Performance analysis of space multiplexing by superposed signal in multi-dimensional VLC system

Liang Qiao; Xingyu Lu; Shangyu Liang; Jiao Zhang; Nan Chi

doi:10.1364/OE.26.019762

1. Introduction

For future communication, many studies have indicated that communication terminals will spend approximately 80 percent of the times indoors. As a result, high-speed data rate services in indoor will contribute to the majority of the ever increasing traffic and motivates us to continue exploring new spectrum and new technologies [1]. Visible light communication (VLC), which utilizes the light emitting diode (LED) based lighting infrastructure for data transmission [2,3], has a number of distinct advantages: cost-benefit, energy-efficient transmission, high security, and low latency. For these reasons, VLC is envisioned to fill a complementary role along with wireless communication to offload traffic.

However, the commercial LED has finite modulation bandwidth, which prohibits high transmission rate of VLC [4–7]. To overcome this limitation, many methods have been proposed in the former literature. Among them, one straightforward method is to form a multiple input multiple output (MIMO) or multiple input single output (MISO) with several LEDs [8–10], as they can be used for illumination while improving the quality of the signal. Since the signal must be non-negative and real and VLC systems always employ simple intensity modulation and direct detection (IM/DD) techniques, additional modulation orders can be generated in VLC-MIMO systems [11,12]. The reason is that only the signal intensity is used to convey information and the received signal can be viewed as a superposition of signals from each individual LED transmitter.

In addition, since VLC is generally line-of-sight (LOS) and cannot propagate through walls, the correlation between spatial channels employing non-imaging receiver is high. The de-correlation from different spatial streams becomes the bottleneck and ill-conditioned channel matrix may occur in the MIMO VLC system [13]. Thus, the traditional MIMO techniques cannot be directly applied to the VLC system [14]. However, the channel state information (CSI) is often assumed perfect [15–20], such as the method of the singular value decomposition (SVD) [21] or orthogonal circulant matrix transform (OCT) [22], these two algorithms demand the size of receiver arrays has to be several distances to ensure that the channel matrix is full-rank. It means that only one stream data can be decoded in our formulated SR-MIMO model.

Is there any other alternative? Interestingly, overlay coding (OC) and superposition coded modulation (SCM) are potential solutions as they are proven technologies that offer high transmission data rate. The OC is used in the image sensors for pixelated MIMO VLC channels utilizing imaging receivers. The SCM is widely investigated in the wireless communication to improve the system performance. However, they share the same character: transmit superposed signal. The author in [23] proposes a suboptimal detection method named sum detection. A linear detection algorithm named fast low-complexity maximum likelihood (ML) is employed for the double-layer in imaging VLC system [24]. In this reference, when the receiver image sensor installed is far from the transmitter, the high spatial-frequency for the lower layer fails to be detected. As a result, only the upper-layer data with high priority can be recovered. Inspired by these, the superposed signal can act a savior to realize space multiplexing in the Multi-dimensional IM/DD system. However, VLC is a bandwidth-limiting system, and the modulation bandwidth of commercial LEDs is merely around 20 MHz without pre-equalization. By using well-matching pre-equalization, the acquired bandwidth in our system can be increased to 250MHz. If we employ the code division multiple accesses (CDMA) in the VLC system, the transmitted signal is formed by the spread code multiplied by the original data and then need more bandwidth to transmit these symbols. Therefore CDMA scheme is not suitable for high-speed VLC system, but for relatively lower speed VLC system, it can be a solution for multi-user access [25].

In this paper, we propose a single receiver multiple-input-multiple-output (SR-MIMO) model to realize the space multiplexing. Since the spatial channel correlation in this model is quite high in the LOS scenario, it is difficult for one receiver to separate the superposed signals by utilizing the traditional MIMO algorithm. In this model, one transmitted signal stream is multiplied by a proportionality factor to generate various superposed constellation structures. To separate this superposed signal, we design a modified detection algorithm, which joint the successive interference cancellation (SIC) and the look-up table (LUT). Simulation results have verified the feasibility of the proposed model. Experiments demonstrate that the measured Bit Error Rates (BERs) are under 7% pre-forward error correction (pre-FEC) limit of 3.8 × 10⁻³ at the data rate of 1.5Gbit/s in 1.3-m indoor LOS scenario. In addition, an optimal range of voltage difference between the two LEDs is illustrated in this system, in which the space multiplexing can be realized. Finally, we show and analyze several of the superposed constellation structures for the SR-MIMO model.

