Experimental demonstration of a two-path parallel scheme for m-QAM-OFDM transmission through a turbulent-air-water channel in optical wireless communications

Lu Zhang; Han Wang; Xu Zhao; Fang Lu; Xiaoming Zhao; Xiaopeng Shao

doi:10.1364/OE.27.006672

1. Introduction

In recent years, due to the broad unregulated bandwidth and insensitivity to electro-magnetic waves, the optical wireless communications (OWC) technology has obtained increasing attention. And it was extensively researched. Broadly speaking, the scope of OWC covers visible light communications (VLC), free-space optical communications (FSO), and underwater optical wireless communications (UOWC). For point-to-point communications, although the transmitter, channel, receiver and application scenarios are different for these technology to some extent, these technology has the same goal, which is providing higher data rates, occupying lower bandwidth and transmitting longer distance. Unlike fiber optical communications, to achieve the higher spectral efficiency (SE) and longer transmission distance, the scheme of high order modulation formats together with coherent detection is not a better choice for OWC considering available devices, cost and complexity et. al. In OWC, for the linear and multi-path channel, the scheme of OFDM together with intensity-modulation and direct detection (IM/DD) seems more popular. For example, in VLC, Chien-Ju Chen et. al. [1] achieved 2.16 Gb/s data rates transmission with an OFDM signal modulated green LED. Bernhard Schrenk and Christoph Pacher [2] reported an interesting all-led VLC system with 1-Gb/s data rates using OFDM. Guowu Zhang et. al [3] proposed a low-complexity frequency domain nonlinear compensation for OFDM based VLC systems with LED. In UOWC, Hassan M. Oubei et. al [4] reported a 4.8-Gbit/s 16-QAM-OFDM transmission based on compact 450-nm laser. Yifei Chen et. al [5] experimentally demonstrate a high-speed air-water transmission employing 32-QAM-OFDM modulated 520-nm laser diode. Chao Fei et. al. [6] experimentally demonstrated a 15-m 7.33-Gb/s OFDM transmission using post nonlinear equalization.

To improve the performance of OFDM in OWC, some research focused on compensating light source's imperfection, like LED's limited modulation bandwidth or nonlinearity [7]. Chen Chen et. al. [8] employed an adaptive digital pre-frequency domain equalization technique to optimize the modulation bandwidth of the OFDM signal. Xingyu Lu et. al. [9] used machine learning based pre-distortion to compensate the non-linear effects of LED devices. Yingjun Zhou et. al. [10] applied power exponential software pre-equalization to achieve 2.08-Gbit/s VLC transmission over 1m free-space distance. Mitigation of clipping noise [11,12] and peak-to-average power ratio (PAPR) [13–16] of OFDM signal is another way to improve the performance. Chao Li et. al. [11] used decision-aided reconstruction to reduce the clipping noise for an OFDM-based UOWC system, and 2 dB SNR gain is achieved in their experiments. A tone reservation based scheme was proposed by Jurong Bai et. al. [14] to reduce the PAPR for DCO-OFDM indoor VLC. Remarkable PAPR reduction efficiency was observed through simulation. A PAPR reduction technique combining orthogonal circulant matrix transform precoding with peak-clipping was reported by Siyi Dong et. al. [15], 3.02-Gbit/s ACO-OFDM with bit error rate of $5 \times 10^{- 5}$ was achieved. Besides device's imperfection, clipping noise and PAPR, intercarrier interference (ICI) can also cause large performance penalty for OFDM. In fiber optical communications, ICI reduction technique has been researched widely [17–20] for coherent optical OFDM transmission, where laser phase noise results in ICI dominantly. Even for IM/DD-OFDM in fiber optical communications, laser phase noise can also lead to ICI and performance impairment [21–24]. However, for OFDM in OWC, because IM/DD scheme is adopted, ICI brought by laser phase noise can be ignored. M. R. H. Mondal et. al [25] analyzed the effect of ICI caused by vignetting for pixelated multiple-input multiple-output OWC systems using spatial OFDM. In OWC, the ICI primarily arises from synchronization error caused by frequency offset, phase offset or timing error. We have ever proposed an OFDM symbol frame structure to mitigate the ICI for m-QAM-FODM transmission in UOWC, and remarkable performance improvement was achieved in the experiments [26]. In this paper, a two-path parallel scheme for m-QAM-FODM transmission in OWC was proposed, and the principle of this scheme was derived theoretically. The feasibility was demonstrated by a transmission experiment based on a turbulent air-water channel. Significant performance improvement was observed in the experiment. The proposed scheme can effectively extend the transmission distance while high spectra efficiency (SE) is kept. It can be widely applied on the scenarios including FSO communications, VLC and UOWC.

2. Principle

In the proposed scheme, two-path parallel baseband OFDM signals were generated at the transmitter end. On the first path, the baseband OFDM signal was given as

x_{k} = \sum_{n = 0}^{N - 1} d_{n} e^{j \frac{2 π}{N} n k}, k = 0, 1, 2, ..., N - 1,

where

d_{n}

was the data symbol mapped by m-QAM or m-PSK format.

