4th International Conference on Communication Engineering and Computer Science (CIC-COCOS’2022) 137

Implementation of a New Algorithm for Drone Control Using BCI system

Ali H. Abdulwahhab 1, Indrit Myderrizi1, Musaria K. Mahmood3

1department Of Electrical-Electronics Engineering, Faculty of Engineering And Architecture, Istanbul Gelisim

University, Istanbul, Turkey
2department Of Avionics, Faculty of Applied Sciences, Istanbul Gelisim University, Istanbul, Turkey
3department Of Energy Systems Engineering, Faculty of Engineering, Ankara Yildirim Beyazit University,
Ankara, Turkey

Abstract: via Support Vector Machine algorithm and Linear Regression

algorithm for attention. Quantified signals are then used to
The brain neurons are responsible of activating the human generate a binary code used for drone control. Binary code is
movement, and generating the electrical bio-signal inside the used as input for a control algorithm based on two-layer control
brain. These neurons features are invested in several to manipulate a drone with 9 possible movements. Experiments
technologies, which are used the mind waves in controlling the are conducted with various individuals where the results show
applications. Brain Computer Interface (BCI) is an interfacing its high performance of the developed algorithm and the signal
technology between the mind and a processor by sensing brain classification methods. The system outperforms most of the
signal and employing it to perform different tasks. The BCI existing systems with an accuracy of 90.37%. Moreover, the
device used in this research is the one-channel NeuroSky algorithms offer a capability of performing 16 commands
mindwave 2. This paper presents an EEG waves-based method making it suitable for various applications.
to control a device “here a drone” using two active signals; the
eye blinking and the concentration level described by attention Keywords: Brain Computer Interface, EEG signals, NeuroSky
signal. The dynamic classification of these signals is performed device, attention level, eye-blink.
Introduction
There are billions of neurons inside the human mind
responsible of human movements, thoughts, emotions and
behaviours. These neurons features are invested in several
technology, which are used the mind activities to implement
different functions [1]. Usually, remote control is used to
perform the tasks and application control. Since the work of
Hans Berger on mind activities in [2], the researchers work
on developing a system to exploit mind internal signals. The
Brain Computer Interface (BCI) system is a communication
technology between the brain and the computer aiming to
create a connection channel for sending and managing the
signals from the human mind to the hardware of the system Fig. 1. BCI system in different fields
[3]. BCI system adopts two brain signal sensing methods
which are; the invasive and non-invasive methods [4]. A hand – free teleoperation system is established to improve
The invasive method demands surgical involvement for remotely robot control in a construction zone using on BCI
implanting the electrodes in the cerebral cortex to acquire the system in [9]. The mind signals are captured using wearable
brain signals providing high quality signals and good SNR [5]. Electroencephalography (EEG) headset device, to generate
The non-invasive method requires installing the electrodes in the control system's digital commands. The system shows an
a mindset device according to the standardized 10-20 ability to direct the robotic in space and under-water with 90%
electrodes map without any surgery [6]. By using the right accuracy. In [10], A BCI system is emulating a painting
tools, every action generates by the brain can be used as input artwork using brain signals. The NeuroSky mindwave 2 is
data to control various applications in medicine, gaming, used to acquire the attention level for creating an artwork
among others, prompting the increased interest on BCI painting composing characters' animation. The system
developing [7]. Recently, controlling a drone, wheelchair for invests to help the museum's visitors in understanding the
persons with disabilities or BCI-based robot is among the traditional paintings of Chinese cultural. A prototype BCI-
most interesting topics as shown in Fig 1 [8]. based wheelchair to help the disabled individuals is
developed in [11]. With a resulting accuracy of 73%,
NeuroSky device is utilized. The system includes also,
Arduino Mega, H-bridge and two DC- motor. A BCI-based
home automation system is designed using the EEG signals
in [12]. The detected data are processed by Brain Waves-
Automation-Detection algorithm utilizing Raspberry pi.
Double eye-blink is used to, and attention/ meditation signals
are used for ON/OFF the appliances. In [13], a smart wireless