2. System description

2.1. Superposed signal in VLC-SR-MIMO model

Firstly, considering a VLC-SR-MIMO system, it consists of $2 \times 1$ optical MISO channels with two LEDs in the transmitter and one photo-detector (PD) in the receiver. The original information bits are modulated into source data vector denoted by $x (t) = {[x_{1} (t), x_{2} (t)]}^{T}$ . $h_{k} (t)$ presents the $2 \times 1$ channel matrix. $n_{k} (t)$ is the noise, $k$ is the number of the LED transmitters. Therefore, the received signal can be expressed by:

y (t) = \sum_{k = 1}^{2} h_{k} (t) \otimes x_{k} (t) + n_{k} (t)

Quadrature amplitude modulation (QAM) is utilized. The real or imaginary part of the signal can be expressed as:

A_{X} = \pm 1, \pm 3, \dots, \pm (N - 1)

where

N

is the even modulation order.

In the LED1, the pseudo-random binary sequence is modulated by quadrature phase shift keying (QPSK). The LED2 sends the signal with the modulation format of the 16QAM. According to the formula (2), the number of the $N$ is two and four, respectively. Therefore, the real or imaginary part of the QPSK signal loaded into the LED1 can be expressed as:

A_{X_{1}} = \pm 1

Correspondingly, the real or imaginary part of the 16QAM signal loaded into the LED2 is:

A_{X_{2}} = \pm 1, \pm 3

A proportionality factor $ε$ is defined and multiplied by the $A_{X_{1}}$ . Then, the real or imaginary part of the QPSK signal in LED1 is changed into $ε A_{X_{1}}$ . The received signal can be viewed as a superposition of signals from each LED, and either in the real part or the imaginary part can be expressed as:

A_{Y} = (ε A_{X_{1}} - 3, ε A_{X_{1}} - 1, ε A_{X_{1}} + 1, ε A_{X_{1}} + 3)

From the formula (5), we can access to different kinds of constellation structures by setting the $ε$ .

Figure 1 is an example of the superposed signal when the $ε$ is set as four. In this diagrammatic drawing, the corresponding modulation symbols of the QPSK signal are $(-4+4 i, -4-4 i, 4-4 i, 4+4 i)$ . In this way, the power of the QPSK signal is higher than the power of the 16QAM signal. A 64QAM signal is generated in the receiver. The circle with a cross curve drawn by a dotted line is the center of each 16QAM signal in each quadrant. The distance between the circle dotted line and coordinate axis can be changed by the $ε$ . The design of the transmitter is similar to the power allocation algorithms.

Fig. 1 Diagrammatic drawing of the superposed constellation in SR-MIMO.

Download Full Size | PDF

2.2 Detection algorithm

In theory, the subsection 2.1 elaborated the transmitter design in the SR-MIMO model. A unique constellation mapping can be achieved between the received signal and the two transmitted signals by setting the $ε$ .

In our MISO experimental setup, the transmission length is fixed to 1.3 meters. Due to the relatively short transmission distance and the short distance between two closely located transmitters, we can assume each channel gain is approximately the same. The signals from two transmitters are super-imposed in the receiver. Therefore, the MISO channel can be simplified as the SISO model for channel estimation and Zero-Forcing (ZF) is employed. Then, in the frequency domain, the channel matrix $H$ can be simplified as:

H = \frac{Y}{X} = \frac{Y}{X_{1} + X_{2}}

Let $\tilde{X}$ denotes the estimated signal of $X$ . It can be expressed by:

\tilde{X} = W_{z f} Y

where

W_{z f} = {(H^{H} H)}^{- 1} H^{H}

,

Y = \sum_{k = 1}^{2} (H_{k} X_{k} + N_{k})

. Then, the formula (7) can be written as:

\begin{array}{l} \tilde{X} = W_{z f} Y \\ = W_{z f} (\sum_{k = 1}^{2} (H_{k} X_{k} + N_{k})) \\ = \sum_{k = 1}^{2} X_{k} + {\tilde{N}}_{z f} \end{array}

where

{\tilde{N}}_{z f} = {(H^{H} H)}^{- 1} H^{H} N

.

