**2. System model**

interface (BCI) system that can be used to detect drowsiness while driving. This system had integrated wireless physiological signal-acquisition and embedded signal-processing modules, and it featured real-time wireless drowsiness detection, long-term daily life EEG monitoring, high computation capacity, and low power consumption. The design and implementation of a Bluetooth-based wearable brain monitoring system was investigated by Sawan et al. [3]. A wireless data recording system was utilized for noninvasive and long-term monitoring of near-infrared spectrometry (NIRS) EEG signals. In addition, the wireless data recording system was applied to the field of invasive cerebral EEG detection. This system has a graphical user interface that is user-friendly and can be used to extract brain activity during dynamic tasks. The advantages of the designed system are portability, wireless connectivity, high throughput, reliable communication, and low-power consumption. Liao et al. [4] proposed a design method for the 16-channel EEG measurement system utilizing dry spring-loaded sensors, a Bluetooth-based acquisition system, and a size-adjustable wearable soft cap. Vos et al. [5] demonstrated an efficient low-cost mobile EEG system that utilizes a P300-based speller for

The development of high-speed, and reliable EEG transmission schemes are interesting research topics. Channel coding is a solution that can be used to achieve a lower error probability for EEG communication. The fundamental design parameters of channel coding are error probability, complexity, and decoding time. Low-density parity-check (LDPC) code is a channel coding technology that was proposed by Gallager [6]. Limpaphayom et al. [7] proposed a power and bandwidth-efficient communication system that utilizes irregular LDPC component codes of block length 100,000, multilevel coding, multistage decoding, 64-quadrature amplitude modulation, and trellis-based signal shaping schemes. The pro-

al. [8] described the concept ofLDPC codes. The regular (v, c) LDPC codes are linear block codes with a sparse parity-check matrix H, In H, the number of nonzero elements in the columns is v, while the number of nonzero elements in the rows is c. In addition, the code rate is defined as 1 − v/c. In an (N,K) LDPC code, the block length is N and the information length is K. Ohtsuki [9] applied LDPC codes to various transmission systems with excellent performance, and illustrated some of the LDPC code designs. LDPC codes are included in the second-generation specification for satellite broadband applications, and in

Filter bank-based multicarrier modulations (FBMC) is an interesting research topic, and it is being considered as a potential candidate for the fifth generation (5G) mobile systems. FMBC is a modified version of orthogonal frequency division multiplexing (OFDM), and a tutorial review of FBMC modulations was discussed by Boroujeny et al. [10]. Compared to cyclic prefix (CP)-based OFDM modulation, FBMC offers better spectral efficiency in multipath channels. Bouhadda et al. investigated the BER performance of nonlinear distortion in high-power amplifiers for FBMC using offset quadrature amplitude modulation (OQAM) [11]. Caus et al. [12] studied the effects of multi-tap filtering on FBMC/OQAM systems to combat intersymbol and inter-carrier interferences due to multipath fading. Caus et al. [12] proposed a low-complexity transmission power estimation method, and their simulation results show that the proposed transmission power estimation method is excellent. Further, Bellanger et al.

[13] proposed an FBMC-based physical layer solution for 5G mobile systems.

at an Eb/No of 6.55 dB. Franceschini et

posed system achieved a bit-error rate (BER) of 10− <sup>5</sup>

wireless BCI.

56 Electroencephalography

IEEE 802.16e.

The proposed FBMC-based EEG transmission scheme is shown in **Figure 1**. The LDPC channel coding, FBMC transmission method, a BPSK or OQAM adaptive modulation, and a power assignment mechanism were used in a new design strategy to achieve low power, high speed, and high quality of service transmission capabilities for wireless EEG systems. We performed a simulation using the irregular LDPC codes set with a block length of 2000 and a rate of 1/2. The digital bit streams of the EEG signal were inputted to the irregular LDPC encoder, and the LDPC bit streams were outputted. The OQAM modulation scheme is used at the lower fading channel in order to achieve high-speed wireless EEG transmission. A BPSK is used at the highest fading channel, to achieve robust EEG communication. The LDPC bit streams were inputted to an OQAM-based or BPSK-based adaptive modulator, and the LDPC bit streams with adaptive modulation were outputted. The LDPC bit streams with adaptive modulation were inputted to the serial to parallel converter, and the 64 parallel LDPC adaptive modulation EEG bit streams were outputted. The 64 parallel LDPC adaptive modulation EEG bit streams were inputted to the digital filter banks (DFBs). The coefficients of the DFBs were determined to be as follows [13].

**Figure 1.** The proposed FBMC-based EEG transmission scheme.

$$\mathbf{H}\_0 = 1, \mathbf{H}\_1 = 0.971960, \mathbf{H}\_2 = \sqrt{2}/2, \mathbf{H}\_3 = 0.235147\tag{1}$$

Further, the 64 DFB output bit streams were inputted to a 64-point inverse fast Fourier transform (IFFT), and the 64 parallel LDPC adaptive modulation FBMC EEG bit streams were outputted. The 64 parallel LDPC adaptive modulation FBMC EEG bit streams were inputted to a parallel to serial converter, and the serial LDPC adaptive modulation FBMC EEG bit streams were output. The proposed power assignment algorithm is summarized below:

Step 1: On the basis of the output information obtained from the object-component Petri Nets (OCPN) model, determine throughputs for EEG packets transmission.

Step 2: Select the appropriate modulation mode to satisfy the requirements for transmission over a wireless EEG transmission system.

Step 3: Assign the initial value of power weighting as 15/30 for EEG packets.

Step 4: Measure the received signal to noise ratio (SNR) for EEG packets.

Step 5: If the measured SNR of the received signal exceeds the threshold SNR at which the required BER for EEG packets is achieved, then update the power weighting as power weighting = power weighting − 1/30; If power weighting ≥1/30, go to Step 4; otherwise, go to Step 7.

Step 6: If the measured SNR of the received signal is less than the threshold SNR at which the required BER for EEG packets is achieved, then update the power weighting as power weighting = power weighting + 1/30; If power weighting ≤1, go to Step 4; otherwise, go to Step 8.

Step 7: Change the modulation mode. If the modulation mode is not OQAM, go to Step 4.

Step 8: Increase the modulation mode. If the modulation mode is not BPSK, go to Step 4.

The carrier sense multiple access with collision avoidance (CSMA/CA) technology were used for multiuser wireless EEG transmission system.
