Open Access
Add to BibSonomy
Abstract
According to our experimental observation, the effect of dead time caused by the quenching circuit of single photon avalanche diode (SPAD) devices could affect the system performance significantly. Hence, we propose a corresponding synchronization scheme and a fast blind union detection (FBUD) algorithm for the received photon-counting pulse waveform by considering the effect of dead time. To verify our proposed theory under the long distance turbulence fading channel, we demonstrate a long distance SPAD-based visible light communication system with a red light-emitting diode status indicator in the practical circuit board. The experimental results indicate that the proposed FBUD algorithm outperforms the conventional dead-time free Poisson multiple-symbol detection method under different conditions. Meanwhile, the bit-to-error ratio gaps of the two algorithms will become more and more larger with the increasing of the data rate.
© 2017 Optical Society of America under the terms of the OSA Open Access Publishing Agreement
1. Introduction
Visible light communication (VLC) is an emerging technology as light-emitting diode (LED) devices will be widely used for illumination and electronic industries [1–6]. Meanwhile, duo to the advantages of easy deployment, high security and compatible spectrum, the VLC technology can be used in some special application scenarios, i.e. hospital, aircraft and coal mine, etc. At present, most of the researches about VLC systems focus on the communication theory under strong illuminance by the photodiode (PD) or avalanche photo diode (APD) receiver, which are constrained by the indoor illumination [1,5,6].
Furthermore, the VLC systems under the weak illuminance condition are attracting extensive attention. The transmitters based on the Gallium Nitride μLED and Micro-LEDs are well studied [7, 8]. In addition, the LED in the practical circuit boards and electronic products are always used as the status indicator. Yet, the status indicator also can be utilized creatively as the light source of VLC system. Whereas, the transmitted optical power of the LED indicator is very low with the typical values range from 10 to 30 mW. Hence, the receiver with high sensitivity and high detection efficiency is needed for the indicator-based VLC system.
Recently, the single photon avalanche diode (SPAD) devices have been widely studied for weak illuminance, outdoor and underwater long distance VLC systems [9–14]. Most of the previous SPAD-based VLC systems are modeled as a Poisson statistics distribution which is not from actual received waveform of SPAD [9–12]. However, according to our experimental observation and some conclusions by previous researchers, the strict output characteristic of SPAD is not Poisson distributed [15,16]. The reason is that once an SPAD has detected a photon, it will not detect any other incoming photons until the process of quenching finishes. This period is called as the dead time. Therefore the dead time sets the upper limit on the number of photons that can be detected per second. Furthermore, it will lead to a considerable counts loss and system performance decreases.
In this paper, for the received photon-counting pulse waveform under the dead time limit, we put forward a new synchronization scheme and a fast blind union detection (FBUD) algorithm. To verify our proposed theory under the long distance turbulence fading channel, we demonstrate a long distance SPAD-based VLC system with a red LED status indicator in the practical circuit board. Especially, the optical power of the indicator is only 10∼30 mW. Finally, after passing different free space transmission distances, the experimental results indicate the proposed FBUD algorithm outperforms the well-established dead time-free Poisson multiple-symbol detection (MSD) method in [17, 18]. Especially, the bit-to-error ratio (BER) gaps of the two detection algorithms will be more obvious when the data rate increases. Based on the proposed optical system, SPAD receiver and FBUD algorithm, a communication distance of 333.362 m with the data rate of 100 Kbps and BER of 1.22 × 10−3 can be achieved.
The proposed LED indicator-based VLC system can be used for the wireless isolation and information transmission between the high voltage electronic equipment and weak voltage control unit in the Smart Grid systems as Fig. 1. The primary advantage is that the VLC technology replaces the traditional cable transmission lines and low frequency wireless communication scheme, which has a more higher isolation degree and the characteristics of avoiding electromagnetic interference. Meanwhile, the proposed system also can be applied into the monitoring of the typical high voltage electric iron tower industry.
2. Experiment and system setup
2.1. The experimental setup of the long distance VLC system
The experimental setup of the long distance SPAD-based weak light communication system is shown in Fig. 2. In the transmitter, a field programmable gate array (FPGA)(ALINX, ZYNQ 7000) is used for the digital signal processing to produce the unipolar on-off keying (OOK) modulation signals. To facilitate our experiment, a single red LED status indicator (304 RUD) connects the FPGA directly and works as the light source. The voltage-to-current (V-I) conversion characteristics curve of the indicator is shown in Fig. 3. When the bit “1” is transmitted, the output voltage of the FPGA circuit is in the linear range of the V-I curve. Meanwhile, when bit “0” is transmitted, the output driving voltage is 0 V.
