All-in-One BPSK/QPSK Switchable Transmission and Reception for Adaptive Free-Space Optical Communication Links

Abstract

Adaptive free-space optical (FSO) communication links have been extensively studied in order to adapt to variable atmospheric channel environments due to factors such as atmospheric turbulence. As a supporting technology, an all-in-one BPSK/QPSK switchable transmission and reception method is proposed and experimentally demonstrated for adaptive modulation format switching in FSO links. The transmission and reception of both modulation formats are realized based on the same IQ modulator and single-photodetector coherent receiver. Simulation and experimental results show that the QPSK signal has a power penalty of about 3–4 dB compared to the BPSK signal with a BER of about 1 × 10⁻³. The basis for format switching is given according to the various atmospheric channel conditions. The proposed method provides a flexible and efficient solution for variable FSO communication environments to improve their performance.

1. Introduction

Free Space Optical (FSO) communications have attracted considerable attention due to their advantages of high capacity, exceptional confidentiality, fast transmission rates, and long-distance capabilities [1]. FSO communications have been regarded as an attractive alternative technology in many applications, especially in inter-satellite and deep space applications [2]. However, FSO communications are greatly affected by the absorption, scattering, and turbulence effects in the atmospheric channel [3,4]. For example, atmospheric turbulence will cause wavefront distortion, atmospheric scintillation, and beam wander of the optical carrier, leading to serious degradation of signal quality and deterioration of link performance. Reducing the impact of atmospheric turbulence on the FSO transmission has been well studied. In particular, all of these atmospheric effects change in real time, which makes the FSO transmission more challenging.

Several methods have been proposed against atmospheric effects, including coherent reception techniques [5], large-aperture reception [6], diversity reception [7], time-domain processing and threshold optimization [8], coherent modulation techniques [9], and adaptive transmission techniques [10]. Using coherent modulation formats, the system sensitivity can be effectively improved. Moreover, adaptive transmission technology has been demonstrated to be an effective way of adjusting transmission parameters according to real-time channel fading [11]. In recent years, many types of adaptive transmission techniques have been conducted, mostly focusing on power, symbol transmission rate, coding efficiency, modulation format, or a combination of these parameters. These parameters have been proposed for tuning to improve the performance of the FSO link in variable atmospherics [12]. For example, the adaptive format switching between OOK and PPM has been performed by determining the threshold region pattern based on the signal-to-noise ratio (SNR) and turbulence intensity selection [13]. The data rate, modulation scheme, and coding rate for intensity modulation/direct detection (IM/DD) systems have also been proposed for changing under different turbulences [14].

In recent years, it has been demonstrated that phase modulation/coherent detection modulation formats, such as binary phase-shift keying (BPSK) and quadrature phase-shift keying (QPSK), are more competitive than the IM/DD formats including on-off keying (OOK) and pulse position modulation (PPM), etc. [15]. Moreover, the spectral efficiency can be improved by using multi-level phase modulation formats. These multi-level coherent modulation formats are capable of carrying multiple bits of information on each symbol, resulting in improvements in the system’s capacity. However, they require higher optical transmitted power to ensure the same bit error rate (BER) [16]. To adapt to the changing channel environment, it is necessary to switch the modulation formats to transmit as much information as possible without compromising the quality of communication. For example, the transmitter enables the QPSK/16QAM modulation format switching by changing the electrical signals on the IQ modulator between OOK and PAM4 [17]. The switching between OOK and BPSK has been proposed by dynamically controlling the bias point of the Mach-Zehnder modulator (MZM) [18]. The bias voltage algorithm of the IQ modulator has been used to switch OOK, BPSK, and QPSK [19]. The switching between five modulation formats including OOK, BPSK, DPSK, QPSK, and DQPSK has been realized by changing the bias voltage of a certain arm and changing the working mode of MZM [20]. Selecting appropriate modulation formats from BPSK, QPSK, and 8-PSK has been suggested based on the estimated SNR [21]. BER is used as a spectrum maximization metric to adapt MPSK modulations in an FSO link [22].

To support such format switchable functions, a transmitter and a receiver that can deal with multiple kinds of modulation formats simultaneously are urgently demanded. The contributions above mainly focus on compatibility in achieving more modulation formats, and how to transmit all these formats in an all-in-one transmission and reception system has not been well researched. Since phase modulation formats have shown better anti-turbulence performance than intensity modulation formats in FSO links, we mainly focus on phase modulation format switching.