Since the $y (t)$ is a superposed signal, after equalization, we utilize the following algorithms to separate it.

2.2.1. Look-up table (LUT)

A look-up table (LUT), as a straightforward method, can separate the superposed signal according to the one-to-one mapping relation. This table lists the ideal superposed signal in the receiver and is valid for a reasonably long period since the VLC scenario is static in most cases. Meanwhile, to improve the reliability of transmission signal which listing in the table, following methods can be considered: low modulation format, time division multiplexing (TDM) [26], space-time block coding (STBC) [11] or repetition coding (RC).

In addition, as shown in Fig. 1, the imaginary part of the superposed signal has eight values under ideal conditions, which are the same as in the real part. For this, we only need carry out 8QAM to encode the ideal superposed signal listed in this table and then expand the 8QAM signal to the 64QAM signal. After this processing, the information redundancy can be reduced by half.

2.2.2. SIC-LUT

Nonlinear effect and electronic equipment often act as the roadblock to distort the signal, then the static LUT scheme cannot address this issue. A nonlinear detector named successive interference cancellation (SIC) is widely used. To better elaborate on this algorithm, we would like to revisit the concept of the SIC. It consists of a set of linear receivers. The detected signal in each time slot is subtracted from the received signal so that the remaining signal with the reduced interference can be used in the subsequent slot [27]. Based on the traditional SIC algorithm, the author in [28] proposed an inter-lighting interference cancellation (ILIC) scheme to reduce the interference between adjacent LEDs. Both LEDs are employed the same modulation order, i.e., QPSK modulation.

In the SR-MIMO system, when the QPSK signal power loaded at the LED1 is higher than that loaded at the LED2, the received superposed signal can be easily divided into four parts. Define $Y_{c}$ is the center point in each quadrant, $v_{y_{c}}$ is the real or imaginary part of the $Y_{c}$ .It can be expressed by:

v_{y_{c}} = \frac{(ε A_{X_{1}} - 3 + ε A_{X_{1}} - 1 + ε A_{X_{1}} + 1 + ε A_{X_{1}} + 3)}{4} = ε A_{X_{1}}

According to the formula (9), we observe that the QPSK signal can be calculated by the center point, which is drawn by the dotted line in Fig. 1.

Then, the ${\tilde{X}}_{1}$ can be achieved by modifying the formula (7):

{\tilde{X}}_{1} = W_{z f_{1}} Y_{c}

Assume that $H_{1}$ and $H_{2}$ denote the channel gain from LED1 and LED2, respectively. Since two LEDs are close to each other, the channel gain satisfies that $H_{1} = H_{2}$ .Subsequently, the formula (8) can be written as:

\begin{array}{l} {\tilde{X}}_{1} = W_{z f_{1}} Y_{c} \\ = W_{z f_{1}} (H_{1} X_{1} + N_{1}) \\ = X_{1} + {\tilde{N}}_{z f} \end{array}

where

W_{z f_{1}} = {(H_{1}^{H} H_{1})}^{- 1} H_{1}^{H}

.

Finally, the $X_{1}$ is then subtracted from the superposed signal:

Y_{2} = Y - H_{1} {\tilde{X}}_{1} + {\tilde{N}}_{z f}

where

Y_{2} = H_{1} X_{2} + N_{2}

.

