Research And Implementation Of Low-power EEG Recording Strategies Based On Microcontroller

EEG recording technologies plays an important role in studies,diagnosis and surgical positioning of neurological diseases such as epilepsy.The traditional EEG devices were focused on improving the number of channels and sampling accuracy.However,the present research emphasis has been changed: gradually increasing the signal sampling rate to study high-frequency oscillation(HFO)signals of EEG;developing implantable,wearable and ambulatory EEG devices with wireless technology;integrating certain signal detection algorithms to reduce manual analysis and judgment.All these three aspects will pose challenges to power consumption,so there is an urgent need to develop low-power technologies for EEG recording.At present,the typical EEG recording system consists of three parts: analog front end,digital control and data processing circuit and wireless transmitter.The digital control and data processing circuit can be implemented in a ASIC or microcontroller.This thesis proposes low-power strategies for microcontroller-based digital signal processing circuits.These strategies mainly consist of three aspects.First,propose dual-cycle conversion mode(DCM),a conversion mode of multi-channel ADC.In this mode,a few channels are distributed higher sampling rate than other channels.The physiological characteristics of high sampling rate channels can be further studied,while the remaining channels are still in general monitoring status.This conversion mode can reduce the bandwidth requirement of EEG device and finally reduce power consumption.In the acquisition test,DCM mode can perfectly record the waveform at a low total sampling rate.Under medium parameters,the estimated power consumption of DCM is about 33% lower than that of the continuous conversion mode.Second,optimize the DMA and improve the hardware's ability to transmit data for multi-channel EEG recording.The first way is to create DMA to data memory fast interface(DDFI)with reducing 50% or 100% waiting periods of microcontroller.Another way is to add the channel data filtering function of DMA which consumed clocks is about 1/8 of software.Third,a software and hardware feedback control based on physiological signal detector is implemented.The active level of the detected effective signal is fed back to the physiological signal detector,and then the software dynamically adjusts the working status of the microcontroller,sampling rate and conversion mode of the ADC controller,and finally the system works in optimum power consumption.The low-power strategies proposed in this thesis were implemented in a SOPC system based on openMSP430.Analog front-end and wireless transmitter can be easily added to the circuit for entire EEG recording system.These low-power strategies are suitable for microcontroller based low-power EEG device.