Open Access
Add to BibSonomy
Abstract
In a practical light emitted diodes (LEDs)-based visible light communication (VLC) system, high-speed transmission is generally limited by the LED bandwidth. To address the bandwidth limitation, a hybrid digital linear and decision-feedback equalization (DFE) is investigated to improve the transmission performance (or spectral efficiency) in the carrier-less amplitude phase modulation (CAP)-based VLC systems. A real-time CAP-VLC transceiver with the hybrid digital equalization is designed, based on which 200 Mb/s transmission is successfully demonstrated over a 15 m VLC link with the commercial red LEDs (bandwidth: 6.5 MHz). In the real-time CAP-VLC system, the baseline wander (BLW) is observed, due to the removal of the low-frequency components with a direct current (DC) block. The BLW effect can be mitigated by increasing the roll-off factor. However, this roll-off factor affects the equalization performance, due to an increased loss in the signal spectrum beyond the system bandwidth. Optimization of the roll-off factor and filter length is required. Experimental results show that, with the optimized real-time transceiver design, the hybrid Wiener/recursive least squares (RLS) and DFE significantly improves the error vector magnitude (EVM) performance compared with the DFE. In addition, the digital signal processing (DSP) complexity is discussed.
© 2019 Optical Society of America under the terms of the OSA Open Access Publishing Agreement
1. Introduction
As a complementary wireless technology to radio frequency (RF)-based solutions, visible light communication (VLC) has attracted growing research interest in the last decade due to the wide deployment of light emitting diodes (LEDs) for the indoor and outdoor infrastructures [1–3]. VLC can offer advantages of unlicensed spectrum, high security and immunity to electromagnetic interference, so that it has been considered as a cost-effective and energy-efficient solution for future high-speed communications. In a VLC system, LEDs can be used for both illumination/signage and communication. However, the limited modulation bandwidth of a commercial LED with a phosphor layer (up to several MHz) is a key factor affecting the transmission capacity [1].
To address the bandwidth challenge, recent research was focused on the following three main solutions: parallel transmission [3–5], analog/digital equalization [6] and high spectral-efficient modulation [7–10]. For the parallel transmission schemes of multiple-input multiple-output (MIMO) or wavelength-division-multiplexing (WDM)-based VLC [3–5], multiple transceivers (LEDs/detectors) are required, which increases the system complexity and energy consumption. The analog equalization can be used to increase the system bandwidth, which is usually realized by suppressing the signal power in the low frequency range. However, the method causes a reduced signal-to-noise ratio (SNR). For a cost-effective and flexible VLC system, carrierless amplitude-phase (CAP) modulation with digital equalization is a promising solution to enhance spectral efficiency and system capacity [3,7–10]. Compared with orthogonal frequency division multiplexing (OFDM), CAP signal has a small peak-to-average power ratio so that the linear region of the power-voltage (P-U) transfer of the LEDs can be fully used to improve energy efficiency. In addition, the complexity of the CAP system is relatively low by taking advantage of high-speed digital signal processing (DSP) for signal generation and recovery.
As the CAP transmission performance is sensitive to non-flat spectral channels [11,12], effective digital equalization is required to mitigate the spectral roll-off effect due to the LED bandwidth limitation. In the previous literature, digital equalization for CAP-VLC were investigated including least mean squares (LMS), recursive least squares (RLS), and modified cascaded multi-modulus algorithm (M-CMMA). Most experimental investigations were conducted with offline signal processing [3,6–8], which does not consider limitation of digital signal processing. Moreover, a linear equalizer may not suffice in practical VLC systems, where crosstalk or reflection arises from imperfect sampling or impedance mismatch (discontinuity) between boards, cables, LED driver circuits [12,13]. The introduction of a nonlinear equalizer could be beneficial to mitigation of the crosstalk and/or reflection effects.