DOI: http://doi.org/10.24086/cocos2022/paper.811
 4th International Conference on Communication Engineering and Computer Science (CIC-COCOS’2022) 138

wheelchair system is controlled to help partial-paralysed Non-

MEG Magnetic Non- Portable
individuals using brain waves based on BCI system. Invasive
NeuroSky mindwave 2 is used to capture the EEG signals Non-
such as attention, meditation and eye-blink. An Arduino fMRI Magnetic Non- Portable
Invasive
micro-controller is adopted to distribute and recognize the
movement commands. The system shows a control accuracy EEG is the most popular non-invasive method of spontaneous
equal to 95%. The brain signals are used for controlling a wave acquisition, and it has several advantages over other
drone based on BCI system in [14]. The EEG mind waves are neuroimaging processes by providing simplicity, low cost,
captured by using Emotiv - Insight headset device. A fast response, and ability to be implemented in many
Raspberry Pi is used to recognize the incoming brain signals, applications [17]. The EEG headset captures the brain waves
and distributes the drone's control commands. The system in different frequency bands using various channels
presents an efficient accuracy equal to 88%. according to the electrodes map. Usually, The EEG signals
are affected by the noise and the other environmentally
The present work purposes a new algorithm for controlling a effects during the acquisition cussing signal distortion and
drone utilizing brain waves skills (attention level and eye- decreasing in the SNR [18].
blink) detected by a TGAM Module provided by NeuroSky
device. The algorithm comprises 2-control layers. In addition, Brain Waves Classification
attention level's dynamic threshold is adopted for improving
the algorithm's accuracy. The brain has different parts produce several patterns of
electric impulses rhythmic [19]. The EEG headset devices
The paper is organized as follows: The general concept of detect these electric impulses rhythmic, which is collected in
BCI system and EEG waves classification is introduced in various frequency bands and amplitude. The brain waves are
section 2. In section 3 the adopted methodology is presented. very complex changing according to the verbs and feelings.
The experimental outcomes and the analysis are shown in 4. The EEG waves are comparatively stable and similar in the
Finally, section 5 concludes the paper. normal people. There are five kinds of typical brain wave [20]:

Delta Waves: The rhythm of these waves is between 0 Hz –

General Framework 3 Hz. They are generated when the person in meditation state
or deep sleep.
BCI system Theta Waves: These waves have rhythm between 4Hz – 7 Hz,
and has less amplitude. They are produced when the person
A typical BCI system comprises four processing steps in deep meditation, imagination or extreme relaxation.
which are signal acquisition and processing, signal feature Alpha Waves: They dominate during the state of relaxed
extraction, translation algorithm, and device command [15]. mental, contemplation, visualization, problem solving and
The brain signals are detected and treatment in the signal resting with eyes closed. The alpha waves have a frequency
acquisition step. They are acquired from the user’s scalp band is between 8 Hz - 12 Hz.
using sensors, working with multi-electrode array. Signals are Beta Waves: the beta waves predominate when the person in
then processed to be suitable for feature extraction. In the alertness state, stable emotion or energy. 13 Hz - 30 Hz is the
feature extraction several techniques are applied for frequency range bands for the beta.
classifying and analysing signals. Once the signals are Gamma Waves: These waves are generated in a high-level
analyzed the next step is the translation algorithm. The information processing. It has frequency range band between
signals are translated into device commands according to the 31 Hz -100 Hz with low amplitude as compared to the other
user’s intention [15]. There are several types of BCI neuro- brain waves.
imaging processes as listed in Table 1.
Methodology
Electroencephalogram (EEG) The brain signals are extracted and processed by TGAM
module provided by NeuroSky company. A Graphical User
The EEG is an observation technique to read and record the Interface (GUI) using Processing + is designed to project and
brain signals. The brain signals are classified based on their supervise the coming signals from the EEG headset device.
electrical activity into three types: spontaneous activity, Eye-blink and attention level are adopted as signals to derive
Evoked Potentials (EP) and the bioelectric events produced the control mechanism of a drone. They are recorded in Excel
by a single neuron [16]. database using GUI. The collected eye-blink data is classified
Table 1.various neuroimaging methods [8] by machine learning based on Support Vector Machine (SVM)
Neuro- classification algorithm to locate the threshold value. The
Measured Possibility Portability
imaging applied blinking is sorted logically ("1" or "0") by Artificial
Movement Measures Measures
Process Neural Network (ANN) trained by the obtained threshold
ECoG Electrical Invasive Portable based on signals’ intensity.
INR Electrical Invasive Portable The attention levels are associated to the concentration of the
Non- individual under test and the observation interval. Linear
EEG Electrical Portable
Invasive
DOI: http://doi.org/10.24086/cocos2022/paper.811
 4th International Conference on Communication Engineering and Computer Science (CIC-COCOS’2022) 139

