Bio-Cyber Interface Parameter Estimation with Neural Network for the Internet of Bio-Nano Things

Recently, there is growing interest in the Internet of Bio-Nano Things (IoBNT) based on biological communication with the advent of communication engineering and nanotechnology. One of the IoBNT's significant challenges is modeling and simulating of bio-cyber interface for linking electromagnetic signal- based internet to the biochemical signal-based bio-nano-network. Gaining an understanding of the bio-cyber interface is vital for the design of the IoBNT framework. In this paper, an artificial neural network (ANN) approach is proposed for parameter estimation of the bio-cyber interface. This choice's motivation is the complexity of the mathematical model of the bio-cyber interface system and the possibility of using ANN methods in this perspective. The idea in this work is for ANN and mathematical modeling to complement each other. The proposed approach is based on two directions: using a non-linear least square for fitting electro-bio and bio-electro interface model parameters, then applying ANN to the output of the first direction to learn the design behind the model parameters(to acquire the parameter estimator). The research work proves that ANN can learn (train) from the model parameters. Furthermore, the training network can predict the model parameters for a given system. The results show that ANN achieves effective performance for parameter estimation of the bio-cyber interface. Finally,the performance analysis of the proposed bio-cyber interface device is evaluated by employing a pharmacokinetics compartmental model for molecule diffusion over the blood vessel system.