In this paper, we propose and experimentally demonstrate an all-in-one BPSK/QPSK switchable transmission and reception method for real-time variable FSO links to achieve stable communications. In Section 2, principles of the switchable modulation format generation, transmission in atmospheric channels, and link performance evaluation are introduced. In Section 3, the all-in-one BPSK/QPSK switchable transmission and reception system is set up based on an IQ modulator and a single-photodetector (SPD) coherent receiver, which is considered as a competitive coherent receiving technology, as only a photodetector and an analog-to-digital converter are required [23]. The structure is efficiently simplified by processing the two kinds of modulation formats using the same series of devices. In Section 4, switching between 5 Gbit/s BPSK and 10 Gbit/s QPSK signals is demonstrated under moderate turbulence. The transmission performance is measured for both BPSK and QPSK by using a rotating phase plate to simulate the atmospheric environment. BER and constellation diagrams are also simulated based on the phase screen method for the BPSK and QPSK signals. The experimental results agree well with the simulation results. In Section 5, the transmission performance is analyzed for the two kinds of modulation formats under different atmospheric turbulences. The power penalty of the QPSK signal is about 3–4 dB compared to the BPSK signal at a BER of 1 × 10⁻³. The basic criterion for BPSK/QPSK switching is obtained. In Section 6, a conclusion is provided.

2. Operation Principles

2.1. Switchable BPSK/QPSK Generation

The transmitter encodes the data and modulates the electrical signal onto an optical carrier. The optical carrier is often served by a continuous-wave laser, whose field is expressed as $E_{in}(x,y,t)$ and power is $P\left(t\right)={\left|E_{in}(x,y,t)\right|}^2$. The MZM is based on the interference of light and the output optical power of MZM is [24]:

$$\begin{gathered} E_{out}\left(x,y, t\right)=\frac{1}{2}{E}_{in}\left(x,y,t\right)\cdot \left(e^{j{\varphi }_1\left(t\right)}+e^{j{\varphi }_2\left(t\right)}\right) \\ =E_{in}(x,y,t)\cdot \mathrm{c}\mathrm{o}\mathrm{s}\left(\frac{{\varphi }_1\left(t\right)-{\varphi }_2\left(t\right)}{2}\right)\cdot e^{j\frac{\left({\varphi }_1\left(t\right)+{\varphi }_2\left(t\right)\right)}{2}} \end{gathered}$$

(1)

where ${\varphi }_1\left(t\right)$ and ${\varphi }_2\left(t\right)$ are the phase shifts of the upper and lower arms of MZM, respectively, which can be further expressed as:

$${\varphi }_1\left(t\right)=\frac{V_1\left(t\right)}{V_{\pi 1}}\pi$$

(2)

$${\varphi }_2\left(t\right)=\frac{V_2\left(t\right)}{V_{\pi 2}}\pi$$

(3)

where $V_1\left(t\right)$ and $V_2\left(t\right)$ denote the external voltage of the two arms, and $V_{\pi 1}$ and $V_{\pi 2}$ denote the half-wave voltage. Usually, the MZMs operate in push-pull mode, with the voltage applied to the upper and lower arms equal in magnitude but opposite in direction as in ${\varphi }_1\left(t\right)=-{\varphi }_2\left(t\right)=\varphi \left(t\right)$. Then, Equation (1) can be rewritten as:

$$E_{out}\left(x,y,t\right)=E_{in}\left(x,y,t\right)\cdot \cos \left[\varphi \left(t\right)\right]$$

(4)

An IQ modulator often contains two MZMs, which are called I-path and Q-path, operating in push-pull mode and a $90°$ phase shifter is used. Therefore, Q-path has one more phase shift of π/2 than I-path, and the output QPSK signal synthesized by the two paths is:

$$E_{out}\left(x,y,t\right)=\frac{1}{2}{E}_{in}\left(x,y,t\right)\cdot [\cos ({\varphi }_I\left(t\right))+j\mathrm{c}\mathrm{o}\mathrm{s}({\varphi }_Q(t))]$$

(5)

where ${\varphi }_I\left(t\right)$ and ${\varphi }_Q\left(t\right)$ are the phase shift of the two MZMs respectively. Generally, ${\varphi }_I\left(t\right)=0 \mathrm{o}\mathrm{r} \pi$, and ${\varphi }_Q\left(t\right)=0 \mathrm{o}\mathrm{r} \pi$. However, if they are set as ${\varphi }_I\left(t\right)=0 \mathrm{o}\mathrm{r} \pi$, and ${\varphi }_I\left(t\right)\equiv {\varphi }_Q\left(t\right)$, the BPSK format will be generated, which can be regarded as a special case of the QPSK format.