The signal is deteriorated by the linear and nonlinear noises in the VLC-MIMO system. For the cases of the lower SNR and larger nonlinear distortion, the received signals may departure greatly from the center point. In the traditional SIC system, the higher power signal is demodulated and subtracted from the received signal, but the residual noise of the signal spreads into the 16QAM signal. Therefore, to improve the system performance, the residual noise must be compensated. In this model, the table stores all ideal superposed signal, we can compensate the deviation by removing the difference between the actual signal and the ideal signal. For this reason, an improved detection algorithm combining with the LUT and SIC is proposed in this paper, as shown in Fig. 2.

Fig. 2 Algorithm flow of the SIC-LUT.

Download Full Size | PDF

For the sake of clarity, the steps of proposed SIC-LUT are as follows:

1) Firstly, find and calculate the center point according to the static LUT;
2) Secondly, correct the distorted center points;
3) Thirdly, decode and equalize the $X_{1}$ according to the formula (11);
4) Fourthly, calculate the $X_{2}$ by the formula (12);
5) Finally, demodulate the $X_{1}$ and the $X_{2}$ by the Monte Carlo, respectively.

Since we take the jitter of the signal into account for the SR-MIMO model, the SIC-LUT has the potential to improve the performance against the severe impact of the nonlinear effect. Notably, in the first step, the center point located in each quadrant is calculated by the formula (9). In the second step, the “correct” means subtract the difference between the ideal center point and the actual center point. Subsequently, the $X_{1}$ and the $X_{2}$ are decoded by the formula (11) and the formula (12) and demodulated by the Monte Carlo, respectively.

Combining with this equalization method and the detector algorithm, we can separate the superposed signal rather than re-modulating the high power signal in [28]. However, either the SIC algorithm or the SIC-LUT algorithm can work only with the case of $x_{1} (t) = {\tilde{x}}_{1} (t)$ . That is to say that the QPSK signal has a significant influence on the 16QAM signal.

3. Simulation and experimental setup

3.1. Simulation

A simulation with an additive white Gaussian noise (AWGN) is performed. A performance comparison is done in form of the SNR and the BER of the former three detection algorithms, which including LUT, SIC, and SIC-LUT. In this simulation, the $ε$ is four, and the corresponding modulation symbols are $(-4+4 i, -4-4 i, 4-4 i, 4+4 i)$ . The received signal either in imaginary or real part is then superposed into ${\pm 7, \pm 5, \pm 3, \pm 1}$ . By using any of these three algorithms, the caused BER performance is similar, which is shown in Fig. 3. When the SNR is higher than 18dB, all these algorithms can successfully decode the lower power signal, which is shown in Fig. 3(ii). In fact, if the 64QAM signal can be demodulated with quasi-error free in the SR-MIMO system, the superposed signal can be separated by one receiver.

Fig. 3 BER performance of different detection algorithms with SNR.

Download Full Size | PDF

3.2. Experimental setup

In subsection 3.1, we have performed a simulation of the proposed model and detection algorithm. However, some elements in the actual VLC channel, such as nonlinear effect, cannot be reflected in the aforementioned simulation. Thus, a VLC-SR-MIMO structure is experimentally designed, which is shown in the Fig. 4.

Fig. 4 Block diagram and experimental setup in the VLC-SR-MIMO system.

Download Full Size | PDF

Two independent QPSK and 16QAM signals are modulated and encoded by offline MATLAB (Matlab 2016a) with the same parameters set as follows: fast fourier transform (FFT) size of 256, Orthogonal Frequency Division Multiplexing (OFDM) number of 200, cyclic prefix (CP) length of 12.5%, training symbol of one OFDM symbol, and up-sample of 4. Subsequently, the transmitted signal is generated using an arbitrary waveform generator (Tektronix AWG520) with a sampling rate of 1GSamle/sec. Before the two signal streams transmitted through the optical channels, each of them will be firstly fed into a pre-equalizer and amplified by an electrical amplifier (EA), then superimposed onto the LED bias current by the aid of a bias-tee. The transmitted signal is carried by the red light of RGB-LEDs, while the green, blue and yellow light LEDs are turned off. The distance between LED1 and LED2 is 1.0m. Two aspheric lenses are in front of the PD, to make sure that the light transmits along the 1.3-meter perpendicular direction. The received signal is a linear combination of the two transmitted signal streams. The VLC signal is detected by a PD and then converted to an electrical signal.