In a practical VLC system, a photodetector is usually AC-coupled to a transimpedance amplifier to remove the received DC photocurrent induced by the ambient/artificial light and continuous wave light from the intensity modulator [14,15]. A direct current (DC) block is commonly added or integrated into amplifiers or analog-to-digital converters (ADCs) to not only use the full range of ADC input, but also reduce the noise harmonics from fluorescent light sources, which is less than 1 MHz. The removal of the DC and low-frequency components results in baseline wander (BLW) or inter-symbol-interference (ISI) in a baseband transmission system [15]. In general, there are two methods of mitigating the BLW effect: quantized feedback baseline restoration [16] and transmission of a small spectral content at low frequencies. The former method increases the system complexity, whilst the latter method is preferable since it can be easily implemented by optimization of signal spectrum. Although CAP is passband signal that has a notch at DC in the frequency domain, the low frequency components of the signal can be affected by the BLW because the roll-off factor is usually set at a low value to fully use the limited bandwidth of LEDs. Therefore, BLW is also an important factor considered for the design of the real-time CAP-VLC transceiver.
In recent years, VLC has been considered as a promising technique for outdoor applications including intelligent transport system (ITS). Compared with the RF-based ITS that offer a limited RF band of 30 MHz in Europe and 75 MHz in United States, VLC-based ITS can provide unlicensed wide bands for high-speed vehicle-to-vehicle/infrastructure (V2X) transmission [17]. However, the bit rate reported from the previously experimental V2X demonstrations is up to 55 Mb/s (or 10 Mb/s) at a distance of 1.5 m (or 20 m) [18,19]. The transmission distance significantly decreases with increasing the bit rate [20], which is limited by the modulation bandwidth of the LED as discussed above. In general, a relatively long VLC link is preferable for outdoor V2X applications. To enhance the transmission performance in practical V2X-based VLC systems, we designed an FPGA-based real-time CAP transceiver with hybrid linear and decision-feedback equalization (DFE), which was experimentally demonstrated over ≤15 m VLC links with red traffic LEDs [21]. Together with linear equalization (normalized LMS (NLMS), RLS or Wiener filtering) [22,23], an adaptive DFE is used to greatly improve the transmission performance of 200 Mb/s CAP signal as compared to the conventional DFE. In this paper, based on the previous work, extended investigation is made on detailed discussions about real-time transmission performance of the 200 Mb/s CAP signal with the hybrid digital equalization and DSP complexity. In addition, in order to mitigate the BLW effect a large roll-off factor is preferable, which however affects the equalization performance due to a large loss in the signal spectrum beyond the system bandwidth. The optimization of the roll-off factor and filter length is required. The impact of the BLW on the CAP-VLC transmission performance is numerically and experimentally investigated over a 15 m VLC link with the commercial red LEDs with a bandwidth of 6.5 MHz.
2. CAP-VLC system with hybrid digital equalization
2.1 CAP transmission in VLC system
CAP is a multilevel and multidimensional modulation technique, in which a pair of orthogonal filters instead of sinusoidal carriers is used to generate a real-valued CAP signal for intensity modulation/direct detection (IMDD) transmission. The CAP signal with the orthogonal filters fI(t) and fQ(t) can be written as [12]
where is a baseband shaping filter, is carrier frequency, dI(t) and dQ(t) are real and imaginary parts of output signal from M-ary quadrature amplitude modulation (QAM) mapping, respectively. The operator ’⊗’ represents convolution. In this paper, the square root raised cosine (SRRC) filter is used as a baseband shaping filter, which is given by where Ts is symbol period, n is an integer, and α is a roll-off factor that determines the shape of the CAP signal spectrum.For an IMDD-based VLC system, the electrical CAP signal is added with a constant, C, for a positive signal before intensity modulation or electrical-to-optical conversion (EOC). After free-space light propagation over a certain distance, the detected optical signal is converted into electrical signal at the receiver. Assuming that the EOC and optical-to-electrical conversion (OEC) are linear, the received electrical signal can be expressed by [24,25]
where η and h(t) are attenuation coefficient and impulse response of the VLC link, respectively. w(t) is additive white Gaussian noise (AWGN) with a variance of σ2. The value of η is mainly determined by the slope of P-U function of LEDs, optical loss of free-space light transmission, responsivity of a photodiode (PD) and electrical amplification/attenuation. For a PD detector, the noise at the receiver is usually dominated by shot noise and thermal noise [26]. To avoid ISI due to limited bandwidth of LEDs, the received CAP signal is equalized before or after the matched filters. After that, the transmitted data is recovered using an inverse procedure of the transmitter.2.2 Hybrid digital equalization at CAP receiver
Thanks to the advance of DSP, CAP signal can be digitally generated and recovered. Figure 1 shows the schematic diagram of an LED-based VLC system including digital CAP transmitter and receiver with the proposed hybrid digital equalization. Considering that the LED-based VLC link is a slowly time-varying channel, the hybrid digital equalizer consists a linear equalizer and a DFE.
