, choosing the most effective forwarding nodes. The effectiveness of the suggested FRLOR technique was assessed and contrasted with current algorithms like EEFLPOR, POR, and DPOR based on Expected Number of Transmissions (ENT), Execution Time, End-to-End Delay (E۲E Delay), Packet Delivery Ratio (PDR), and Energy Consumption. Simulation outcomes indicate that the integration of fuzzy reasoning with reinforcement learning greatly improves routing efficiency and network performance in comparison to conventional approaches.

کلیدواژه ها:

Wireless sensor networks ، Candidate selection ، Opportunistic routing ، Fuzzy systems ، Reinforcement Learning