No Doze EEG Sleep Detector

Presentation Transcript

No Doze EEG Sleep Detector Ben Schneider David Mahr Kuang Hau Tan ECE 445 Senior Design Nov 30, 2006 Project #9

Introduction Our rest locator fills in as an existence sparing gadget that will ready drivers that are sluggish or have nodded off No successful rest finder for drivers is as of now accessible Drowsiness adds to 1,000,000 crashes every year (1/6 of all auto collisions) 1,500 fatalities 71,000 wounds $12.5 billion

Introduction EEG gathers brainwaves by measuring voltage differentials crosswise over scalp Brain waves shift in recurrence and plentifulness Categorized base on waveform as alpha, beta, delta, and theta waves Type of wave (recurrence of waveform) is demonstrative of a man's mindfulness level Beta waves → individual is wakeful and caution (>12Hz) Alpha waves → onset of rest (8–12 Hz) Theta waves → light rest (4–8 Hz) Delta waves → profound rest (1–4 Hz) Our gadget utilizes the recurrence of EEG flag to decide the sharpness level of client and to analyze rest

Device Features Alerts client in under a moment after rest onset Only 3 anodes required 3 caution volume levels (expanding with time snoozing) Portable (uses non-legitimacy battery) Wireless association with base unit Visual and sound cautions

System Overview Block chart

Our Device Base Alarm Unit Head Gear Unit

System Overview (cont.) Signal Conditioning Wireless Module Buzzer Audio Module EEG Sensor LED Logic Base Alarm Unit Head Gear Unit

Hardware Overview EEG sensor circuit Measures brainwaves Filters and opens up flag Signal Processing Detects recurrence of EEG flag Controls bell and LED mindfulness bar Wireless Module Relays PIC yields to LED circuit

Hardware Overview (cont.) LED Awareness Bar Color reflects client of momentum readiness level Audio Module Start up song Constant tune amid low mindfulness state Alarm motion amid rest Buzzer Circuit Alerts dozing client 3 Volume levels

EEG Module

Purpose Amplify brainwaves an adequate sum so as to be perused by PIC. Keep up good SNR proportion all through enhancement handle. Represent DC balance and overabundance commotion all through circuit through utilization of simple channels.

Challenges Brainwaves are on the request of 10uV Requires a lot of increase Greater pick up = Greater commotion Power line impedance (60Hz) DC counterbalance enhancement (Dipole in Eyes, Shift of wires) Single Channel EEG Most EEG utilize numerous channels and encompass whole head with terminals In enthusiasm of straightforwardness and accommodation of client, we wish to utilize slightest measure of cathodes and wires as could be expected under the circumstances

Instrumentation Amplifier A sort of differential intensifier used to enhance the distinction between two info signals. Voltage contrasts between data sources are opened up and sent to the yield Similarities on information sources are dismisses and weakened. Essential qualities High Common-Mode Rejection proportion (CMRR) Low DC counterbalance Low Drift Low Noise

Gain Desire scope of 1 Vpp to be handled by PIC Because brainwaves measured on the scalp are on the request of 10uV, we intensify by a pick up of 121,825. Max pick up of every instrumentation speaker is 10,000. Utilize two in arrangement (G1=4941,G2=24.656)

Gain 1 st I. Amp Rg=10 Ohms G=4941 2 nd I. Amp Rg=3.3k || 5.1k =2003.57 Ohms G2=24.656

Common-mode Rejection Ratio (CMRR) Ad=Differential Gain Ac=Common-mode Gain Measures inclination of the gadget to reject enter signals regular to both info leads. High CMRR is vital when measuring little voltage differentials To Test This… Look at yield when sources of info are shorted to decide normal mode pick up. Look at pick up of a test flag to decide differential pick up

1 st Instrumentation Amplifier Input flag: 20Hz, .001V Output: 20Hz, 4.930V Common-mode Gain: 200mV Theoretical Gain: 4941 Actual Gain: 4930 CMRR 87.83dB

2 nd Instrumentation Amplifier Input flag: 20Hz, .1V Output: 20Hz, 2.75V Common-mode Gain: <<.01mV Theoretical Gain: 24.656 Actual Gain: 27.5 CMRR >>200dB

Filters Active versus Latent Filters Sharper shorts Less subject to criticism, go about as support Respond faster to unconstrained DC balances

Passive High-pass Filters Cutoff recurrence of .34Hz Used before instrumentation intensifiers Simulation Actual

Active Low-pass Filters Cutoff recurrence of 2.34Hz Used after instrumentation speakers Simulation Actual

Active 6 th Order Low-pass Filter I.C. made by National Semiconductor Cutoff Frequency of 39.44Hz Used instantly before sending sign to PIC Simulation Actual

Active Low-pass and High-go in Series Simulation Actual

EEG Data

EEG Data Adjusted oscilloscope to 50s/div (500sec=8.33 Min) Imported degree information into Matlab Calculated normal recurrence each 5 sec.

