Performance Evaluation of Fuzzy-Based Routing Protocols for Opportunistic Networks

Abstract

AbstractOpportunistic networks are a class of delay-tolerant networks that can provide communication facilities in situations where no end-to-end path is available between the source and destination nodes and intermittent connectivity prevails. In this context, the design of routing protocols for such type of networks is a challenge. In the literature, various routing mechanisms for opportunistic networks have been introduced, in particular, those that involve fuzzy logic in the approach used for selecting the appropriate nodes for message routing from source to destination. This chapter studies the performance evaluation of three recently proposed fuzzy-based routing protocols for opportunistic networks using the ONE simulator, namely, the Reinforcement Learning-based Fuzzy Geocast (RLFGRP), the Fuzzy-based Check-and- Spray Geocast Routing Protocol (FCSGRP) for opportunistic networks, and the Multi-Agent Reinforcement Learning Congestion Control (MARL-CC), in terms of delivery ratio, average latency, and overhead ratio, under varying number of hosts, buffer size, and time to live (TTL). The considered nodes’ mobility models are the Shortest Path Map-Based Movement (SPMBM) model and the INFOCOM 2006 real mobility traces. Simulation results show that RLFGRP outperforms the FCSG and MARL-CC under the predefined metrics.