Heterogeneous Multi-Agent System for Brain-Computer Interaction in Routing and Forwarding With Memristive Neuron Networks

In this research, we aimed to design a novel brain-computer interaction (BCI) approach in routing and forwarding mechanisms to realize a networking multi-modal and multi-brain linked control. In recent years, the field programmable gate array (FPGA) has become a popular choice to construct a heterogeneous network for routing and forwarding processes due to its ability to complete a high performance computation in single node. Conventional BCI mode focus on capturing the dynamic EEG signals from multiple electrode channels and this concept neglects the existing complexity and diversity of connecting information between two brain regions. To this moment, we build a heterogeneous multi-agent system (HMAS) architecture to simulate the memristive neuron networks (MNN) with a set of logic units for multi-agent in the Internet. Specifically, we first feed the non-linear features into an intelligent router to adjust their routing and forwarding processes. Subsequently, at one-time step, the output of all testing nodes is written into a heterogeneous logic space, which mainly consists of the sub-agent, forward port, and memory memristive neuron. In the FPGA, each programmable cell selectively integrates and stores a huge number of motion information between multiple interacting channels. We can achieve the networking cooperative functions of forward and route through FPGA large-scale computation. According to the extensive experimental results on several data, we validate the effectiveness of the proposed HMAS-BCI architecture to compare the optimization method of the multiple simulations.