Regression Method (LRM) is utilized to classify the attention 𝑉𝑖

𝑟𝑎𝑤 𝑣𝑎𝑙𝑢𝑒 ×
level, yielding a dynamic threshold. 4-bit eye-blink codes are 𝑣𝑜𝑙𝑡𝑎𝑔𝑒 = 2𝑁 (1)
generated using four sequential blinks, and employs together 𝐺
with attention level in controlling a drone’s movements. Fig
2 presents the block diagram of the adopted method. where G is the gain of the post-amplifier, 𝑉𝑖 is the maximum
Signal
input voltage equal to 1.8 V, and N is the resolution in bits of
Acquisition TGAM
brain signals Treatment
determination / Drone control A/D. The TGAM module output the
classification

Fig.2. Brain-drone interface block diagram Signals Classification

Acquisition Signals and TGAM treatment The attention and meditation level are related to personal
mental and physical states. The attention level is increased
NeuroSky mindwave 2 with the concentration, while meditation requires more
uses TGAM module to extract and process the brain signals training, sustainable mental and physical relaxation process
in this system. The TGAM module uses single channel with during the experiment. This promotes the use of attention and
dry electrode to capture the brain signals and eye-blink from not meditation [24]. The eye-blink reading can be acquired
the user’s scalp at Fp1 position as clarified in Fig 3. from the raw wave EEG. The NeuroSky TGAM module reads
the eye-blink as an unsigned single byte with range strength
between 1-255 integer value. The binary code designed in the
present work requires the correct classification of attention
and eye-blinking signal to drive the correct control action.

Attention Level Classification

The human's concentration levels cannot continue for a
long time in a linear or semi-linear high’s levels. After 10-15
seconds, the human’s concentration levels start lacking
causing irregular fluctuations [25]. For an accurate definition
of the attention level’s threshold value, Linear Regression
Method (LRM) is applied. This classification of collected
Fig.3. NeuroSky mindwave device [21]. attention levels provides a dynamic threshold value
separating between strong and weak attention level (binary-1
The TGAM Module comprises three electrical pins which are; and binary-0). The LRM is a statistical method used for
the brain signals acquisition, reference and the GND. The representing a relationship between two continuous variables
module uses potential difference between the extraction or factors by providing experimental data's best linear
electrode and the electrode of reference for eliminating the approximation. The general LRM equation’s form is given by:
noise. The TGAM’s TX pin is connected to the Bluetooth’s 𝑆 = 𝑎𝑇+𝑏 (5)
RX pin during the work, that together are supplied with 3.3 V where, S is the dependent variable, T is the independent
to realize the communication between the modules [22]. variable, a is the slop, and b is the y intercept. The collected
attention level’s data are used to calculate the LRM
The EEG signals suffer from weak amplitude, and noise equation’s constants (a and b) using the following equations:
sensitivity during the capturing stage. Therefore, the TGAM (∑𝑁 ̅ 𝑁 2 𝑁 𝑁 ̅
𝑖=1 𝑆𝑖 )(∑𝑖=1 𝑇𝑖 )−(∑𝑖=1 𝑇𝑖 )(∑𝑖=1 𝑇𝑖 𝑆𝑖 )
𝑎= (6)
𝑁(∑𝑁 2 𝑁
𝑖=1 𝑇𝑖 )−(∑𝑖=1 𝑇𝑖 )
2
module works on processing and filtering the captured signals
using pre-treatment circuit before transmitting to the PC as
𝑁(∑𝑁 ̅ 𝑁 𝑁 ̅
𝑖=1 𝑇𝑖 𝑆𝑖 )(∑𝑖=1 𝑇𝑖 )(∑𝑖=1 𝑆𝑖 )
shown in Fig 4. 𝑏= (7)
𝑁(∑𝑁 2 𝑁
𝑖=1 𝑇𝑖 )−(∑𝑖=1 𝑇𝑖 )
2