It is notable that, the two signal streams are transmitted from two independent LED transmitters and then detected by the PD. The properties of the two channels are similar. Afterwards, the received electrical OFDM signal is amplified by an amplifier and then received by Agilent 54855A Digital Oscilloscope with 2GSample/sec sampling rate for signal demodulation. Finally, the offline MATLAB performs the following works, including synchronization, down-sampling, equalizing, decoding, separating and de-mapper. And the last step is to count the BER by bit-by-bit comparison between transmitted and received data, which is also a conventional approach to count the actual BER.

4. Experimental results and discussion

In this section, we would like to illustrate some experimental results to verify the practicability of the proposed SR-MIMO model and the detection algorithms. As formerly discussed, the $ε$ is related to the power difference between the QPSK signal and the 16QAM signal. It decided by the VPP voltage and the DC voltage. For clarity, some proper nouns need to be defined, the difference in DC voltage and VPP voltage are simplified to the “Diff-DC” voltage and the “Diff-VPP” voltage, respectively. In the LED1, the “QPSK signal with higher power” is simplified to the “4-SHP”. The “16QAM signal with lower power (16-SLP)” represents the signal loaded at the LED2.

Firstly, we aim at finding the proper operating range of the “Diff-DC” voltage and the “Diff-VPP” voltage, which can realize the space multiplexing in the VLC-SR-MIMO system. Figure 5(A) shows the BER curves of the 16-SLP versus the Diff-DC voltage. By utilizing the proposed SIC-LUT algorithm, the BER of the 16-SLP in the rectangular region drawn by the blue dotted line is less than the BER threshold. The two signal streams can be successfully decoded in the range of ${0 .06v, 0 .08v}$ . The space multiplexing can be realized when the Diff-DC voltage is part of this interval. In addition, the data rate can be multiply improved. However, the traditional LUT and the SIC algorithm cannot decode the 16-SLP, as they never consider to compensate for the jitter that generated by the nonlinear effect.

Fig. 5 BER versus Diff-DC voltage and Diff-VPP voltage.

Download Full Size | PDF

Figure 5(B) shows the BER curves of the 16-SLP versus the Diff-VPP voltage. Since the transmitted signal is alternating current signal and superimposed on the DC, compared with the DC voltage, the signal power difference is more likely to be affected by the Diff-VPP voltage. In this figure, when the Diff-VPP voltage is in the region of ${0 .175v, 0 .215v}$ , the receiver can decode the two non-relational data streams utilizing the modified SIC-LUT algorithm. The BER performance with the detection algorithm of the LUT is the worst. In principle, the traditional SIC detector can pull down the BER below the threshold. For example, a BER of 3.94 × 10⁻³ is achieved with the $D i f f - V P P = 0.195 v$ . Nevertheless, we hardly access to a lower BER by this algorithm. Fortunately, the improved SIC-LUT detector can broaden the operation range of the Diff-VPP voltage.

In the Fig. 5, the BER of the 4-SHP is still below the threshold. SIC-LUT can outperform the other two schemes due to the elimination of linear and nonlinear noise through the SIC-LUT process. A data rate of 1.5Gbps is successfully achieved, which equals the rate shown in [25]. As we process and separate the superposed signal in one receiver, the channel correlation is much higher than [25]. In the rectangles of both in the Fig. 5(A) and the Fig. 5(B) figures, the value of the $ε$ is almost 4.5. When $ε \leq 4$ , the BER performance of the 16-SLP gets well along with the increase of the $ε$ . Similarly, when the $ε$ beyond four, the BER performance gets worse along with the increase of the $ε$ , since the 16QAM signal is swamped by the QPSK signal. Compared with the traditional SIC, the SIC-LUT can be treated as an optimization algorithm.

