Related Works

A BCI is a direct communication pathway between the brain of a human and an

external device without muscles (hands, fingers and voice). By means of the

technology, humans’ intention can be translated to a control signal to communicate

with external devices, namely directly interaction between neurons and devices [7]. It

is known that BCIs have been commonly used to be an assistive device for the severe

motor disabilities, to execute daily routine as long as their motor cortices were still

intact. More than that, BCIs also have great benefits in the other fields such as

military and entertainment [8]. Nowadays, BCIs have already been seen as an

important technology that certainly brings us significant progress in the future.

Noninvasive electroencephalogram (EEG)-based BCI, is one of the mostly

popular investigated directions, due to it provide high spatial-temple resolutions, ease

of acquisition and more viable. There has been many EEG sources used for BCIs

control, such as event-related synchronization/ desychronization (ERS/ ERD), mu,

alpha and beta rhythm, slow cortical potential (SCP), visual evoked potentials (VEP),

SSVEP and P300 evoked potential [7, 9, 10]. According the previous studies [11-13] ,

they both indicated that the method of SSVEP is an indeed promising way to study

the BCI studies, comparing with others. That is, in this study, we focused on SSVEP

in BCI application.

In recent years considerable concern has arisen over the system based on VEP in

BCI research [11, 12, 14-20]. VEP is an electrical potential fluctuation caused by

visual stimulation. The potential can be recorded via the scalp electrodes placed on

heads, usually generated from the occipital region because of brain functioning

structure. On the basis of stimulation flickering approaches [21], BCIs based on VEP

could be essentially categorized by three groups: time modulated VEP (t-VEP),

frequency modulated VEP (f-VEP) and pseudorandom code modulated VEP (c-VEP).

F-VEP was also called SSVEP because these repetitive stimuli with a certain fixed

frequency provide continuously stable signals to human brain over eyes. So far,

human EEG in response to flickers have been already explored to 100 Hz in 2001 [22]

that was a significant reference to realize mankind’s responses to the SSVEP. The

study did a physiology exploration whereas practical application should be concerned.

Apart from the frequency difference, the contours of stimuli also give human optical

nerve cell different level effects. In [23] has further shown that there have been three

types in SSVEP research: light stimuli, single graphics stimuli, and pattern reversal

stimuli. Different kinds lead to different responses strength as well. The pattern

reversal stimulus seems to evoke stronger SSVEP signals than others.

Because of higher information throughput and less training time than other brain

signals [24, 25], SSVEP-based BCI has become valuable applications. The first

SSVEP-based BCI system was proposed in 1996 [26] though this system had only

one stimulus (13.25 Hz) produced by fluorescent lights. The article also pointed that

human factors would be a huge challenge but provide potential opportunities for more

complex and efficient BCI system. In addition, the different colors were involved in

the stimulation design of SSVEP in 2001 [27]. The study discussed about brain

responses of mixed-colors with two different stimulation frequencies, which still

evoked successful SSVEP. Based on frequency of the stimulus, SSVEP-based BCI

system could also approximately classify into 3 ranges: low frequency (1-12 Hz),

medium frequency (12-30 Hz), and high frequency (30-60 Hz) [23]. Low frequency

and medium frequency systems have predominantly been applied in SSVEP-based

BCI system to date because it was more easily detectable.

Heretofore, there were two kinds of devices that have been commonly used to

rendering signals, LED [28-35] and CRT [11, 12, 18, 36-38]. Furthermore, [39]

indicated that LED was more suitable for highly complicated BCI system (more

targets). In contrary, the number of targets was limited by using a CRT monitor

refresh rate. The refresh rate R means the number of times that the monitor redraws

the screen per second, is usually below 100 Hz (for LCD monitors it is usually 60 Hz),

and only lower than R/2 Hz frequencies can be used as targets [40].

An offline analysis has shown that utilizing the high frequency band could be

very promising [41] even though human cannot perceive stimulus flickering.

Therefore, the high frequency band can be highly expected to be applied in

SSVEP-based BCI system in the future and should definitely and deeply be

researched. It was able to finish some specific tasks on a reliable system [25, 42] for a

decade development.

In practical, system performance is always in considering. The research usually

evaluated performance by two indicators, signal-to-noise ratio (SNR), discussed in

offline analysis and transfer information rate (ITR, bit rate) [43], discussed in

real-time system. Both indicators are closely related that SNR improvement will

enhance detection accuracy, and ITR will improve in the following. Actually, ITR

was an evaluation in telecommunication about messages containing user information

from a source to a sink. BCIs can also be seen as a communication channel between

brains and external devices. As a result, ITR can therefore be regard as a criterion of

evaluation for BCI systems performance. The bit rate B can be express as

1]

N is the number of targets and P is the accuracy of target selections. B multiplied by

selecting speed is the transfer rate (bits per minutes) [12]. For example, there are 4 (1)

targets, one selection is performed every 2 s; the ITR of the online system can reach

41.17 bits/m if the detection accuracy is 90 %.

Figure 1. BCI system based on SSVEP

Figure 1 shows basic architecture of SSVEP-based BCI system. The screen is

divided different blocks. Each block has own frequency and represent a command that

can be decided by system designer. When user gazed on one of these blocks, the

electrode cap would receive the signal. By means of certain detection techniques, BCI

system would know which block the user has been gazing on. Then, the system would

implement suitable action that the user desired to do.