Digital
Analog
AFE
A/D
Logic UART Bluetooth
Digital where, 𝑇𝑖 is data collection time, 𝑆̅𝑖 is the average of the
Signal Converter Signal
circuit experimental attention levels of the five individuals and N is
Fig. 4. Pre-treatment circuits block diagram the number of readings.
Essentially the pre-treatment circuit has Analog Front End
(AFE) stage, analog-to-digital convertor (A/D), digital logic Eye-Blink Strength Classification
circuit. Once the AFE stage accomplishes, the signals pass to The drone’s control commands are executed using eye-blink
the A/D converter for sampling at 512 Hz, and coded with 12 signal based on the intensity of blink. In order to define an
bits. The digital logic circuit is used to output the serial data accurate blink threshold separating on-blink represented by
via Universal Asynchronous Receiver and Transmitter 1-binary and off-blink for 0-binary, many tests on different
(UART) interface. Typically, the UART output is connected users must be performed. The Support Vector Machine (SVM)
to a Bluetooth module UART input for transmitting the algorithm is utilized to classify the collected data. The output
signals to the PC [23]. The equation of converting raw value of the SVM is the optimal eye-blink intensity used for
to the voltage is given by: classification.

Algorithm Development
DOI: http://doi.org/10.24086/cocos2022/paper.811
 4th International Conference on Communication Engineering and Computer Science (CIC-COCOS’2022) 140

The developed algorithm is based on multi-layer Start

order. It is a two control layers where the first is performed

Eye-blink code (EBC)
using the eye-blink codes (EBC), while the second is detection

activated by the attention level (ATT) as depicted in Fig 5.

The EBC represent the First Control Layer (FCL), while the EBC=1111
in 5 sec
ATT represents the Second Control Layer (SCL). No

Yes
According to the coming signals from the brain to generate
Attention (ATT) detection
EBC, the approved algorithm is employed to control various

Yes
drone's movements as follow; 0000/ Land, 1111/Take-off,
1001/Up, 0110/Down, 1100/Left, 0011/Right, 1110/Forward, ATT > y
During 3 sec
0001/Backward and 1010/Stop. All the movements are

Yes
executed using both layer in a successive manner except the Drone Takeoff
stop motion, which is executed by using only the EBC
because it can be a critical movement. EBC detection during 5 sec.
Regardless of the blink strength, a timer of 5 seconds is
triggered with the first active blink for generating the EBC. Yes EBC= 1010 No
Once the generating EBC, the ATT is detected based on the
movement to be executed, and detects during a surveillance Drone Stop
Timer sec
period of 7 seconds. The ATT must be greater than detected No ATT = y
"S" during 3 seconds to execute the SCL, else the device
Yes
carries on with the current movement.
Once the devices are coupled, the drone is ready for receiving No
ATT > y
During 3 sec
the Take-off command. After Take-off, the drone waits to

Yes
receive for the next movements' commands based on the Yes
Drone (Forward-Backward-
user's intention. As soon as the drone is received the land's Left-Right-Up-Down)
No Drone Land

command "0000", the drone is down, and turns off after 15

Yes
seconds if not Take-off again.
Timer sec
EBC=1111?