Fig. 1 Schematic diagram of a CAP-VLC system with hybrid digital equalization. EA: electrical amplifier. Sync.: synchronization.
At the CAP transmitter, a pseudo random binary sequence (PRBS) is used as a data source for the input of the QAM mapping. The output complex sequence is upsampled at a rate of Γ for inphase/quadrature filters to form a discrete CAP signal. Γ is an integer number. A preamble sequence (128 symbols) is added in the beginning of each frame (120000 symbols) for synchronization and channel estimation. The analog CAP signal from the DAC is then amplified to drive LEDs with a Bias-Tee for illumination and communication.
At the receiver, optical light of the CAP signal is detected by an avalanche photodiode (APD) with a convex lens for high optical gain. To improve electrical SNR, an amplifier and a low pass filter (LPF) are used before feeding the received signal into the ADC. After digitizing the analog CAP signal, digital recovery of the CAP signal is carried out at the CAP receiver. The linear equalizer is applied after the ADC at a sampling rate of fs. The sampling rate is equal to Γ⋅ BC, where BC is symbol rate of CAP signal. The output signal is written as
where hL is impulse response of a FIR filter with a length of D. Here, three different algorithms are used, including normalized LMS (NLMS), RLS and Wiener filtering. Prior to the linear equalizer, synchronization is carried out to identify the start of frames in the received sequence for channel estimation. After the matched inphase/quadrature filters, the CAP signal is downsampled for adaptive DFE before QAM demapping. The adaptive DFE is used to compensate the dynamic signal distortion due to the slowly varied channel response and imperfect sampling at the receiver. The equalized signal is given bywhere a and b are forward and backward coefficients, respectively. q(n) is the received signal after downsampling. For simplicity, LMS algorithm is used to update the coefficients, a and b. Error vector magnitude (EVM) is calculated by comparing the received and transmitted signals. It is worth noting that since the frequency response of the LEDs is relatively stable, the coefficients for the linear equalizer can be calculated offline and stored in the memory of FPGA. In general, the transmission performance is determined by not only the 3-dB bandwidth of LEDs, but also the noise at the transmitter and receiver including quantization, shot and thermal noise. To quantify the equalization performance, a ratio of baud rate to 3-dB bandwidth of LEDs, BL, is defined aswhere Rb is bit rate. In practice, the signal bandwidth is usually larger than 3-dB bandwidth of the LEDs for high-speed transmission, which results in β>1.2.3 Baseline wander
To model the BLW, a high-pass filter (HPF) is introduced in the VLC link. The frequency response of the VLC link is thereby given by [27]
where fL (fP) is the 3-dB bandwidth of the LEDs (photodetectors) that is modelled with an LPF, kL and kP are the fitted coefficients, u(t) is the Heaviside step function, and hB(t) is the impulse response of the HPF. As an example, Fig. 2 shows the frequency responses of a VLC link and in-phase/quadrature filters. For a relatively large roll-off factor, the BLW does not affect the signal spectrum due to a small amount of signal power in the low frequency range. However, the wide signal spectrum may suffer large loss in the high frequency range beyond the system bandwidth. Therefore, there is a trade-off between the attenuation in the low and high frequency ranges. In the next section, the impact of the BLW on the CAP transmission performance with the hybrid digital equalization is discussed, based on which the roll-off factor is optimized for different bit rates.3. Transmission performance of CAP-VLC signals
Following the descriptions on the principle of the hybrid digital equalization and the BLW in a CAP-VLC system, in this section, a real-time CAP transceiver with the DSP design is developed in Fig. 3(a). To ensure that the sampling frequency offset between the transmitter and receiver is small enough, a single FPGA board with a DACs and an ADC is used for both digital signal generation and recovery. A high-level modulation format of CAP16 is used to improve spectral efficiency for high bit rate VLC systems with bandwidth-limited LEDs. With the FPGA-based CAP transceiver, experimental investigation of the real-time CAP16 transmission performance is undertaken over ≤15 m VLC links. Table 1 lists default values of key system parameters in the experiment or simulation unless explicitly mentioned.