2.2. Atmospheric Channel Model

Since atmospheric turbulence is an irregular random movement, it is difficult to determine the specific atmospheric refractive index value at a certain point and a certain time. The Kolmogorov turbulence model provides the wind speed structure function and refractive index power spectral density, which has been proven to be consistent with a number of experimental conclusions [25].

For long-distance FSO communication systems, the signal light can be approximated as a plane wave. Due to the fluctuation of the refractive index, the phase of the received light will fluctuate after transmitting a distance of L. The phase power spectral density can be expressed as [26]:

$${\Phi }_{\varphi }\left(\left|\mathit{\kappa }\right|\right)=0.49r_0^{-\frac{5}{3}}{\left|\mathit{\kappa }\right|}^{-\frac{11}{3}} ,\frac{1}{L_0}\ll \left|\mathit{\kappa }\right|\ll \frac{1}{l_0}$$

(6)

where $\mathit{\kappa }=\left({\kappa }_x,{\kappa }_y, {\kappa }_z\right)$ denotes the spatial frequency, $L_0$ denotes the outer scale of turbulence, $l_0$ denotes the inner scale of turbulence, and $r_0$ denotes the atmospheric coherent length, which can be expressed as:

$$r_0={\left[0.423k^2{\int }_0^L{C}_n^2\left(z\right)\mathrm{d}z\right]}^{-\frac{3}{5}}$$

(7)

$$C_n^2\left(z\right)=0.813{\sigma }_R^2{k}^{-\frac{7}{6}}{L}^{-\frac{11}{6}}$$

(8)

where $k=2\pi /\lambda$ denotes the wave number, $C_n^2\left(z\right)$ is the refractive index construction constant, ${\sigma }_R^2$ is Rytov variance, and $L$ is the link distance. When the change in phase caused by atmospheric turbulence is small, the phase change along the transmission path can be equivalent to a phase screen [27]. Equation (6) shows a statistical average of the power spectral density. For atmospheric turbulence at a certain moment, the two-dimensional power spectrum of its phase fluctuations is a sample value in the random process, and the mean square value of the random process is ${\Phi }_n$. A series of phase screens of atmospheric turbulence can be obtained by power spectral inversion with Fourier transform as in [28]:

$${\phi }_i(x,y)=\mathrm{I}\mathrm{F}\mathrm{F}\mathrm{T}\left(C_i\cdot \sqrt{{\Phi }_n\left(\left|{\mathit{\kappa }}_{\mathit{x}}+{\mathit{\kappa }}_{\mathit{y}}\right|\right)}\right)$$

(9)

where $i=\mathrm{1,2},3\dots$, and ${\{C}_i\}$ is a two-dimensional complex random matrix series that obeys normal distribution in the frequency domain. The role of the $i$-th phase screens can be expressed as:

$$E_{pout}\left(x,y,t\right)=E_{pin}\left(x,y,t\right)\mathit{exp}\left[j{\phi }_i\left(x,y\right)\right]$$

(10)

where $E_{pout}\left(x,y,t\right)$ is the light field after passing through the phase screen, ${\phi }_i\left(x,y\right)$ is the $i$-th phase modulation due to the phase screen, and $E_{pin}\left(x,y,t\right)$ is the light field incident on the phase screen. For this transmission model, the $E_{pin}\left(x,y,t\right)$ before the first phase screen is the output of transmitter $E_{out}\left(x,y,t\right)$ as in Equation (5), so the role of the first phase screens can be expressed as:

$$E_{pou{t}_1}\left(x,y,t\right)=\frac{1}{2}{E}_{in}\left(x,y,t\right)\left[\cos ({\varphi }_I\left(t\right)\right)+j\mathrm{c}\mathrm{o}\mathrm{s}\left({\varphi }_Q\right(t\left)\right)]\mathit{exp}\left[j{\phi }_1\left(x,y\right)\right]$$

(11)