Experimental Results

No
The system is evaluated through a practical experiment
including five participants. The practical experiment is Device OFF

executed in a convenient position and quiet ambience free of Fig. 5. Drone control algorithm
the negative factors affecting.
Table 2 Calculation of LRM constants
4.1 Evaluation of dynamic attention threshold
𝑇𝑖 𝑆𝑖I 𝑆𝑖II 𝑆𝑖III 𝑆𝑖IV 𝑆𝑖V 𝑆̅𝑖 𝑇𝑖2 𝑇𝑖 𝑆̅𝑖
A period of 11 seconds is adopted to collect the attention 1 38 24 54 29 56 40.2 1 40.2
levels' data for five individuals experimentally as shown in 2 66 72 56 53 93 68 4 136
Table 2. After calculating the LRM’s constants by the 10
equations 5, 6, and 7, the attention level’s dynamic threshold 3 83 91 74 77 85 9 255
0
is credible as: 10 10 383.
𝑆 = 3.0273𝑥 + 65 (8) 4 96 96 87 95.8 16
0 0 2
The adopted mechanism of the attention level for controlling 10 10 10
is as follows; achieving strong value (logic "1") by raising 5 93 84 95.4 25 477
0 0 0
over the attention level more than the threshold value for three 10 548.
second. 6 88 97 81 91 91.4 36
0 4
641.
7 91 85 96 96 90 91.6 49
2
8 87 88 94 96 80 89 64 712
667.
9 90 93 90 90 88 74.2 81
8
10
10 97 81 90 83 95 89.2 892
0
10 12 102
11 81 84 97 96 93
0 1 3

In agreement with the attention level's work mechanism, the

attention level shows an improvement with the time
according to the dynamic threshold as compared to the static
threshold as clarified in Fig 6. In Table 3, the comparison
DOI: http://doi.org/10.24086/cocos2022/paper.811
 4th International Conference on Communication Engineering and Computer Science (CIC-COCOS’2022) 141

between the thresholds value shows the gradually adapts of

the dynamic threshold with the attention level's changes over
the time as compared with the static threshold.

Fig. 7. Eyed-blink reading for five persons

Fig.6. Dynamic and static thresholds of attention levels

Table 3 Comparison of threshold types

Static threshold (85) Dynamic threshold
𝑇 𝑆𝑖I 𝑆𝑖II 𝑆𝑖III 𝑆𝑖IV 𝑆𝑖V 𝑆𝑖I 𝑆𝑖II 𝑆𝑖III 𝑆𝑖IV 𝑆𝑖V
1 38 24 54 29 56 38 24 54 29 56
2 66 72 56 53 93 66 72 56 53 93
10 10
3 83 91 74 77 83 91 74 77
0 0
10 10 10 10
4 96 96 87 96 96 87
0 0 0 0
10 10 10 10 10 10 Fig. 8 Distribution of collected eye-blink signals
5 93 84 93 84
0 0 0 0 0 0 4.3 Experimentation of the developed algorithm
10 10 Ones the eye-blink and attention dynamic thresholds are
6 88 97 81 91 88 97 81 91
0 0 defined, the drone control via the algorithm can start. Each
7 91 84 96 96 90 91 84 96 96 90 participant makes three trials for each motion during the
*Cells in red present faulty readings of attention level experiment, and the executed motion's average time is
calculated. Table 4 shows different average times for each
4.2 Experimental evaluation of Eye-blink threshold participant in performing motions, and the demanding
average elapsed time to implement each movement.
The five persons are asked to generate six consecutive In Table 5, the total performance accuracy of the practical
blinking of different strength in random way as shown in Fig experiment, the average accuracies of each participant and
7. The SVM algorithm is utilized to classify the collected data, each movement in the experiment is listed.
and results an optimal threshold equal to 72 eye-blink The proposed algorithm's results show an efficient accuracy
intensity separating between the strong and slight eye-blinks equal to 90.37 with 9.63% error rate providing higher
reading. After the threshold's definition, the collected data is performance than the other algorithms. Moreover, the
sorted logically ("1" and "0") using the trained ANN with the commands' number of controlling the drone's motion has
obtained threshold as shown in Fig 8. been significantly raised.
The Logically classification is used to generate the command
code, its composed of 4-bit generating by four sequential Table 4 Average elapsed time for every motion
reading of eye-blink, this provides 16 commands are; 0000, elapse
0001, 0010, 0011, 0100, 0101, 0110, 0111, 1000, 1001, 1010, Motion P1 P2 P3 P4 P5
d time
1011, 1100, 1101, 1110, 1111. According to an experimental Take- 12.34
test for different participants, five seconds period has been 13.5 11.8 11 12 13.4
off
fixed to generate each eye-blink code. 10.4 12.6 11.99
Land 13 11.3 12.6
3 6 8
12.4 11.28
Up 12 10 11 11
2 4
11.6 11.5 12.10
Down 12 14 11.3
7 5 4
10.8 11.01
Right 11 11.5 11 10.7
8 6