N

was the IFFT size, and

k

was the subcarrier index. After transmitting over wireless channel, the received signal at the receiver end was

r_{k} = x_{k} \cdot e^{j ϕ_{k}} + w_{k},

where

ϕ_{k}

represented the phase noise caused by synchronization error brought by frequency offset, phase offset or timing error.

w_{k}

is the AWGN (additive white Gaussian noise). At the receiver, the demodulated sequence in the frequency domain after FFT was given by

{\hat{d}}_{n} = \frac{1}{N} \sum_{k = 0}^{N - 1} r_{k} e^{- j \frac{2 π}{N} n k}, n = 0, 1, 2, ..., N - 1.,

Without loss of generality, we ignored the AWGN, then there is

\begin{matrix} {\hat{d}}_{n} = \frac{1}{N} \sum_{k = 0}^{N - 1} (x_{k} \cdot e^{j ϕ_{k}}) e^{- j \frac{2 π}{N} n k} \\ = \frac{1}{N} \sum_{k = 0}^{N - 1} (\sum_{m = 0}^{N - 1} d_{m} e^{j \frac{2 π}{N} m k}) \cdot e^{j ϕ_{k}} \cdot e^{- j \frac{2 π}{N} n k} \\ = \frac{1}{N} \sum_{m = 0}^{N - 1} \sum_{k = 0}^{N - 1} d_{m} e^{j \frac{2 π}{N} m k} \cdot e^{- j \frac{2 π}{N} n k} \cdot e^{j ϕ_{k}} \\ = \frac{1}{N} \sum_{m = 0}^{N - 1} \sum_{k = 0}^{N - 1} d_{m} e^{j \frac{2 π}{N} m k} \cdot e^{- j \frac{2 π}{N} n k} \cdot e^{j \frac{2 π}{N} \cdot \frac{N}{2 π k} ϕ_{k} \cdot k}, \begin{matrix}  \end{matrix} n = 0, 1, 2, ..., N - 1, \\ = \frac{1}{N} \sum_{m = 0}^{N - 1} \sum_{k = 0}^{N - 1} d_{m} e^{j \frac{2 π}{N} (m - n + \frac{N}{2 π k} ϕ_{k}) k} \\ = d_{n} (\frac{1}{N} \sum_{k = 0}^{N - 1} e^{j ϕ_{k}}) + \frac{1}{N} \sum_{\begin{array}{l} m = 0 \\ m \neq n \end{array}}^{N - 1} \sum_{k = 0}^{N - 1} d_{m} e^{j \frac{2 π}{N} (m - n + \frac{N}{2 π k} ϕ_{k}) k} \\ = d_{n} \cdot ψ (0) + ψ (m - n) \end{matrix}

where

ψ (0) = \frac{1}{N} \sum_{k = 0}^{N - 1} e^{j ϕ_{k}},

ψ (m - n) = \frac{1}{N} \sum_{\begin{array}{l} m = 0 \\ m \neq n \end{array}}^{N - 1} \sum_{k = 0}^{N - 1} d_{m} e^{j \frac{2 π}{N} (m - n + \frac{N}{2 π k} ϕ_{k}) k}

In Eq. (4), the first term is the desirable data symbol

d_{n}

with weighting factor

ψ (0)

, and the second term

ψ (m - n)

represents the crosstalk from the undesired data symbols. Because of the phase noise

ϕ_{k}

, the carriers' orthogonality was lost.

On the second path, the baseband OFDM signal was written as

{x^{'}}_{k} = \sum_{n = 0}^{N - 1} d_{n}^{*} e^{j \frac{2 π}{N} n k}, k = 0, 1, 2, ..., N - 1,

where

d_{n}

was the data symbol on the first path, and

d_{n}^{*}

was the complex conjugation of

d_{n}

.

{(\cdot)}^{*}

represented the complex conjugation operation.

N

was the IFFT size, and

k

was the subcarrier index. After propagating through wireless channel, the received signal can be represented as

{r^{'}}_{k} = {x^{'}}_{k} \cdot e^{j ϕ_{k}} + {w^{'}}_{k}

where

ϕ_{k}

also represented the phase noise induced in the transmission. Moreover, we assumed that the second path signal

{x^{'}}_{k}

experienced the same phase noise as the first path signal.

{w^{'}}_{k}

was the AWGN. Similarly, at the receiver, the demodulated sequence in the frequency domain after FFT can be written as

{\hat{d}}^{'}_{n} = \frac{1}{N} \sum_{k = 0}^{N - 1} {r^{'}}_{k} e^{- j \frac{2 π}{N} n k}, n = 0, 1, 2, ..., N - 1.