Table 1. Key parameters of the CAP-VLC system
3.1 Experimental setup
In the real-time CAP-VLC system, the analog CAP signal is generated with a 16-bit DAC operating at a maximum sampling rate of 200 MS/s. As an example, a commercial red traffic LED module with a 3-dB bandwidth of 6.5 MHz in Fig. 3(b) is used as a light source. Compared with the other VLC systems with blue/yellow/white LEDs, a red LEDs-based VCL system has relatively large LED modulation bandwidth and high APD responsivity at long wavelengths. Therefore, red LEDs are preferable for high bit rate or long distance transmission. After the intensity modulation, the optical signal propagating from the LED module is detected by a commercial APD (Hamamatsu, C12702-12) at the receiver. As a plano-convex lens (diameter: 10cm) is placed in front of the APD for long distance transmission, the receiver has a limited field of view (FoV), which may require an effective tracking scheme for dynamic VLC links [28]. The received signal passing an amplifier and an LPF is then digitized by a 14-bit ADC for digital recovery of the CAP signal as mentioned in Section 2. Γ is set to 4. For the conventional DFE, the coefficients for DFE are usually estimated with training symbols in an initial state of each incoming data frame. Here, the adaptive DFE is adopted by updating the coefficients with estimated values in the preceding data frame, which increases the number of training symbols for a relatively accurate channel estimation.
Figure 4(a) shows a measured P-U curve of the LEDs. The output optical power linearly increases with bias voltage in the bias voltage range of ≥7 V. At a bias voltage of 8.2 V, the measured frequency response of the VLC link has a 3-dB bandwidth of 6.5 MHz as seen in Fig. 4(b). In the real-time CAP-VLC system, the major source for the BLW is the ADC, which suffers large attenuation at DC. With the channel model in Eq. (10), the simulated frequency response is well fitted with the experimental one with fL = 6.5 MHz, fP = 40 MHz, kL = 0.93, kP = 1.25. The BLW is modelled with a Butterworth HPF with a stopband frequency of 0.1 MHz, a passband frequency of 0.5 MHz, a passband ripple of 10 dB, and stopband attenuation of 31.8 dB.
3.2 BLW effect
With the aforementioned channel model and measured frequency response of the VLC system, the impact of BLW on the CAP transmission performance is initially investigated in the simulation (Matlab). The receiver noise is simulated with the following APD parameters: noise-equivalent power (NEP) of 2pW/Hz1/2, excess noise factor of 3.5 [29]. Considering that the low frequency components of the CAP signal spectrum depend on the bit rate and α, the curves of EVM performance with the hybrid Wiener and DFE versus α are plotted in Fig. 5(a). It is seen that the EVM performance with the BLW degrades at α<0.5 and α<0.7 for bit rates of 50 Mb/s (β = 1.9) and 25 Mb/s (β = 0.96), respectively. For the bit rate of 200 Mb/s (β = 7.7), the EVM with the BLW is almost the same as that without BLW in the range of α = 0.05-0.95. This can be explained that the increase in bit rate and/or α can alleviate the BLW effect. However, a large value of α results in wide spectrum of CAP signal, which causes much loss beyond the 3-dB bandwidth as shown in Fig. 2. Therefore, the EVM of the 200 Mb/s CAP signal slightly degrades with increasing the α. A tradeoff between the bit rate and α should be considered.
Fig. 5 EVM performance versus the roll-off factor in the (a) simulation and (b) experiment. Distance: 3 m.
