Feedforward Neural Network-Based Data Aggregation Scheme for Intrabody Area Nanonetworks

Abstract

An intrabody area nanonetwork (intra-BANN) is a set of nanoscale devices, which have outstanding cellular level precision and accuracy for enabling noninvasive healthcare monitoring and disease diagnosis. In this article, we design a novel feedforward neural networks (FFNNs) based data aggregation scheme that integrates the attributes of artificial intelligence to boost the computational intelligence of intra-BANNs for prolonged network lifetime. In the proposed scheme, data division and labeling are performed to transmit detected information using two different types of packets with different sizes to save energy resources and to avoid redundant data transmission. FFNN-based periodic data transmission exploits the fitness function approximation characteristics of FFNN to increase the transmission probability of critical information with minimum energy consumption and delay, whereas our proposed event-driven data transmission also ensures the transmission of high priority data with minimal delay and storage overhead. The detailed evaluation and comparison of our proposed framework with three existing schemes conducted using the Nano-Sim tool highlight that our proposed scheme performs 50%–60% better than state-of-the-art schemes in terms of residual energy, delay, and packet loss.