Without loss of generality, the AWGN

{w^{'}}_{k}

was also ignored, then yielded

\begin{matrix} {\hat{d}}^{'}_{n} = \frac{1}{N} \sum_{k = 0}^{N - 1} ({x^{'}}_{k} \cdot e^{j ϕ_{k}}) e^{- j \frac{2 π}{N} n k} \\ = \frac{1}{N} \sum_{k = 0}^{N - 1} (\sum_{m = 0}^{N - 1} d_{m}^{*} e^{j \frac{2 π}{N} m k}) \cdot e^{j ϕ_{k}} \cdot e^{- j \frac{2 π}{N} n k} \\ = \frac{1}{N} \sum_{m = 0}^{N - 1} \sum_{k = 0}^{N - 1} d_{m}^{*} e^{j \frac{2 π}{N} m k} \cdot e^{- j \frac{2 π}{N} n k} \cdot e^{j ϕ_{k}} \\ = \frac{1}{N} \sum_{m = 0}^{N - 1} \sum_{k = 0}^{N - 1} d_{m}^{*} e^{j \frac{2 π}{N} m k} \cdot e^{- j \frac{2 π}{N} n k} \cdot e^{j \frac{2 π}{N} \cdot \frac{N}{2 π k} ϕ_{k} \cdot k}, \begin{matrix}  \end{matrix} n = 0, 1, 2, ..., N - 1. \\ = \frac{1}{N} \sum_{m = 0}^{N - 1} \sum_{k = 0}^{N - 1} d_{m}^{*} e^{j \frac{2 π}{N} (m - n + \frac{N}{2 π k} ϕ_{k}) k} \\ = d_{n}^{*} (\frac{1}{N} \sum_{k = 0}^{N - 1} e^{j ϕ_{k}}) + \frac{1}{N} \sum_{\begin{array}{l} m = 0 \\ m \neq n \end{array}}^{N - 1} \sum_{k = 0}^{N - 1} d_{m}^{*} e^{j \frac{2 π}{N} (m - n + \frac{N}{2 π k} ϕ_{k}) k} \end{matrix}

The complex conjugation of

{\hat{d}}^{'}_{n}

was

{({\hat{d}}^{'}_{n})}^{*} = d_{n} (\frac{1}{N} \sum_{k = 0}^{N - 1} e^{- j ϕ_{k}}) + \frac{1}{N} \sum_{\begin{array}{l} m = 0 \\ m \neq n \end{array}}^{N - 1} \sum_{k = 0}^{N - 1} d_{n} e^{- j \frac{2 π}{N} (m - n + \frac{N}{2 π k} ϕ_{k}) k}

After simple superimposition of the sequence

{\hat{d}}_{n}

and

{({\hat{d}}^{'}_{n})}^{*}

,a new sequence

{\tilde{d}}_{n}

was obtained as

\begin{matrix} {\tilde{d}}_{n} = \frac{{\hat{d}}_{n} + {({\hat{d}}^{'}_{n})}^{*}}{2}, \\ = d_{n} {\frac{1}{N} \sum_{k = 0}^{N - 1} (\frac{e^{j ϕ_{k}} + e^{- j ϕ_{k}}}{2})} + \frac{1}{N} \sum_{\begin{array}{l} m = 0 \\ m \neq n \end{array}}^{N - 1} \sum_{k = 0}^{N - 1} d_{m} (\frac{e^{j \frac{2 π}{N} (m - n + \frac{N}{2 π k} ϕ_{k}) k} + e^{- j \frac{2 π}{N} (m - n + \frac{N}{2 π k} ϕ_{k}) k}}{2}), \\ = d_{n} {\frac{1}{N} \sum_{k = 0}^{N - 1} \cos (ϕ_{k})} + \sum_{\begin{array}{l} m = 0 \\ m \neq n \end{array}}^{N - 1} d_{m} {\frac{1}{N} \sum_{k = 0}^{N - 1} \cos (\frac{2 π}{N} (m - n + \frac{N}{2 π k} ϕ_{k}) k)} \\ = d_{n} \cdot ψ^{'} (0) + ψ^{'} (m - n) \end{matrix}

where

ψ^{'} (0) = \frac{1}{N} \sum_{k = 0}^{N - 1} \cos (ϕ_{k}),

ψ^{'} (m - n) = \sum_{\begin{array}{l} m = 0 \\ m \neq n \end{array}}^{N - 1} d_{m} {\frac{1}{N} \sum_{k = 0}^{N - 1} \cos (\frac{2 π}{N} (m - n + \frac{N}{2 π k} ϕ_{k}) k)}

By comparing

ψ^{'} (0)

in Eq. (13) with

ψ (0)

in Eq. (5), it can be seen that the exponential term

\exp (j \cdot ϕ_{k})

in

ψ (0)

becomes

\cos (ϕ_{k})

in

ψ^{'} (0)

, the phase

ϕ_{k}

is eliminated, so the phase noise is reduced. And because

| \cos (ϕ_{k}) | \leq 1

, the weighting factor of

d_{n}

is also reduced. Similarly, comparing

ψ^{'} (m - n)

with

ψ (m - n)

, the exponential term

\exp (j \cdot \frac{2 π}{N} (m - n + \frac{N}{2 π k} ϕ_{k}) k)

is reduced to

\cos (\frac{2 π}{N} (m - n + \frac{N}{2 π k} ϕ_{k}) k)

, the crosstalk from the undesired data symbol is mitigated. Because of the mitigation of the crosstalk and weighting factor, the performance of the proposed two-path scheme can be improved significantly. Based on the principle above, the architecture of the transmitter and receiver was proposed and presented in Fig. 1.

Fig. 1 The proposed architecture of baseband OFDM (a) transmitter and (b) receiver.

Download Full Size | PDF

In Fig. 1(a), the binary information sequence is mapped to data symbol sequence $d_{n}$ using m-QAM or m-PSK format after serial-to-parallel (S/P) conversion. The sequence $d_{n}^{*}$ is generated by applying complex conjugation operation to $d_{n}$ . The two-path sequences $d_{n}$ and $d_{n}^{*}$ , together with their corresponding Hermitian symmetrical sequences, were transformed to two real sequence by N-points IFFT, respectively. For the two-path real sequences, the training sequence is inserted for the purpose of synchronization and channel estimation. Then a cyclic prefix (CP) is added to avoid inter-symbol interference. After parallel-to-serial (P/S) conversion, the two-path baseband sequences $x_{k}$ and ${x^{'}}_{k}$ are generated.