In the following, we reveal and analyze the superposed constellation structures by setting the $ε$ for the SR-MIMO model. Some typical examples are illustrated in the Figs. 6(a)-6(h). Concretely, when $ε = 5.5$ , the Fig. 6(a) shows that the superposed 64QAM is divided into four parts as the power of the QPSK signal are too large. The 16-SLP in each quadrant is then indistinct. When $ε = 4.5$ , the 16-SLP can be decoded by the SIC-LUT, which is shown in Fig. 6(b). We achieve a standard 64QAM signal structure when the value of the $ε$ is four. Wherein, the “standard” means the distance between any two constellation points is equivalent. The 16-SLP can be decoded in an allowable error scope. However, the BER performance is slightly worse than that in Fig. 6(b). Interestingly, if the $ε$ is in the range of ${1,3}$ , a 25QAM structure comes into our sight. The Fig. 6(e) shows a superposed signal case with $ε =1$ . In this way, neither the 16-SLP nor the 4-SHP can be successfully decoded by utilizing any of the previous MIMO detectors.

Fig. 6 Superposed constellations in the receiver.

Download Full Size | PDF

However, transmit diversity technology can act as a savior to decode this 25QAM [11]. With the $ε$ gets smaller and less than one, the following cases appear which is shown in Figs. 6(f)-6(h). At this point, the power of the 16QAM signal is larger than that of the QPSK signal.

Figure 7 depicts the BER curves of all of the case from the Fig. 6. The x-axis represents the proportionality factor $ε$ . When $ε$ less than one, the BER curves are shown the constellation structures from the Fig. 6(h) to Fig. 6(f). It is worth noting that, when $ε =0 .5$ , we acquire a standard 64QAM signal. In this way, the power of the 16QAM signal is larger than the power of the QPSK signal and the 16QAM signal is regarded as the detracted signal from the superposed signal. For the 16QAM signal, a BER of 4.427 × 10⁻³ is achieved by utilizing the SIC-LUT algorithm. Nevertheless, we hardly obtain a lower BER, which is shown in the in the ellipse region drawn by the blue dotted line (i.e., Fig. 7(B)). When the $ε$ is in the range of the ${1,3}$ , which is shown in the region of (C), only transmit diversity technology can decode the superposed signal. As the $ε$ gets larger and greater than three, the power of the QPSK signal is larger than the power of the 16QAM signal.

Fig. 7 BER performance with the $ε$ .

Download Full Size | PDF

When the $ε$ is in the range of the ${4,5}$ , we get an approximate standard 64QAM signal, which is shown in the rectangle region drawn by the blue dotted line (i.e., Fig. 7(A)). A best BER performance in both of the QPSK signal and the 16QAM signal can be got when the $ε =4 .5$ . Correspondingly, the Diff-DC voltage and the Diff-VPP voltage are ${0 .06v, 0 .08v}$ and ${0 .175v, 0 .215v}$ , respectively. Unlike the region of (B), in the region of (A), it is easy to find the central point in each quadrant, and which facilitates the decision. In addition, we intuitively find that, when one signal power is beyond a certain value, this signal has a nice BER performance while the other cannot meet the BER threshold, such as the constellation structures of Fig. 6(a) or 6(h).

In our system, LED1 transmits the QPSK signal while LED2 transmits the 16QAM signal. Depending on whether the power of QPSK or 16QAM is larger, two operation ranges can be found. For a large QPSK case, the best performance can be achieved when the received power of the $x_{1} (t)$ is four times than the $x_{2} (t)$ . Whereas for a large 16QAM case, an ideal BER performance can be achieved when the received power of the $x_{2} (t)$ is twice of the $x_{1} (t)$ . In both cases, an approximate standard 64QAM signal can result in good detection.

5. Conclusion

In this paper, we have presented a high correlation SR-MIMO model to realize the space multiplexing. Different kinds of the superposed signals are generated by multiplying a proportionality factor into the QPSK signal in the transmitter. Then the SIC-LUT algorithm is utilized to separate the superposed signals. The experiment demonstrates that a data rate of 1.5Gbps is successfully achieved in 1.3-m indoor free space transmission with the BER below the 7% FEC limit of 3.8 × 10⁻³. Furthermore, the experiment provides an optimal range of voltage difference between the two LEDs, in which the VLC MIMO system could realize space multiplexing.