To further explore the BLW effect, experimental investigation is made on real-time EVM performance with BLW in the 3 m FPGA-based VLC system with the hybrid Wiener and DFE. As shown in Fig. 5(b), the trend of the EVM performance for bit rates up to 200 Mb/s is similar to that in Fig. 5(a). The overall performance in the simulation is slightly better than the performance in the experiment because of the imperfect DACs/ADCs in the experiment. Both numerical and experimental results indicate that a relatively small (large) roll-off factor is preferable for high (low) bit-rate CAP signals. At the roll-off factor of 0.1, the EVMs for bit rates of 50 Mb/s and 200 Mb/s are almost equal, indicating a minimum EVM for 200 Mb/s CAP-VLC transmission. Therefore, the default roll-off factor is set at 0.1 unless explicitly mentioned [9,30].
3.3 Real-time transmission performance
To distinguish optical effect from electrical components, electrical back-to-back (B2B) transmission is investigated, which means that the output of the DAC is directly connected to the input of the ADC. In the B2B case, the enhancement performance of the DFE is experimentally validated in Fig. 6, which shows EVM performance of the CAP16 signal with or without DFE. The signal bandwidth is varied by adjusting sampling rate of the DAC/ADC. Compared with no DFE, the EVM performance with the DFE is significantly improved at different bandwidth of the CAP signal. The simulated EVM agrees well with the experimental result, which indicates that the relatively high EVM at a relatively low bandwidth is due to the low-frequency distortion of the ADC. The lowest EVM value of −32 dB indicates the feasibility of the DFE.
After the validation of the DFE in the B2B case, experimental investigation of real-time CAP16 signal is undertaken over a red LEDs-based VCL link. For intensity modulation, bias and root mean square (RMS) driving voltages of 200 Mb/s CAP signal are optimized with the hybrid Wiener and DFE in Fig. 7(a). Because of the LED nonlinearity in the very low/high driving voltage region, the best EVM performance of −23.8 dB over a 1m VLC link is achieved at an optimized bias (RMS driving) voltage of 8.2V (0.47V), below which the EVM decreases with increasing bias voltage. The optimum values of bias and RMS driving voltage are adopted in the following experimental investigations.
Fig. 7 (a) Optimization of bias voltage, Vb, and RMS driving voltage of LEDs (1m), (b) EVM performance for different equalization schemes as a function of number of taps for linear equalization (8m).
It is well known that the number of equalization taps affecting the real-time CAP transmission performance depends on the VLC channel. To choose an optimum number of taps for the hybrid digital equalization schemes, Fig. 7(b) shows EVM performance of 200 Mb/s CAP signal over an 8 m VLC link for different equalization schemes as a function of number of taps (D) when La = Lb = 3. Compared with the linear equalization, the hybrid NLMS/RLS/Wiener and DFE schemes significantly improve EVM performance by >4 dB at the optimum number of taps (4-10). It indicates that only linear equalization cannot be used to completely recover the distorted signal. It is noted that a large number of taps is preferable for accurately modelling an impulse response, which however, results in increased noise power after the equalization.
With the optimized number of taps for the linear equalization, the impact of number of taps (La, Lb) for DFE on the CAP transmission performance is numerically investigated over a VLC link in Fig. 8, where the receiver SNR is 20 dB or 35 dB. As seen from Fig. 8 the EVMs for the DFE and hybrid equalization (Wiener and DFE) remain at the minimum value at La≥3 and Lb≥3. The minimum EVMs for the hybrid Wiener and DFE are lower than that for the DFE. Therefore, the DFE with La = Lb = 3 is used in the following study. It is noted that because of the relatively high baud rate of 50 MBd compared with the small LED bandwidth of 6.5 MHz, the minimum EVMs are higher than the inverse of the SNR.
Fig. 8 EVM performance versus number of taps for DFE (a, b) and hybrid Wiener and DFE (c, d). (a, c): SNR = 20 dB, (b, d): SNR: 35 dB (simulation).