In Fig. 1(b), the received two-path baseband sequences $r_{k}$ and ${r^{'}}_{k}$ are synchronized first. Then after S/P conversion, the CP is removed. Next, N-points FFT and channel equalization are performed to generate sequence ${\hat{d}}_{n}$ and ${\hat{d}}^{'}_{n}$ . Applying complex conjugation operation to ${\hat{d}}^{'}_{n}$ , ${({\hat{d}}^{'}_{n})}^{*}$ is yielded. Using the superimposition in Eq. (12), the compensated sequence ${\tilde{d}}_{n}$ is generated. The binary information can be recovered by demodulating ${\tilde{d}}_{n}$ using m-QAM or m-PSK format.

3. Experimental setup

Experiments were carried out to investigate the performance of the proposed scheme. The block diagram and the photo of the experimental setup were shown in Fig. 2 and Fig. 3, respectively. In the experiments, the generation and recovery of two-path baseband OFDM sequences were both by off-line digital signal processing (DSP) using Matlab. At the transmitter end, a pseudorandom binary sequence (PRBS 2¹⁵-1) was mapped into m-QAM symbols $d_{n}$ after serial-to-parallel (S/P) conversion. The complex conjugation operation was performed to obtain ${(d_{n})}^{*}$ . For the two-path symbols sequences $d_{n}$ and ${(d_{n})}^{*}$ , after the operation of Hermitian symmetry, they were modulated onto 100 subcarriers by inverse FFT (IFFT), respectively. The size of the IFFT is 256. Then training sequence (TS) was inserted for synchronization and channel estimation. Cyclic prefix (CP) was added to overcome inter-symbol interference. After parallel-to-serial (P/S) conversion, the two-path baseband OFDM sequences were generated. We tabulated the detailed parameters of the baseband OFDM sequence in Table 1. The two-path baseband OFDM sequences were uploaded to an arbitrary waveform generator (AWG, Agilent 81180A) . The sampling rate of AWG were set to 1-GSa/s, so the bandwidth of the baseband signal was $1 \times (6 + 50) / 256 \approx 218.75$ -MHz. The two-path baseband electrical signals were output by AWG's two channels, and the peak-to-peak voltage amplitude was set to 2-V. Next, the two-path baseband electrical signals from AWG passed through two Bias-Tees (PE1611), where the DC bias current from laser driver was added. At last the biased two-path electrical signals directly modulated a blue-light laser diode (LD) and a green-light LD, respectively. The optical spectra at the bias current of 65-mA was measured by an Andor SR-500i spectrometer and shown in Fig. 4(a). The central wavelength of these two LDs is about 450-nm and 517.5-nm. The curves of optical power versus current for these two LDs were given in Fig. 4(b). Because the bias current of these two LDs were from a same laser driver, a variable resistor was in series with the blue-light LD to balance the current obtained. At last the bias current for the blue-light LD and green-light LD was optimized to 70-mA, and 65-mA, respectively. The corresponding output power was measured as 42-mW and 26-mW, respectively.

Fig. 2 Experimental setup of m-QAM-OFDM transmission over a turbulent-air-water channel based on the proposed scheme. Laser diode (LD), mirror (M1, M2), avalanche photodiode (APD).

Download Full Size | PDF

Fig. 3 Photo of the experimental setup (a) devices and equipments used at the transmitter end (b) green-light and blue light in transmission and devices used at the receiver end (c) atmospheric turbulence simulator with heater and fans and water tank filled with tap-water.

Download Full Size | PDF

Table 1. Parameters of OFDM

View Table

Fig. 4 (a) Optical spectra and (b) optical power versus bias current of blue-light LD and green-light LD.

Download Full Size | PDF

After collimation by lens, the blue light and green light were launched into an atmospheric turbulence simulator with length of 2.5-m first. The simulator can generate turbulent air flow by heater and fans. The two-path light was reflected by 4 mirrors in the simulator, respectively. The equivalent transmission distance was about 7.5-m. If the heater and fans were enabled, the simulator can provide a 7.5-m turbulent air channel, otherwise it just provided a 7.5-m free-space air channel without turbulence. The direction of the two outgoing light from the atmospheric turbulence simulator was then changed by two mirrors M1 and M2, and they got into a glass water tank. The length of the water tank was 1.6-m. The water tank was filled with tap-water, which has an estimated attenuation coefficient of 0.23-m⁻¹ [27]. The blue and green light were reflected by 4 mirrors to extend the propagation distance to 8-m. The propagation distance was variable when the number of the reflection was varied. After passing through the water tank, the output blue and green light were focused into two APDs, respectively. One APD is APD210 produced by Menlo Systems, its bandwidth was about 1-GHz and the active diameter was 0.5-mm. The other APD was APD430A2 from Thorlabs, its bandwidth was 400-MHz and the active diameter was 0.2 mm.