Funding

National Natural Science Foundation of China (NSFC) (61571133); National Key Research and Development Program of China (2017YFB0403603); Key project of Guangdong Province (2016B010111002).

References

1. N. Chi, H. Haas, M. Kavehrad, T. D. C. Little, and X. Huang, “Visible light communications: demand factors, benefits and opportunities,” IEEE Wirel. Commun. 22(2), 5–7 (2015). [CrossRef]

2. Y. Wang, Y. Wang, N. Chi, J. Yu, and H. Shang, “Demonstration of 575-Mb/s downlink and 225-Mb/s uplink bi-directional SCM-WDM visible light communication using RGB LED and phosphor-based LED,” Opt. Express 21(1), 1203–1208 (2013). [CrossRef] [PubMed]

3. Y. Wang, L. Tao, X. Huang, J. Shi, and N. Chi, “8-Gb/s RGBY LED-Based WDM VLC System Employing High-Order CAP Modulation and Hybrid Post Equalizer,” IEEE Photonics J. 7(6), 1–7 (2015). [CrossRef]

4. L. Minh, D. O’Brien, G. Faulkner, L. Zeng, K. Lee, D. Jung, and Y. Oh, “High-Speed Visible Light Communications Using Multiple-Resonant Equalization,” IEEE Photonics Technol. Lett. 20(14), 1243–1245 (2008). [CrossRef]

5. K. Routray, “The Changing Trends of Optical Communication,” Potentials IEEE 33(1), 28–33 (2014). [CrossRef]

6. K. Lee, H. Park, and J. R. Barry, “Indoor Channel Characteristics for Visible Light Communications,” IEEE Commun. Lett. 15(2), 217–219 (2011). [CrossRef]

7. J. Grubor, J. W. Walewski, K. D. Langer, and S. Randel, “Broadband Information Broadcasting Using LED-Based Interior Lighting,” J. Lightwave Technol. 26(24), 3883–3892 (2008). [CrossRef]

8. J. Zhu, F. Liang, K. Zhang, and Y. Zhang, “Space-Collaborative Constellation Designs for MIMO Indoor Visible Light Communications,” IEEE Photonics Technol. Lett. 27(15), 1667–1670 (2015). [CrossRef]

9. L. Qiao, N. Chi, S. Liang, and X. Lu, “MISO visible light communication system utilizing MCMMA aided pre-convergence of STBC decoding,” in Optical Fiber Communication, (OFC, 2018), Th2A.64.

10. L. Qiao and N. Chi, “Enhanced method with superposed signal-based visible light communication system using multiple-input signal-output beamforming,” Opt. Eng. 56(8), 1–7 (2017). [CrossRef]

11. J. Shi, Y. Wang, X. Huang, L. Tao, and N. Chi, “Enhanced performance using STBC aided coding for led‐based multiple input single output visible light communication network,” Microw. Opt. Technol. Lett. 57(12), 2943–2946 (2015). [CrossRef]

12. J. Shi, X. Huang, Y. Wang, L. Tao, and N. Chi, “Improved performance of a high speed 2×2 MIMO VLC network based on EGC-STBC,” in Proceeding of European Conference on Optical Communication, (ECOC, 2015), pp. 1–3. [CrossRef]

13. L. Zeng, D. C. O’Brien, HL. Minh, GE. Faulkner, K. Lee, D. Jung, Y. Oh, and ET. Won, “High data rate multiple input multple output (MIMO) optical wireless communications using white LED lighting,” IEEE J. Sel. Areas Comm. 27(9), 1654–1662 (2009). [CrossRef]

14. H. Burchardt, N. Serafimovski, D. Tsonev, S. Videv, and H. Haas, “VLC: Beyond point-to-point communication,” IEEE Commun. Mag. 52(7), 98–105 (2014). [CrossRef]