After the optimization of the key parameters, transmission performance of the real-time CAP signal is investigated over ≤15 m VLC links. Figures 9(a,b) show EVM performance of the real-time CAP signals at a bit rate up to 200 Mb/s over 3 m and 15 m transmission. The hybrid digital equalization is used to compensate the large power loss in the high frequency range of the received CAP-VLC signal spectrum as seen in Fig. 10(a). Due to the limited bandwidth of the LEDs and roll-off effect of the DACs/ADCs, the electrical spectrum of the received CAP signal over a 15 m VLC link decreases by >30 dB at the frequency of 50 MHz (signal bandwidth) as compared with the flat signal spectrum at the transmitter.
Fig. 10 (a) Electrical spectra of transmitted and received CAP-VLC signals (15 m). (b) Experimental and numerical EVM performance of 200 Mb/s CAP16 signal over a 1 m VLC link.
As shown in Fig. 9, the hybrid equalization schemes outperform the (adaptive) DFE scheme, whilst the hybrid Wiener/RLS and DFE scheme has the best EVM performance. There is up to 5 dB and 10 dB improvement in EVM for the hybrid equalization at the distance of 15 m and 3 m, respectively. The EVM difference between the hybrid equalization and the DFE increases with increasing the bit rate. This indicates the limitation of the DFE in recovering the distorted signal suffering large loss in the high frequency range for high-speed CAP systems. The introduction of a linear equalizer before the DFE can ease the burden of the DFE in mitigating the linear effect so that the overall performance is improved. For the short distance case in Fig. 9(a), the EVM for the hybrid equalization schemes remains almost constant at different bit rates because a relatively large received optical power at a short distance causes an EVM floor. This can be confirmed in Fig. 10(b) that EVM performance of 200 Mb/s CAP signal over 1 m VLC links decreases as received optical power increases up to −20 dBm, beyond which EVM varies slightly with received optical power. Both numerical and experimental results of the EVM performance agree with each other. In order to show the impact of the transmission distance on the EVM performance, the measured distances corresponding to different received optical power are given in Fig. 10(b). The EVM performance significantly degrades with increasing distance. It is predicted that 300 Mb/s CAP64 signal with a spectral efficiency of 6 bit/s/Hz can be achieved for error-free transmission at a BER threshold of 3.8 × 10−3 for the 7% overhead hard-decision forward error correction (FEC) [31,32], which corresponds to an EVM value of −22.5 dBm for CAP64. It is noted that the predicted performance (300 Mb/s) is raw bit rate before deducting the overhead from the FEC.
Table 2 shows the resource consumption in the report of the hardware compilation for the DSP design at the FPGA transceiver. For the traditional/proposed scheme, the linear equalization is realized after/before the matched filers (inphase and quadrature), which results in digital processing of a complex/real-valued signal. Compared with the traditional scheme, the proposed scheme for 200 Mb/s (12.5 Mb/s) CAP transmission saves 3.8% (4.3%) lookup tables (LUTs), 3.5% (4.1%) registers and 15.4% (30.7%) DSP48Es. The large consumption of DSP48Es for the high-speed CAP transceiver is due to the restrict requirement in multipliers at a fast clock frequency.
Table 2. Comparison of DSP complexity
4. Conclusion
Experimental investigation of a real-time 200 Mb/s CAP-VLC transceiver with the hybrid linear equalization and DFE has been made in a 15 m VLC system with the commercial red LEDs. As the BLW causes large loss in the low frequency range that affects the equalization performance, the roll-off factor and FIR filter length have been optimized for the design of the CAP-VLC transceiver. Experimental results have shown that with the optimized real-time transceiver design the proposed hybrid Wiener/RLS and DFE significantly improves the EVM performance compared with the DFE. The DSP complexity with the proposed scheme is 15.4% lower than the traditional scheme for the 200 Mb/s FPGA-based CAP transceivers. In addition, the impact of the BLW on the CAP-VLC transmission performance has been numerically and experimentally investigated. It has shown that the BLW effect is effectively mitigated in the high (low) speed CAP-VLC systems with a small (large) roll-off factor.
