Software Development Of Cranial Nerve Signal Real-time Processing System Platform

Brain-computer Interface (BCI) achieves the transformation of the brain electrical signal to the practical information, which allows human thought controlling external mechanical or electrical equipment. Amputation or due to brainstem or spinal cord injury causes paralysis of the population is very large in the word at present. The BCI provides a new treatment for their rehabilitation. At the same time, the study of BCI also plays an important role in the field of brain and cognition, artificial intelligence and so on. And it also has wide application potential in the treatment of neurological diseases, aerospace and military.Due to the high complexity of the nervous system itself and human gradually in-depth research on neural signals, the amount of neural signal data collected for analyzing and the complexity of computation are both improving. In this condition, traditional BCI based on the common PC CPU may not meet the demand. In this paper, the design based on the embedded system of high performance and large data computing platform and with efficient prediction algorithm to realize the neural signal data of high-speed real-time predicting. This compute platform consists of two parts: the client application and the computing platform. Especially, the computing platform is composed of two parts, the main-board and the sub-board. The main-board is mainly responsible for the communication with the PC and complete the computing platform of the work process control, neural signal preprocessing and feature vector extraction, and etc. The sub-board is for the training of the neural signals and predictions. The software of the main-board and the sub-board are both Linux, using the technology of network programming, multithreaded programming, and so on to keep high efficiency and stability.Experimental results show that the system has completed brain signals for real-time processing, and it supports for not less than 96 channels of neural signal data predicts. The system uses the RapidIO for high-speed data transmission, and produces series of multi-core processors parallel speed optimization by using the Keystone of TI company, which makes the prediction of neural predicting speed reached 100 miliseconds. It satisfies the calculation speed of the online real-time neural predicting system requirements.