15. J. Y. Wang, J. Dai, R. Guan, L. Jia, Y. Wang, and M. Chen, “Channel capacity and receiver deployment optimization for multi-input multi-output visible light communications,” Opt. Express 24(12), 13060–13074 (2016). [CrossRef] [PubMed]

16. H. Park, C. Ko, and S. Alouini, “On the power and offset allocation for rate adaptation of spatial multiplexing in optical wireless MIMO channels,” IEEE Trans. Commun. 61(4), 1535–1543 (2013). [CrossRef]

17. T. Fath and H. Haas, “Performance comparison of MIMO techniques for optical wireless communications in indoor environments,” IEEE Trans. Commun. 61(2), 733–742 (2013). [CrossRef]

18. A. H. Azhar, T. Tran, and D. O’Brien, “Demonstration of high-speed data transmission using MIMO-OFDM visible light communications,” in Proc. IEEE Globecom Workshop OWC, (2010), pp. 1052–1056. [CrossRef]

19. A. H. Azhar, T. Tran, and D. C. O’Brien, “A gigabit/s indoor wireless transmission using MIMO-OFDM visible-light communications,” IEEE Photonics Technol. Lett. 25(2), 171–174 (2013). [CrossRef]

20. A. Burton, H. L. Minh, Z. Ghassemlooy, E. Bentley, and C. Botella, “Experimental demonstration of 50-Mb/s visible light communications using 4 × 4 MIMO,” IEEE Photonics Technol. Lett. 26(9), 945–948 (2014). [CrossRef]

21. J. Zhao, LK. Chen, and Y. Hong, “Experimental Demonstration of Performance-enhanced MIMO-OFDM Visible Light Communications,” in Optical Fiber Communication, (OFC, 2017), Th1E.2.

22. Y. Hong, J. Xu, and L. K. Chen, “Experimental investigation of multi-band OCT precoding for OFDM-based visible light communications,” Opt. Express 25(11), 12908–12914 (2017). [CrossRef] [PubMed]

23. S. Nishimoto, T. Nagura, T. Yamazato, and Yendo, “Overlay coding for road-to-vehicle visible light communication using LED array and high-speed camera,” in Proc. IEEE ITSC (2011), pp. 1704–1709.

24. Y. Zhu, J. Zhu, K. Zhang, and Y. Zhang, “A Double-Layer VLC System with Low-Complexity ML Detection and Binary Constellation Designs,” IEEE Commun. Lett. 19(4), 561–564 (2015). [CrossRef]

25. C. Yang, Y. Wang, Y. Wang, X. Huang, and N. Chi, “Demonstration of high-speed multi-user multi-carrier CDMA visible light communication,” Opt. Commun. 336, 269–272 (2015). [CrossRef]

26. Y. Wang and N. Chi, “Indoor gigabit 2×2 imaging multiple-input–multiple-output visible light communication,” Chin. Opt. Lett. 12(10), 12–15 (2014).

27. P. W. Wolniansky, G. J. Foschini, G. D. Golden, and R. A. Valenzuela, “V-BLAST: an architecture for realizing very high data rates over the rich-scattering wireless channel,” in Ursi International Symposium on Signals, Systems, and Electronics, (IEEE, 2002). 295–300.

28. K. Kim, K. Lee, and K. Lee, “An inter-lighting interference cancellation scheme for MISO-VLC systems,” Int. J. Electron. 104(8), 1377–1387 (2017). [CrossRef]

Performance analysis of space multiplexing by superposed signal in multi-dimensional VLC system

Abstract

1. Introduction

2. System description

2.1. Superposed signal in VLC-SR-MIMO model

2.2 Detection algorithm

2.2.1. Look-up table (LUT)

2.2.2. SIC-LUT

3. Simulation and experimental setup

3.1. Simulation

3.2. Experimental setup

4. Experimental results and discussion

5. Conclusion

Funding

References

Cited By

Figures (7)

Equations (12)

Optics Express