Funding
National Key Research and Development Program of China (2017YFB0403604); Key Program of National Natural Science Foundation of China (61631018); Key Research Program of Frontier Sciences of CAS (QYZDY-SSW-JSC003); Fundamental Research Funds for the Central Universities (WK2100060022).
Acknowledgement
The authors would like to thank Information Science Laboratory Center of USTC for the hardware & software services.
References
1. H. Elgala, R. Mesleh, and H. Haas, “Indoor optical wireless communication: potential and state-of-the-art,” IEEE Commun. Mag. 49(9), 56–62 (2011). [CrossRef]
2. A. Hussein and J. Elmirghani, “Mobile multi-gigabit visible light communication system in realistic indoor environment,” J. Lightwave Technol. 33(15), 3293–3307 (2015). [CrossRef]
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), 7904507 (2015).
4. K. Liang, C.-W. Chow, and Y. Liu, “RGB visible light communication using mobile-phone camera and multi-input multi-output,” Opt. Express 24(9), 9383–9388 (2016). [CrossRef] [PubMed]
5. K. Werfli, P. Chvojka, Z. Ghassemlooy, N. B. Hassan, S. Zvanovec, A. Burton, P. A. Haigh, and M. R. Bhatnagar, “Experimental demonstration of high-Speed 4 × 4 imaging multi-CAP MIMO visible light communications,” J. Lightwave Technol. 36(10), 1944–1951 (2018). [CrossRef]
6. N. Chi, Y. Zhou, S. Liang, F. Wang, J. Li, and Y. Wang, “Enabling technologies for high-speed visible light communication employing CAP modulation,” J. Lightwave Technol. 36(2), 510–518 (2018). [CrossRef]
7. F F. M. Wu, C. T. Lin, C. C. Wei, C. W. Chen, Z. Y. Chen, H. T. Huang, and S. Chi, “Performance comparison of OFDM signal and CAP signal over high capacity RGB-LED-based WDM visible light communication,” IEEE Photonics J. 5(4), 7901507 (2013). [CrossRef]
8. G. Stepniak, L. Maksymiuk, and J. Siuzdak, “Experimental comparison of PAM, CAP and DMT modulations in phosphorescent white LED transmission link,” IEEE Photonics J. 7(3), 7901708 (2015). [CrossRef]
9. J. L. Wei, C. Sanchez, and E. Giacoumidis, “Fair comparison of complexity between a multi-band CAP and DMT for data center interconnects,” Opt. Lett. 42(19), 1 (2017). [CrossRef] [PubMed]
10. P. A. Haigh, A. Burton, K. Werfli, H. L. Minh, E. Bentley, P. Chvojka, W. O. Popoola, I. Papakonstantinou, and S. Zvanovec, “A multi-CAP visible light communications system with 4.85 b/s/Hz spectral efficiency,” J. Sel. Areas Commun. 33(9), 1771–1779 (2015). [CrossRef]
11. M. I. Olmedo, T. Zuo, J. B. Jensen, Q. Zhong, X. Xu, S. Popov, and I. T. Monroy, “Multiband carrierless amplitude phase modulation for high capacity optical data links,” J. Lightwave Technol. 32(4), 798–804 (2014). [CrossRef]
12. A. Abdolhamid and D. Johns, “A comparison of CAP/QAM architectures,” in Proc. IEEE International Symp. on Circuits and Systems (IEEE, 1998), 316.
