Performance Study of Various Routing Protocols in Opportunistic IoT Networks
S. P. Ajith Kumar, Hardeo Kumar Thakur · Apple Academic Press eBooks · 2023
Opportunistic Internet of Things (OppIoT) networks have been one of the effective evolutions of delay tolerant networks. OppIoTs are different from IoT network as end-to-end routing path may not be available from the source to the destination. It operates in an intermittent, mobile communication topology, employing hops on a store-carry-forward manner, and peer-to-peer transmission. In this network, routes are building dynamically, whenever messages are routed from the source to destination(s), any potential node will act as the next hop, bringing the message nearer to the destination. This is necessary to develop routing protocols that can maximize the message delivery possibility and minimize the delivery failure possibility. Many protocols like Epidemic, PROPHET, Spray and Wait, MLPROPH, KNNR, GMMR, etc… used in OppIoT networks. Most of these protocols suffer from overheads like higher power and memory requirements, large delays, generalization in their approach in a dynamic, mobile environment, 234 misrepresentation of the node attributes in the wrong cluster, etc. Machine learning (ML) is a tool for learning from histories used to select the next hop to forward message under store-carry-forward scheme to reach the message to destination. The trained ML model leverages information on delivery predictability with respect to buffer availability, success rate, transmission speed, node strength, and message live time.