If the whole transmission path is divided into $\left(n+1\right)$ segments with $n$ phase screens, there will be:

$$E_{pou{t}_n}\left(x,y,t\right)=E_{pou{t}_{n-1}}\left(x,y,t\right)\mathit{exp}\left[j{\phi }_n\left(x,y\right)\right]$$

(12)

2.3. Link Performance Evaluation

In an FSO link using phase modulation, the SNR can be expressed as:

$$SNR=\frac{P_r^2{R}^2}{{\sigma }_n^2}$$

(13)

where ${\sigma }_n^2$ is the variance of the channel noise current, $P_r$ is the received optical power, and $R$ represents the responsivity of the receiver. Because of the atmospheric turbulence, the received power on each micro-element area of the receiving plane fluctuates. Taking the example that the whole transmission path is divided into $\left(n+1\right)$ segments with $n$ phase screens, the received power on the receiving plane can be expressed as:

$$P_r\left(x,y,t\right)={\left|E_{pou{t}_n}\left(x,y,t\right)\right|}^2\cdot L_a$$

(14)

where $L_a$ denotes the link loss due to the atmospheric attenuation. For the first phase screen, $E_{pi{n}_1}\left(x,y\right)$ is uniformly distributed in the plane, so the relationship between it and the transmitted power can be expressed approximately as:

$$E_{pi{n}_1}\left(x,y\right)=\frac{P_t}{A_t}$$

(15)

where $P_t$ is the transmitted optical power, and $A_t$ is the area of the transmitted plane. Generally, bit error rate (BER) is often used to measure the performance of communication systems. The average BER of the BPSK and QPSK signals can be expressed as in Equations (16) and (17), respectively [29]:

$$BE{R}_B=\frac{1}{2}erfc\left(\sqrt{\frac{E_b}{N_0}}\right)$$

(16)

$$BE{R}_Q=erfc\left(\sqrt{\frac{E_b}{N_0}}\right)$$

(17)

where $E_b$ is bit energy, $N_0$ is noise power spectral density, and the ratio $E_b/N_0$ is a normalized version of SNR. So, the average BER of the BPSK signal and QPSK signal can be further written as:

$$BE{R}_B=\frac{1}{2}erfc\left(\sqrt{SNR+10{\log }_{10}\frac{T_{sym}}{T_{samp}}}\right)$$

(18)

$$BE{R}_Q=\frac{1}{2}erfc\left(\sqrt{SNR+10{\log }_{10}\frac{T_{sym}}{T_{samp}}+10{\log }_{10}2}\right)$$

(19)

where $T_{sym}$ is symbol period and $T_{samp}$ is sampling period.

In simulation, the BER can also be obtained as:

$$BER=\frac{number of errorneous bits}{number of total bits}$$

(20)

When a signal light with a certain SNR transmits through a small area of the phase screen, the signal light will be affected by the atmospheric turbulence in this area, as shown in Equation (11). After demodulating and being compared with the transmission data, the BER of the signal light transmitting in this area of the phase screen can be obtained. The same method can be used to obtain the BERs of the signal lights at different positions on the phase screen. The average BER on the phase screen can be on behalf of the BERs of the signal lights transmitting through the whole phase screen. Similarly, for the signal light transmitted through multiple phase screens as in Equation (12), the BER can also be calculated.

With Equation (15) and iteration over $E_{pou{t}_n}\left(x,y\right)$ by Equation (12), the relationship between the BER and the transmitted power $P_t$ can be finally obtained. Thus, under given atmospheric turbulence conditions, the BERs of the BPSK and QPSK modulation formats vary with the transmitted power $P_t$, respectively.

3. Experimental Setup

Figure 1 shows the experimental setup of the BPSK/QPSK switchable transmission and reception in a FSO link. A laser source with a wavelength of 1550 nm is used as the optical carrier, which enters an IQ modulator (FTM7961EX, Fujitsu, Kawasaki, Japan). A 10 Gb/s QPSK signal is generated by uploading two pseudo-random binary sequences (PRBS, I(t) and Q(t)) generated by the error analyzer (N10000A, Keysight, Beijing, China) with lengths of 2¹⁵ − 1 and a rate of 5 Gb/s on the modulator. Moreover, if the electrical signal on one of the two paths is turned off, a 5 Gb/s BPSK signal will be generated. Therefore, the BPSK/QPSK format switching can be realized in this way.

Table 1. Equipment and system parameters.