DOI: http://doi.org/10.24086/cocos2022/paper.811
 4th International Conference on Communication Engineering and Computer Science (CIC-COCOS’2022) 142

10.6 10.9 11.46 Hand." 2020 5th International Conference on

Left 13 11.7 11
7 5 4 Communication and Electronics Systems (ICCES). IEEE,
Forwar 12.3 10.8 11.76 2020.
13 10.3 12.3 B. Blankertz et al "Boosting bit rates and error detection for
d 3 7
Backwa 11.6 11.2 10.9 11.17 the classification of fast-paced motor commands based on
10 12 single-trial EEG analysis," EEE Transactions on Neural
rd 7 3 5
Stop 3 3.8 4 3.5 3.7 3.6 Systems and Rehabilitation Engineering, vol. 11, no. 2, pp.
127-131, Jul 2003.
Table 5 Drone control algorithm accuracy Mohd Azali, N. N, “Classification of eeg signals for human
Motion P1 P2 P3 P4 P5 Accur computer interface (hci) application,” Ph.D. dissertation,
acy Dept. Elect. Eng., Tun Hussein Onn Univ., Malaysia, 2015.
Takeof 3|3 3|3 3|3 2|3 3|3 93.33 G. R. a. G. P. Muller-Putz, "Control of an electrical
f % prosthesis with an SSVEP-based BCI," IEEE Transactions on
Land 3|3 3|3 3|3 3|3 2|3 93.33 biomedical engineering, vol. 55, no. 1, pp. 361-364, 2007.
% R. a. A. V. D. Rai, "Fragmentary shape recognition: A BCI
Up 2|3 3|3 3|3 3|3 3|3 93.33 study," Computer-Aided Design, vol. 71, no. 1, pp. 51-64,
% Feb 2016.
Gu, Xiaotong, et al. "Eeg-based brain-computer interfaces
Down 3|3 2|3 2|3 3|3 3|3 86.67
(bcis): A survey of recent studies on signal sensing
%
technologies and computational intelligence approaches and
Right 2|3 3|3 3|3 3|3 3|3 93.33
their applications." IEEE/ACM transactions on
%
computational biology and bioinformatics (2021).
Left 3|3 3|3 3|3 2|3 2|3 86.67
Liu, Yizhi, Mahmoud Habibnezhad, and Houtan Jebelli.
%
"Brain-computer interface for hands-free teleoperation of
Forwar 3|3 2|3 2|3 3|3 3|3 86.67
construction robots." Automation in Construction 123 (2021):
d %
103523.
Backw 3|3 2|3 3|3 3|3 2|3 86.67 Chen, Zitong, et al. "Paint with Your Mind: Designing EEG-
ard % based Interactive Installation for Traditional Chinese
Stop 3|3 2|3 3|3 3|3 3|3 93.33 Artworks." Proceedings of the Fifteenth International
% Conference on Tangible, Embedded, and Embodied
92.59 92.59 85.18 92.59 88.89 90.37 Interaction. 2021.
% % % % % % Abuzaher, Mazen, and Jamil Al-Azzeh. "Mind-Wave
Wheelchair System." International Review on Modelling and
Conclusion Simulations (IREMOS). 2021, Vol.14, iss.1, pp.18-23.
The people suffering from spinal injury or reductions of Raja, P. Dinesh Anton, et al. "Feature extraction and
motor skills will have the ability to perform tasks and classification of EEG signal based anomaly detection and
communicate with the society by using BCI system. New home automation for physically challenged/impaired people
algorithm/method using EEG waves collected and transferred using neurosky mindwave headset." AIP Conference
by a BCI system is presented. The case study in this research Proceedings. Vol. 2207. No. 1. AIP Publishing LLC, 2020.
is focused into controlling a drone wit many movements, but Al-Wesabi, Fahd N., et al. "Smart Mind-based Approach to
results can be generalized for many applications. A two-layer Control Wheelchair Wirelessly." Sensors and Materials 32.4
control algorithm based on eye-blink, and attention levels is (2020): 1533-1555.
tested foe many volunteers. Two original methods are used MARIN, Iuliana, Myssar Jabbar Hammood AL-
for signal classification which are the SVM, and LRM. BATTBOOTTI, and G. O. G. A. Nicolae. "Drone Control
Dynamic signal classification is positively reflected to the based on Mental Commands and Facial Expressions." 2020
algorithm execution accuracy. 12th International Conference on Electronics, Computers and
The algorithm is validated by a test experiment using Artificial Intelligence (ECAI). IEEE, 2020.
single channel NeuroSky module. The proposed algorithm Tiwari et al., “Sensitive Brain-Computer Interface to help
shows a high performance with 90.37% accuracies. Moreover, manoeuvre a Miniature Wheelchair using
the algorithms offer a capability of performing 16 commands Electroencephalography,” in IEEE International Students'
making it suitable for various applications. Conf.on Electrical, Electronics and Computer Science
References (SCEECS), Bhopal, India, 2020, pp. 1-6.
Namdev, Kavita, and Mohd Maroof Siddiqui. "Different Malmivuo, J., & Plonsey, R. (1995). Bioelectromagnetism:
types of electrical signals produced by human principles and applications of bioelectric and biomagnetic
body." International Journal of Advance Research In Science fields. Oxford University Press, USA.
And Engineering 4.02 (2015). L. a. F. X.-A. a. L. Y. Bi, "EEG-based brain-controlled
Berger, Hans. "Über das elektroenkephalogramm des mobile robots: a survey," IEEE transactions on human-
menschen." Archiv für psychiatrie und nervenkrankheiten, machine systems, vol. 43, no. 2, pp. 161-176, 2013.
1929, 87, (1) pp.527-570. Zhang, Lei.:’Real-time feature extraction for multi-channel
Dave, Maitri, et al. "Non Invasive Mindwave Control System EEG signals time-frequency analysis’. 8th Int. IEEE/EMBS
For Effectively Monitoring The Neuroprosthetic
DOI: http://doi.org/10.24086/cocos2022/paper.811
 4th International Conference on Communication Engineering and Computer Science (CIC-COCOS’2022) 143