The two-path electrical signals from APDs were sampled by a digital oscilloscope (Teledyne Lecroy WavePro 254HD-MS), which had 2.5-GHz bandwidth, 20-GSa/s sampling rate and 12-bits resolution. The captured samples were then processed by offline DSP. In the offline processing, synchronization was first performed to identify the frame's header. Next, S/P conversion and CP removal followed. After that, the a 256 points fast Fourier transformation (FFT) was carried out to transform the time domain signal to frequency domain signal. The channel estimation came after FFT. The channel multiplication coefficients were calculated and used for compensation of each subcarrier's linear distortion. Right after channel equalization, the sequence ${\hat{d}}_{n}$ and ${\hat{d}}^{'}_{n}$ were obtained. Then noise suppression was achieved using Eq. (12). The binary sequence was recovered after m-QAM demodulation. BER obtained by counting the error bits directly was used to evaluate the performance.

4. Results and discussions

To verify the performance of the proposed scheme, first we disabled the heater and fans in the atmospheric turbulence simulator, so the channel became a non-turbulent-air-water channel. The number of refection in water tank was varied to gain different propagation distance. We transmitted the two-path parallel baseband OFDM signals synchronously through the channel by letting the two channels of AWG share a common transmission clock. The baseband signal of path-A in Fig. 1(a) modulated the green-light LD. The baseband signal of path-B in Fig. 1(a) modulated the blue-light LD. The performance of the proposed scheme was compared with the performance of the conventional OFDM signal from path-A. At the receiver end, the received two-path electrical signals after APDs were captured by two synchronous channels of digital oscilloscope. The transmitter and receiver were asynchronous, because the clocks of AWG and oscilloscope were independent. When the propagation distance in water tank was 1.6-m, the captured waveform with 64-QAM format was shown in Fig. 5. It can be seen that the two-path signals aligned well. The corresponding frequency spectra of the waveform in Fig. 5 was shown in Fig. 6.

Fig. 5 The captured waveform when two-path OFDM signals were transmitted synchronously.

Download Full Size | PDF

Fig. 6 The corresponding frequency spectra of the captured waveform in Fig. 5.

Download Full Size | PDF

When the propagation distance in water tank was varied from 1.6-m to 8-m, the BER curves for 16-QAM and 64-QAM formats were obtained. They were shown in Fig. 7. In Fig. 7, the curves of '16QAM' and '64QAM' were the results of conventional OFDM signal from path-A in Fig. 1(a). The curves of '16QAM-Parallel' and '64QAM-Parallel' were the results of the proposed two-path parallel scheme. Seen from Fig. 7, significant improvement of BER performance was brought by the proposed two-path parallel scheme, which completely outperformed the conventional single-path scheme. The BER of '16QAM-Parallel' and '64QAM-Parallel' was far lower than 7% FEC limit of $3.8 \times 10^{- 3}$ . And the BER of '64QAM-Parallel' was even lower than that of '16QAM' with the proposed two-path scheme.

Fig. 7 Curves of BER versus transmission distance over a non-turbulent-air-water channel when two-path OFDM signals were transmitted synchronously.

Download Full Size | PDF

When we configured the two channels of AWG to use separate transmission clock, which meant the two-path parallel signals were transmitted asynchronously, the second experiment was carried out. The channel in this experiment was still non-turbulent-air-water channel. And the transmitter and receiver were still asynchronous. The captured waveform with 64-QAM format was shown in Fig. 8. There was about 1.9-us time delay between the signals of path-A and path-B. When the transmission distance was varied, the curves of BER versus distance were obtained and displayed in Fig. 9. Similarly, significant performance can still be observed. The BER of '16QAM-Parallel' and '64QAM-Parallel' was also far lower than 7% FEC limit of $3.8 \times 10^{- 3}$ . Comparing '16QAM-Parallel' and '64QAM-Parallel' in Fig. 9 with '16QAM-Parallel' and '64QAM-Parallel' in Fig. 7, it can be seen that the difference between each curve was tiny, hence it can be concluded that the proposed two-path parallel scheme was not sensitive to the time delay between the two paths. The proposed two-path parallel scheme worked well whatever the two-path signals were transmitted synchronously or asynchronously.

Fig. 8 The captured waveform when two-path OFDM signals were transmitted asynchronously.

Download Full Size | PDF

Fig. 9 Curves of BER versus transmission distance over a non-turbulent-air-water channel when two-path OFDM signals were transmitted asynchronously.

Download Full Size | PDF

To further verify the performance of the proposed scheme, we enabled the heater and fans in the atmospheric turbulence simulator, then the channel became a turbulent-air-water channel. The temperature of atmospheric turbulence simulator was set to maximum value, 50°C。Because the low frequency component was dominant in the power spectra density (PSD) of atmospheric turbulence [28,29], we used two Si amplified detectors (Thorlabs' PDA100A-EC), whose bandwidth was DC to 2.4-MHz, to measure the fluctuated signals of the two paths first. We transmitted the two-path parallel light with constant optical power into the channel. The output power of blue-light LD and green-light LD was 42-mW and 26-mW, respectively. Because the water channel here just resulted in attenuation of light, it didn't contribute to signal's fluctuation, so the length of water channel was just configured to 1.6-m. The captured fluctuated waveform of blue light was presented in Fig. 10(a). Figure 10(b) was the statistical histogram of the fluctuated waveform in Fig. 10(a), and the fitted curve was obtained by fitting the histogram based on lognormal distribution. According to the fitted curve, the mean of the waveform in Fig. 10(a) was 0.0286-V, the variance $σ^{2} = 0.00536$ , and the scintillation index was $S I = 0.00535$ , which means the generated turbulence was weak for the blue light. The power spectra density (PSD) of the waveform in Fig. 10(a) was given in Fig. 10(c). Seen from the PSD, most of the power concentrated below 20-Hz, so it can also be concluded that the generated turbulence was weak. The results of green light was shown in Figs. 11(a)–11(c). Figure 11(a) was the captured waveform, and Fig. 11(b) showed the histogram and fitted curve. According to the fitted results, the mean was 0.0274-V, the variance was $σ^{2} = 0.00495$ , the scintillation index was $S I = 0.00497$ . Seen from PSD in Fig. 11(c), the power also primarily concentrated below 20-Hz. Accordingly, the generated turbulence was also weak for green light.

