A frequency domain multiplexing scheme based on kernel density estimation for neural communication systems

Abstract

Transmitting information in engineered neural communication systems is a promising solution to delay-sensitive applications for the Internet of Bio-Nanothings (IoBNTs). As widely used in wired and wireless communication systems, introducing multiplexing into neural communication system could improve channel transmission efficiency. In this paper, we model a neural communication system for IoBNTs and propose a neural signal multiplexing scheme for this system, based on frequency-division multiplexing (FDM) principles. The whole system including channel modeling, neural encoding, demultiplexing scheme, and decoding method using kernel density estimation (KDE) are presented. The optimal parameters for KDE and bit error probability are analyzed, and the performance of the proposed strategy is evaluated in terms of error rate and mutual information rate. The work can help researchers better understanding the underlying mechanism of neural multiplexing and pave the way for the implementation of IoBNT applications.