Equipment	Parameters	Symbols	Values	Units
BPF	bandwidth	B	0.8	nm
PD	responsivity	$R$	0.85	A/W
Rotating phase plate	aperture diameter	NA	3.3	cm
Oscilloscope	sampling rate	S	20	GS/s
Laser	wavelength	λ	1550	nm
Local oscillator	wavelength	λ0	1550.15	nm

shows the equipment and system related parameters in the experiment. The modulated optical signal is injected into a collimator (Collimator 1, F280FC-1550, Thorlabs, Newton, NJ, USA) and expanded through an 8× beam expander. It is sent into a simulated atmospheric channel, which is served by a rotating phase plate. After FSO transmission, the optical signal is collected by another collimator (Collimator 2, F280FC-1550, Thorlabs, Newton, NJ, USA) and coupled into an optical fiber. It is amplified by an EDFA and filtered by a band-pass filter with a bandwidth of 0.8 nm, and then mixed with the local oscillator into a 50:50 power coupler for frequency mixing. An SPD coherent receiver is used to convert the optical signal into an electrical signal, which is then acquired by a real-time oscilloscope (DSOZ594A, Keysight, Beijing, China) at a sampling rate of 20 GS/s. The captured data are then transmitted to a computer and handled by off-lined digital signal process (DSP) algorithms.

In this experiment, the rotating phase plate, as shown in Figure 2a, is used to simulate the atmospheric turbulence. Figure 2b shows the measured probability density function (PDF) of the normalized received power, where a total of 1200 data points are collected at a sampling interval of 0.1 s. Rytov variance ${\sigma }_R^2$ is often used to characterize the intensity of atmospheric turbulence and the degree of refractive index fluctuation [30]. The turbulence level can be classified according to the Rytov variance: ${\sigma }_R^2<<1$ is considered as weak turbulence, ${\sigma }_R^2\approx 1$ is considered as moderate turbulence, and ${\sigma }_R^2>>1$ is considered as strong turbulence. In our experiment, the equivalent Rytov variance ${\sigma }_R^2$ of the rotating phase plate is calculated as 0.4665, and can be considered as moderate turbulence.

4. Experimental and Simulation Results

As shown in Figure 3, the back-to-back (BTB) eye diagrams of the generated optical signals are measured using an optical sampling oscilloscope (86100D, Agilent, Santa Clara, USA), where the signal power is fixed at 3.0 dBm. The eye diagrams of the QPSK (Figure 3a) and BPSK (Figure 3c) signals are clearly observed by tuning the modulated data on the IQ modulator. Their constellation diagrams are shown in Figure 3b,d, where the error vector magnitude (EVM) values are calculated to be 0.4329 and 0.4056, respectively. It is feasible to realize the switching between BPSK and QPSK using the proposed method.

The loss of the FSO link is evaluated in this experiment. When the transmitted optical power after the IQ modulator is 3 dBm, a receiving power of −39.70 dBm is measured. The link loss is evaluated to be 42.70 dB. This loss seems to be heavy, and the reason mainly lies in the fact that the beam is greatly expanded by an 8x beam expander to simulate the beam spreading over a long-distance FSO link and only a small part of the light is collected by the receiver. Moreover, the amplitude and phase are affected by the inserted rotating phase plate, which will reduce the coupling efficiency between free space and the optical fiber.

Figure 4 shows the measured BER curves for the BPSK and QPSK signals. In the experiment, the transmitted optical power is adjusted using a variable optical attenuator (VOA), and the wavelength of the local oscillator light is set to be 1550.15 nm. The insets show the constellation diagrams of the received BPSK and QPSK signals. Compared to the BTB case, it can be seen that the constellation diagrams degrade a little. When the transmitted power is 3 dBm, the EVMs of the BPSK and QPSK signals are 0.5358 and 0.6239, respectively. They are larger than the BTB values, which indicates the distortion in the FSO transmission. If a BER of 1.0 × 10⁻³ is considered, the required transmitted powers of the BPSK and QPSK signals are fitted to be 2.05 dBm and 5.12 dBm according to the measured data. Compared to the BPSK signal, the power penalty is obtained as 3 dB. This means that the BPSK signal has better turbulence flexibility than the QPSK signal.

