Bionic Conventional Deep Learning Model-Based Optimal Routing in Opportunistic IOT Networks.

Opportunistic Network (OppNet) IoT is a subsection of Mobile Adhoc Network (MANET) in which the connection between nodes is not regulated. In MANET, the message is transmitted to the destination with the known routing path, whereas OppNet can transmit data without having a predefined path for data transmission. Estimating the path between sources and destinations is complicated due to the lack of infrastructure and the frequently changing environment. In this research, efficient routing with the detection or classification of nodes is accomplished with a multi-hop routing-based deep learning model. The EPRoPHET routing algorithm is based on a deep learning strategy in which the energy-efficient routing decision is made based on node classification. The deep learning model optimized deep convolutional neural network (DCNN) is utilized to classify reliable and unreliable nodes based on their ability to deliver the message. The hyperparameters used in the DCNN are updated with the Bird Swarm bionic Model (BSBM). Information, such as node movement, location, distance between nodes, and energy status, is considered when estimating the delivery probability. The decision about the forwarder node is taken with the memory of individual nodes and the previous routing information. The performance of a proposed approach is evaluated and compared with the existing state-of-the-art approaches. For 150 nodes, the proposed model achieved a better delivery probability of 0.96, an overhead ratio of 15.6, a latency of 4,100ms and an energy consumption of 37.5J, respectively. The higher performance obtained with the EPRoPHET routing algorithm represents the efficiency of a proposed approach. [ABSTRACT FROM AUTHOR]