Fig. 10 (a) The fluctuated signal captured at the receiver end when the constant optical power from blue-light LD propagated through the turbulent-air-water channel. (b) histogram of the signal in (a) and the fitted curve of the histogram. (c) power spectra density of the signal in (a).

Download Full Size | PDF

Fig. 11 (a) The fluctuated signal captured at the receiver end when the constant optical power from green-light LD propagated through the turbulent-air-water channel. (b) histogram of the signal in (a) and the fitted curve of the histogram. (c) power spectra density of the signal in (a).

Download Full Size | PDF

Based on the generated weak turbulence and the water tank, the turbulent-air-water channel was available. By sharing a common clock, two channels of the AWG synchronously transmitted the two-path parallel baseband OFDM signals into the turbulent-air-water channel. When the number of reflection in water tank was varied, the curves of BER versus transmission distance were acquired and shown in Fig. 12. In Fig. 12, like the results in Fig. 7, '16QAM-Parallel' and '64QAM-Parallel' were superior to '16QAM' and '64QAM', the performance improvement was remarkable.

Fig. 12 Curves of BER versus transmission distance over a turbulent-air-water channel when two-path OFDM signals were transmitted synchronously.

Download Full Size | PDF

When the common clock was changed to separate clock, the two-path parallel baseband OFDM signals were asynchronously transmitted into the turbulent-air-water channel. The time delay between path-A and path-B was about 3.7-us at this time. The BER results were shown in Fig. 13. Similar trend as in Fig. 12 was seen. The superiority of '16QAM-Parallel' and '64QAM-Parallel' remained, which further proved that the proposed two-path parallel scheme was immune to the time delay between the two paths. When the length of water channel was 8-m, the recovered constellations of '16QAM', '16QAM-Parallel', '64QAM' and '64QAM-Parallel' were shown in Fig. 14. The improvement of constellations was obvious when comparing Figs. 14(a) and 14(b) with 14(c) and 14(d), respectively.

Fig. 13 Curves of BER versus transmission distance over a turbulent-air-water channel when two-path OFDM signals were transmitted asynchronously.

Download Full Size | PDF

Fig. 14 Recovered constellations of 16-QAM and 64-QAM for conventional single-path OFDM signals ((a), (b)) and the proposed two-path scheme ((c), (d)) when the length of water channel was 8-m and atmospheric turbulence simulator was enabled.

Download Full Size | PDF

To make more performance comparison, we conducted another two experiments where we transmitted the two paths of conventional OFDM signals with 8-QAM constellation, which has nearly the same spectral efficiency (SE) as the two-path parallel transmission scheme with 64-QAM constellation. In the first experiment, sampling timing error were introduced by configuring the arbitrary waveform generator (AWG) at the transmitter end and the digital oscilloscope (DSO) at the receiver end running independently, and no timing error compensation was used. The BER was measured. Then in the second experiment, the sampling timing error were removed by synchronizing the AWG and the DSO using a common clock generator. The BER was measured again. In the two experiments, the channel was configured to turbulent-air-water channel, and the length of the water channel varied from 3.2-m to 8-m. At last, the BER comparison was shown in Fig. 15. In Fig. 15, the curve '8QAM-Asyn' was obtained in the first experiment, and the curve '8QAM-Syn' was obtained in the second experiment. When the SE was nearly the same, '64QAM-Parallel' outperformed '8QAM-Asyn' when timing error existed in two paths of conventional OFDM signals with 8-QAM constellation. Furthermore, '64QAM-Parallel' was very close to '8QAM-Syn', which meant the proposed two-path parallel scheme with 64-QAM constellation had nearly the same performance as the two paths of conventional OFDM signals with 8-QAM constellation and without timing error.

Fig. 15 Comparison of BER performance over a turbulent-air-water channel when two paths of conventional OFDM signals with 8-QAM constellation were transmitted synchronously.

Download Full Size | PDF

By comparing Fig. 12 and Fig. 13 with Fig. 7 and Fig. 9, respectively, the differences between the results obtained in turbulent-air-water channel and the results acquired in non-turbulent-air-water channel were such tiny that they can be ignored. The reason was that the generated turbulence in the experiments was weak, the fluctuation frequency was about 20-Hz, which meant the coherent time of the turbulent-air-water channel was about 50-ms. The coherent time was far longer than the duration of one OFDM data frame (72.192-us) in this paper, hence the fade caused by turbulence was very limited. Additionally, the phase fluctuation induced by the turbulence cannot affect the received OFDM signal, because the intensity modulation and direct detection (IM/DD) was used. For the proposed two-path parallel scheme in Fig. 1, the key point is to guarantee the same or highly correlated phase noise is introduced in path-A and path-B. At the receiver end, the oscilloscope used two synchronized channels to capture the received baseband OFDM signals, which introduced the same timing error to the two-path signals. At the transmitter end, the same training sequences were inserted in the baseband signals of path-A and path-B, and the transmission paths of path-A and path-B were close to each other, which ensured the received signals from path-A and path-B had nearly the same signal-to-noise ratio (SNR). Thus, similar synchronization error was introduced in path-A and path-B. In summary, according to the experiments described above, the proposed two-path parallel scheme in Fig. 1 manifested its superiority over the conventional single-path OFDM scheme. It provides a better choice for m-QAM-OFDM transmission in FSO communications, VLC and UOWC than conventional single-path OFDM transmission when it is necessary to keep longer transmission distance and high spectra efficiency (SE) at the same time.