The transmission performance in the atmospheric channel can be simulated using the phase screen model. According to the experimental conditions, the simulation parameters are set as follows: the incident light carrier is a Gaussian beam with a wavelength of 1550 nm, the radius of the beam waist is 5.0 mm, and the Rytov variance ${\sigma }_R^2$ is 0.4665. The transmission distance is divided into 10 segments and a phase screen is set in each segment. Then, 100 sets of data (each containing 1 × 10⁶ symbols) are recorded and the BER is calculated, as shown in Figure 4. The simulation results agree well with the experimental results. At the BER of 1 × 10⁻³, the simulated light powers are 2.12 dBm and 5.19 dBm for the BPSK and QPSK signals, respectively. The differences between the simulation and experimental results are within 0.07 dB.

5. Determination of the Switching Basis

In our experiment, the transmission performance is measured for the atmospheric condition of the Rytov variance of 0.4665. In order to obtain the switching basis, more BER performances under different atmospheric conditions are required. As shown in Figure 5, the BER curves are simulated as the transmitted power with different values of ${\sigma }_R^2$. Figure 5a,b show the BER curves for the QPSK and BPSK signals under weak turbulences. At the BER of 1 × 10⁻³, the required transmitted power is 3.96 dBm and 0.88 dBm, respectively when the Rytov variance is 0.0199, while these are 4.01 dBm and 0.91 dBm when the Rytov variance increases to 0.0995. Under moderate turbulences shown in Figure 5c,d, the transmitted powers of 4.82 dBm for the QPSK signal and 1.59 dBm for the BPSK signal are required for the Rytov variance of 0.3982. On the other hand, they increase to 7.68 dBm and 4.35 dBm when the Rytov variance increases to 0.9955. In Figure 5e,f for strong turbulences, the transmitted power required for the QPSK and BPSK signals rises to 10.25 dBm and 7.00 dBm (11.55 dBm and 7.50 dBm) for the Rytov variance of 3.9819 (9.9548). As shown in Figure 6, the power penalty of the QPSK signal is about 3–4 dB compared to the BPSK signal, which provides the possibility of improving the comprehensive performance, considering both capacity and quality by switching the modulation format between QPSK and BPSK.

In Figure 6, the required transmitted powers are obtained for the certain BER of 1 × 10⁻³ under different atmospheric turbulences. The simulation results are summarized in Figure 6, where the variation of the BER is plotted against the transmitted power by fixing the BER at 1 × 10⁻³. It can be seen that the required transmitted power for the QPSK signal is higher than that of the BPSK signal. However, the QPSK signal can transmit more data since each symbol carries multiple bits. Therefore, the QPSK signal is preferred for use if the communication performance can be ensured. For a real FSO link, the transmitted power that can be provided by the optical carrier is always limited. Assuming that the maximum permitted transmitted power is 5 dBm, the range of the Rytov variance that can be supported by the QPSK is obtained as 0–0.4624 to ensure the required BER. However, the range is enhanced to 0–1.1858 for the BPSK signal. As a result, when the Rytov variance ranges from 0 to 0.4624, both BPSK and QPSK signals can be selected and the QPSK signal is preferred for improvement of the capacity. When ${\sigma }_R^2$ ranges from $0.4624$ to $1.1858$, only BPSK can meet the BER requirement. When ${\sigma }_R^2$ exceeds $1.1858$, neither QPSK nor BPSK can meet the BER requirement of 1 × 10⁻³ under the given transmitted power.

6. Conclusions

An all-in-one BPSK/QPSK switchable transmission and reception method has been proposed and experimentally demonstrated for adaptive FSO communication links. Using the same series of transmitter and receiver, both the QPSK and BPSK are generated and the transmission performances are measured via a rotating phase plate simulating the atmospheric channel. Experimental and simulation results show that the QPSK signal performance is lower than the BPSK signal and the power penalty is about 3–4 dB at the BER of 1 × 10⁻³. By simulating the required transmitted power to ensure the BER performance, a suitable Rytov variance range is obtained for the BPSK and QPSK signals. Furthermore, the switching criteria is determined as the atmospheric environment varies. In a real BPSK/QPSK switchable FSO transmission system, the atmospheric environment can be real-time monitored to evaluate the channel condition. The required modulation formats will be selected according to the switching criteria. Then, the transmitter and receiver will be switched to the suitable operating mode in order to realize the format transmission. The switching can be real-time realized in the future to keep the link performance stable. Moreover, it is also possible to select higher-order modulation formats to further improve the communication capacity. Our proposed switchable BPSK/QPSK FSO transmission technology may have potentially in real FSO communication systems of the future.