Signal Processing Module

Purpose of Module Determine the attention to the client and yield in like manner to different modules Accomplished through finding the recurrence of the EEG flag > 12 Hz → High Alertness > 8Hz & < 12 Hz → Low Awareness < 8 Hz → Sleeping

Algorithm I Determine the normal counterbalance of the EEG motion (more than 256 ms) Sum all information focuses and after that correct move by 8 bits Count the quantity of zero intersections New ADC esteem > normal balance and old ADC < normal balance (or the other way around) then zero intersection has happened Frequency = number of zero intersections (since more than 500 ms) Problems: Variation in yield close limit frequencies EEG flag must be counterbalanced with the goal that all focuses are more prominent than zero however under 5 V Voltage spikes throw off normal esteem

Algorithm I (cont.)

Algorithm II Eliminate DC balance in EEG circuit and redress contribution to the PIC Determine the quantity of positive pinnacles that happen (more than 500 ms) If ADC (t) < ADC (t+1) > ADC (t+2), crest has happened Frequency = 2*(number of pinnacles) Problems: Noise presents positive pinnacles that are not a piece of the recurrence check PIC sees a recurrence higher than the genuine esteem

Algorithm II (cont.)

Algorithm III Determine the quantity of falling edges in amended EEG motion (more than 500 ms) Old ADC > 0 and New ADC = 0 Frequency = 2*(number of falling edges) Sampling rate expanded to diminish uncertainty at move frequencies (40 tests/s → 125 specimens/s) Problems: Noise can present extra zeroes at low recurrence Not an issue if the flag is appropriately increased

Algorithm III (cont.)

Control of LED Awareness Bar and Buzzer Circuit: Transmitter Power and Sleep Indicator Signals LED Bar when an adjustment in mindfulness level happens ↑ Awareness level → T.P. & S.I. high ↓ Awareness level → T.P. high & S.I. low Buzzer 0 and Buzzer 1 Non-resting client → no caution essential Buzzer 0 and Buzzer 1 low Sleeping client → need to sound alert Progression of Buzzer 0 and Buzzer 1 yields isolated by 1.5 seconds (01→10→11)

Buzzer Module

Volume Level Alarms client in dynamically expanding volume (3 levels) Volume evaluated through the voltage provided to the ringer Levels isolated through utilization of a voltage divider Inputs from PIC select the volume of signal Increases in volume at regular intervals until most extreme volume is come to

Wireless Module Components: - HP3 LINX RF combine - LM 555 - Max 232 Receiver LM555 MAX 232 Transmitter

Wireless Module HP3 LINX RF Pair 902-928MHz band, FM/FSK Modulation,15 ft run Outputs Sleep Indicator (DATA) and Transmitter Power (RSSI) MAX 232 Modifies Sleep Indicator and Transmitter Power Produces Level 5V LM555 Modifies Transmitter Power into usable clock hail Produces postponement of 12ms.

Clock Flag To PIC 1 Pin 38 (Sleep Indicator) To LED circuit (Sleep Indicator) To PIC 1 Pin 37 (Power Down) Clock Flag Sleep Indicator 10K Ω To D-Flip-Flop (Clock Flag) 1µF

LED Logic Circuit Consists of OR,AND,NAND and D-FlipFlops and LEDs D-Flip Flop OR doors LED Indicator AND entryways NAND entryways

LED Logic Circuit 2 D-FlipFlop control prompts to LED (Yellow/Red) 3 Color LED (blend of 2 leads), Orange is produced using Yellow and Red LED Color Status

K-delineate LED Logic Circuit Legend

LED occasions

5V From Yellow LED Lead 5V From Red LED Lead To PIC 2 (Pin34) 74F08PC SN74LS74AN To Yellow LED Lead To Red LED Lead 5V Clock Flag (from LM555) SN74LS32N SN74F00N Sleep pointer (from MAX232)

Audio Module PIC, Oscillator and LM386N LM 386N PIC16F877A Oscillator

V S 5MHz Oscillator V S V S From PIC Pin 35 From Red LED Lead 1µF PIC 16F877A 1K Ω PIN36 V S PIN35 5100 Ω PIN10 PIN34 From Output of LED Logic PIN11 PIN33 V S PIN12 PIN32

Audio Module Initially motivated by the Theremin instrument Sound is created by means of the PIC by recreating square waveforms with recurrence from 0.1KHz to 2KHz By making delays in the PIC, the pitch, length of tone and nature of tone can be changed 3 Melodies: Start up, Warning, Danger Increasing & Decreasing Frequency, Continuous Variable Frequency, Discrete with postponements

Flow Chart Of Events Wireless Module Transmission Power Sleep Indicator MAX 232 Transmission Power Sleep Indicator LED Status Yellow/Orange/Red LM555 LED Logic Circuit Clock Flag LED status LED Audio Module Yes