13. S. Gondi and B. Razavi, “Equalization and clock and data recovery techniques for 10-Gb/s CMOS serial-link receivers,” IEEE J. Solid-State Circuits 42(9), 1999–2011 (2007). [CrossRef]
14. A. Street, K. Samaras, D. O’Brien, and D. Edwards, “Closed form expressions for baseline wander effects in wireless IR applications,” Electron. Lett. 33(12), 1060–1062 (1997). [CrossRef]
15. Z. Ghassemlooy, “Investigation of the baseline wander effect on indoor optical wireless system employing digital pulse interval modulation,” IET Commun. 2(1), 53–60 (2008). [CrossRef]
16. F. Waldhauer, “Quantized feedback in an experimental 280-Mb/s digital repeater for coaxial transmission,” IEEE Trans. Commun. 22(1), 1–5 (1974). [CrossRef]
17. M. Uysal, Z. Ghassemlooy, A. Bekkali, A. Kadri, and H. Menouar, “Visible light communication for vehicular networking: performance study of a V2V system using a measured headlamp beam pattern model,” IEEE Veh. Technol. Mag. 10(4), 45–53 (2015). [CrossRef]
18. Y. Goto, I. Takai, T. Yamazato, H. Okada, T. Fujii, S. Kawahito, S. Arai, T. Yendo, and K. Kamakura, “A new automotive VLC system using optical communication image sensor,” IEEE Photonics J. 8(3), 1 (2016). [CrossRef]
19. A. Cailean and M. Dimian, “Current challenges for visible light communications usage in vehicle applications: a survey,” IEEE Comm. Surv. and Tutor. 19(4), 2681–2703 (2017). [CrossRef]
20. K. Cui, G. Chen, Z. Xu, and R. Roberts, “Experimental characterization of traffic light to vehicle VLC link performance,” IEEE Globecom Workshops, Houston, TX, USA, 808–812 (2011).
21. Y. Mao, X. Jin, W. Liu, C. Gong, and Z. Xu, “Demonstration of real-time CAP transceivers with hybrid digital equalization for visible light communication,” in Proc. Asia Commun. and Photonics Conf. (ACP), 2017, paper M1G.4. [CrossRef]
22. S. Haykin, Adaptive Filter Theory, 4th Edition (Prentice Hall, 2002).
23. J. Chen, J. Benesty, Y. Huang, and S. Doclo, “New insights into the noise reduction wiener filter,” IEEE Trans. Audio Speech Lang. Process. 14(4), 1218–1234 (2006). [CrossRef]
24. F. Miramirkhani and M. Uysal, “Channel modeling and characterization for visible light communications channel modeling and characterization for visible light communications,” IEEE Photonics J. 7(6), 1 (2015). [CrossRef]
25. Z. Ghassemlooy, W. Popoola, and S. Rajbhandari, Optical Wireless Communications: System and Channel Modelling (CRC Press INC, 2012).
26. F. Xu, M. A. Khalighi, and S. Bourennane, “Impact of different noise sources on the performance of PIN- and APD-based FSO receivers,” in Proc. International Conf. on Telecom. (IEEE, 2011), 211–218.
27. H. Minh, D. O’Brien, G. Faulkner, L. Zeng, K. Lee, D. Jung, Y. Oh, and E. Won, “100-Mb/s NRZ Visible Light Communications Using a Postequalized White LED,” IEEE Photonics Technol. Lett. 21(15), 1063–1065 (2009). [CrossRef]
28. F. Liu, M. Chen, W. Jiang, X. Jin, and Z. Xu, “Effective auto-alignment and tracking of transceivers for visible light communication in data centres,” in Proc. SPIE Photonics West OPTO, San Francisco, California (2019).
29. R. Yang, X. Jin, M. Jin, and Z. Xu, “Experimental investigation of optical OFDMA for vehicular visible light communication,” in Proc. of European Conf. on Optical Communication (ECOC) (IEEE, 2017), 1–3. [CrossRef]
30. L. Tao, Y. Wang, Y. Gao, A. P. T. Lau, N. Chi, and C. Lu, “Experimental demonstration of 10 Gb/s multi-level carrier-less amplitude and phase modulation for short range optical communication systems,” Opt. Express 21(5), 6459–6465 (2013). [CrossRef] [PubMed]
31. T. Nguyen, S. Le, M. Wuilpart, and P. Mégret, “Experimental demonstration of a low-complexity phase noise compensation for CO-OFDM systems,” IEEE Photonics Technol. Lett. 30(16), 1467–1470 (2018). [CrossRef]
32. G. Tzimpragos, C. Kachris, I. Djordjevic, M. Cvijetic, D. Soudris, and I. Tomkos, “A survey on FEC codes for 100 G and beyond optical networks,” IEEE Comm. Surv. and Tutor. 18(1), 209–221 (2016). [CrossRef]