5. Conclusions

In this paper, we proposed a two-path parallel scheme for m-QAM-OFDM transmission in optical wireless communications. The principle of the proposed scheme was theoretically derived first. The derivation show that the proposed scheme can improve the performance by mitigating the ICI induced by phase noise. We experimentally demonstrated the feasibility of the proposed scheme by transmitting two-path OFDM signals synchronously or asynchronously over a turbulent-air-water channel. The results show that the proposed scheme can achieve remarkable BER performance improvement whatever the OFDM signals of the two paths were transmitted synchronously or asynchronously. Furthermore, even in the turbulent-air-water channel, the proposed scheme still had excellent BER performance, which manifested its stability. In some limited scenarios of FSO communications, VLC and UOWC, the proposed scheme can help to extend the transmission distance effectively while high spectra efficiency (SE) is kept.

Funding

National Natural Science Foundation of China (NSFC) (61605149); Natural Science Foundation of Shanxi Province under Grant No. 2018JM6074; the Fundamental Research Funds for the Central Universities (XJS17095).

References

1. C.-J. Chen, J.-H. Yan, D.-H. Chen, K.-H. Lin, K.-M. Feng, and M.-C. Wu, “A 520-nm Green GaN LED with High Bandwidth and Low Current Density for Gigabits OFDM Data Communication,” in Optical Fiber Communication Conference (San Diego, California, 2018), paper Th2A.18. [CrossRef]

2. B. Schrenk and C. Pacher, “1 Gb/s All-LED Visible Light Communication System,” in Optical Fiber Communication Conference (San Diego, California, 2018), paper M1F.4. [CrossRef]

3. G. Zhang, J. Zhang, X. Hong, and S. He, “Low-complexity frequency domain nonlinear compensation for OFDM based high-speed visible light communication systems with light emitting diodes,” Opt. Express 25(4), 3780–3794 (2017). [CrossRef] [PubMed]

4. H. M. Oubei, J. R. Duran, B. Janjua, H.-Y. Wang, C.-T. Tsai, Y.-C. Chi, T. K. Ng, H. C. Kuo, J. H. He, M. S. Alouini, G. R. Lin, and B. S. Ooi, “4.8 Gbit/s 16-QAM-OFDM transmission based on compact 450-nm laser for underwater wireless optical communication,” Opt. Express 23(18), 23302–23309 (2015). [CrossRef] [PubMed]

5. Y. Chen, M. Kong, T. Ali, J. Wang, R. Sarwar, J. Han, C. Guo, B. Sun, N. Deng, and J. Xu, “26 m/5.5 Gbps air-water optical wireless communication based on an OFDM-modulated 520-nm laser diode,” Opt. Express 25(13), 14760–14765 (2017). [CrossRef] [PubMed]

6. C. Fei, J. Zhang, G. Zhang, Y. Wu, X. Hong, and S. He, “Demonstration of 15-M 7.33-Gb/s 450-nmUnderwater Wireless Optical Discrete Multitone Transmission Using Post Nonlinear Equalization,” J. Lightwave Technol. 36(3), 728–734 (2018). [CrossRef]

7. M. N. Raed, H. Elgala, and T. D. C. Little, “A Novel Method to Mitigate LED Nonlinearity Distortions in Optical Wireless OFDM Systems,” in Optical Fiber Communication Conference (Anaheim, California, 2013), paper JW2A.69.

8. C. Chen, W.-D. Zhong, and D. Wu, “Indoor OFDM Visible Light Communications Employing Adaptive Digital Pre-Frequency Domain Equalization,” in Conference on Lasers and Electro-Optics (San Jose, California, 2016), paper JTh2A.118. [CrossRef]

9. X. Lu, M. Zhao, L. Qiao, and N. Chi, “Non-linear Compensation of Multi-CAP VLC System Employing Pre-Distortion Base on Clustering of Machine Learning,” in Optical Fiber Communication Conference (San Diego, California, 2018), paper M2K.1. [CrossRef]

10. Y. Zhou, S. Liang, S. Chen, X. Huang, and N. Chi, “2.08Gbit/s Visible Light Communication Utilizing Power Exponential Pre-equalization,” in 25th Wireless and Optical Communication Conference (WOCC, 2016).

11. C. Li, Z. Xu, C. Yang, Q. Yang, and S. Yu, “Experimental Demonstration of Clipping Noise Mitigation for OFDM-Based Underwater Optical Wireless Communications,” in Asia communications and photonics conference (Guangzhou, China, 2017), paper M1G.2.