Conf. on Neural Engineering (NER), Shanghai, China, May

2017, pp. 493-496
Berry, Richard B., and Mary H. Wagner. Sleep medicine
pearls E-book. Elsevier Health Sciences, 2014.
Morshad et al., "Analysis of Brain Wave Data Using
Neurosky Mindwave Mobile II," in Proc. of the Int. Conf. on
Computing Advancements, Dhaka, Bangladesh, 2020, pp.
585-590.
Making Sense of EEG Bands, "NeuroSky," 2015.
[Online].http://neurosky.com/2015/05/greek alphabet-soup-
making-sense-of-eegbands/.[Accessed 20 Dec 2020].
Zhang, Lu, Qingsong Lv, and Yishen Xu. "Single channel
brain-computer interface control system based on TGAM
module." 2017 10th International Congress on Image and
Signal Processing, BioMedical Engineering and Informatics
(CISP-BMEI). IEEE, 2017.
Think Gear measurements (. P. TGAM), "NeuroSky," 2014.
[Online].
Available:http://support.neurosky.com/kb/science/thinkgear
measurements-mindset-protgem. [Accessed 20 Dec 2020].
D. V. Nuys, "Meditation, attention, and hypnotic
susceptibility: A correlational study," International Journal of
Clinical and Experimental Hypnosis, vol. 21, no. 2, pp. 59-69,
1973.
Abdulwahhab, Ali Hussein Abdulwahhab. Improved
algorithm for EEG–based BCI application. MS thesis.
İstanbul Gelişim Üniversitesi Lisansüstü Eğitim Enstitüsü,
2021.

DOI: http://doi.org/10.24086/cocos2022/paper.811