12. A. J. Lowery, “Enhanced Asymmetrically Clipped Optical ODFM for High Spectral Efficiency and Sensitivity,” in Optical Fiber Communication Conference (Anaheim, California, 2016), paper Th2A.30. [CrossRef]

13. Y. Hei, J. Liu, H. Gu, W. Li, X. Xu, and R. T. Chen, “Improved TKM-TR methods for PAPR reduction of DCO-OFDM visible light communications,” Opt. Express 25(20), 24448–24458 (2017). [CrossRef] [PubMed]

14. J. Bai, Y. Li, Y. Yi, W. Cheng, and H. Du, “PAPR reduction based on tone reservation scheme for DCO-OFDM indoor visible light communications,” Opt. Express 25(20), 24630–24638 (2017). [CrossRef] [PubMed]

15. S. Dong, J. He, M. Chen, Q. Chen, R. Deng, J. Ma, and L. Chen, “Performance improvement of ACO-OFDM system using OCT precoding combined with digital peak-clipping,” in Conference on Lasers and Electro-Optics (San Jose, California, 2018), paper Th3I. 1. [CrossRef]

16. J. Wang, Y. Xu, X. Ling, R. Zhang, Z. Ding, and C. Zhao, “PAPR analysis for OFDM visible light communication,” Opt. Express 24(24), 27457–27474 (2016). [CrossRef] [PubMed]

17. J. Zhang, X. Chen, D. Zeng, H. Yang, X. Yi, and K. Qiu, “ICI cancellation using symmetric subcarrier pairs with opposite weightings in CO-OFDM systems,” in Asia communications and photonics conference (Shanghai, China, 2014), paper ATh3A.121. [CrossRef]

18. X. Yi, B. Xu, J. Zhang, Y. Lin, and K. Qiu, “Theoretical calculation on ICI reduction using digital coherent superposition of optical OFDM subcarrier pairs in the presence of laser phase noise,” Opt. Express 22(25), 31192–31199 (2014). [CrossRef] [PubMed]

19. X. Yi, X. Chen, D. Sharma, C. Li, M. Luo, Q. Yang, Z. Li, and K. Qiu, “Digital coherent superposition of optical OFDM subcarrier pairs with Hermitian symmetry for phase noise mitigation,” Opt. Express 22(11), 13454–13459 (2014). [CrossRef] [PubMed]

20. X. Hong, X. Hong, J. Zhang, and S. He, “Low-complexity linewidth-tolerant time domain sub-symbol optical phase noise suppression in CO-OFDM systems,” Opt. Express 24(5), 4856–4871 (2016). [CrossRef] [PubMed]

21. W.-R. Peng, “Analysis of Laser Phase Noise Effect in Direct-Detection Optical OFDM Transmission,” J. Lightwave Technol. 28(17), 2526–2536 (2010). [CrossRef]

22. W.-R. Peng, J. Chen, and S. Chi, “On the Phase Noise Impact in Direct-Detection Optical OFDM Transmission,” IEEE Photonics Technol. Lett. 22(9), 649–651 (2010). [CrossRef]

23. L. Zhang, Y. Ming, and J. Li, “Suppression of laser phase noise in direct-detection optical OFDM transmission using phase-conjugated pilots,” Opt. Commun. 403, 197–204 (2017). [CrossRef]

24. W.-R. Peng, I. Morita, and H. Tanaka, “Digital Phase Noise Estimation and Mitigation Approach for Direct-Detection Optical OFDM Transmissions,” in European Conference on Optical Communication (Torino, Italy, 2010), paper Tu.3.C.3. [CrossRef]

25. M.R.H. Mondal and J. Armstrong, “Analysis of the effect of vignetting on MIMO optical wireless systems using spatial OFDM,”, J. Lightwave Technol. 32(5), 922–929 (2014). [CrossRef]

26. L. Zhang, H. Wang, and X. Shao, “Improved m-QAM-OFDM transmission for underwater wireless optical communications,” Opt. Commun. 423, 180–185 (2018). [CrossRef]

27. J. W. Giles and I. N. Bankman, “Underwater optical communications systems. Part 2: basic design considerations,” in Proceedings of IEEE Military Communications Conference (IEEE, 2005), 1700–1705. [CrossRef]

28. E. J. Lee and V. W. S. Chan, “Part 1: Optical Communication Over the Clear Turbulent Atmospheric Channel Using Diversity,” IEEE J. Sel. Areas Comm. 22(9), 1896–1906 (2004). [CrossRef]

29. N. D. Chatzidiamantis, G. K. Karagiannidis, and M. Uysal, “Generalized Maximum-Likelihood Sequence Detection for Photon-Counting Free Space Optical Systems,” IEEE Trans. Commun. 58(12), 3381–3385 (2010). [CrossRef]

Parameters	Value
Number of data-carrying subcarriers	50
FFT size	256
Length of cyclic prefix (CP)	32
Number of subcarriers for frequency gap near DC	6
Number of training symbols for synchronization	2
Number of training symbols for channel estimation	8
Number of OFDM symbols (excluding training symbols)	256
Net data rate of 16-QAM modulation format	668.34-Mbps
Net data rate of 64-QAM modulation format	1-Gbps

Experimental demonstration of a two-path parallel scheme for m-QAM-OFDM transmission through a turbulent-air-water channel in optical wireless communications

Abstract

1. Introduction

2. Principle

3. Experimental setup

4. Results and discussions

5. Conclusions

Funding

References

Cited By

Figures (15)

Tables (1)

Equations (14)

Optics Express