The LED-based VLC system is highly directional, and the received optical power is highly dependent on the “line-of-sight” path. Hence, the distance and the offset between the LED indicator and the receiver could highly affect the received signal quality. To realize the long distance transmission, an optical receiver system should be well designed. For this reason, a telescope (Bosma, Mk24020) is utilized creatively to condense the visible light from the transmitter by its reversed direction link in this experiment. Although the telescope is an optical instrument by its forward direction link that is always used to observe the remote objects.
Furthermore, a red light band-pass filter with peak value 630 nm is utilized in front of SPAD to filter out the background noise as much as possible. The background light before SPAD is measured by an illuminance meter (Everfine, Z-10). Finally, a commercially available SPAD (Excelitas Technologies, SPCM-AQRH-15-FC) receives the light signals and connects to a digital storage oscilloscope (Agilent, DSO9064A). The SPAD and the red light filter locate above the eyepiece of the telescope. In addition, they are fixed by some 3D printing materials. This well-designated structure could make the received light spot of LED indicator to be located in the effective detection area of SPAD. The other used parameters of SPCM-AQRH-15-FC in this experiment are listed in Table 1.
Table 1. Some used theoretical parameters for the SPCM-AQRH-15.
The distances between the transmitter and receiver are measured by the laser rangefinder (Swiss Technology, D810 touch). Although the maximum tested distance of D810 is 250 m, we can still measure the long distance by subsection and then sum the value of each subsection.
2.2. The received photon-counting pulse waveform and dead time
To observe the received photon-counting pulse waveform, the output maximum photon counts in one symbol interval and the practical dead time, we measure the data rates Rb=0.5, 1, 2, 5, 10, 50 Mbps respectively. The distance L between the indicator and SPAD receiver is nearly 1 cm.
Figure 4 presents some observed waveforms about the received photon-counting pulse sequences and Fig. 5 shows the practical pulse period under different data rates. To balance the authenticity of the received waveforms and the friendliness of presentation, we show the screenshots of the oscilloscope and plot other figures from the saved data. The achievable maximum received photon counts are shown in Table 2. From Fig. 4, 5 and Table 2, we could obtain the following conclusions:
- The received waveforms of SPAD-based receiver are some continuous photon-counting pulse sequences during one symbol interval, which is clearly different from the traditional waveforms of PD-based or APD-based receiver. In addition, we also could find that the amplitudes of each pulse is almost equal.
- The measurement results denote the practical dead time ranges from 25 to 28 ns which is longer than the theoretical value 20 ns. The reason is that the quenching circuit, the after pulsing and the environment temperature will affect the dead time. Meanwhile, the dead time is dependent of optical power and data rate. Hence, the maximum photon counts are lower than the theoretical value Kmax under different data rates.
- When Rb=50 Mbps, the measured photon counts tend to saturate and the maximum photon count is 1. When Rb >50 Mbps, the counted photons will be clipped which causes a serious clipping distortion problem. In other words, the data rate of single SPAD-based VLC system is restricted to a few tens of Mbps.
- The effect of dead time makes the practical output of SPAD is not Poisson distributed.
Table 2. The tested photon counts.
3. Synchronization design and detection algorithm under dead time limit
According to the previous analysis about SPAD receiver, we present the synchronization scheme and detection algorithm for the received photon-counting pulse waveform of the long distance SPAD-based VLC system.
3.1. The symbol synchronization for the received photon-counting pulse waveform
In this subsection, we present the photon-counting pulse synchronization (PCPS) scheme and parameters estimation for the received pulse waveform. Meanwhile, the symbols of one frame are consisted by fixed preamble with lt and random data with ld, where lt and ld are the lengths of synchronization sequence and data sequence, respectively. In the following experiment, the m-sequence is applied to the synchronization with lt=127 and ld=4873.
The proposed method is similar to the traditional low-pass filtering sampling. We sample the received photon-counting pulse sequence randomly and count the photon number in every time interval Td. y1,j is the photon counts of the jth symbol after the 1th sampling. Next, we count the photon number as y2,j after delaying Td/N time for the received pulse sequence. We repeat the process until delaying (N − 1)Td/N time and mark as yN,j. In fact, the delaying interval index N follows 1 < N ≤ Kmax , where K max =⌊1/δ⌋ + 1=⌊Td/τ⌋ + 1 is the maximum photon number in one symbol duration, δ = τ/Td is dead time ratio and τ is the fixed dead time. Finally, the optimal sampling point is the maximum value under the correlation operation between the sampling photon-counting value and the synchronous sequence, i,e.
where sm,j is the jth symbol of m-sequence and sm,j ∈ (0, 1).After the synchronization, we will perform parameters estimation by the following formula
where λ1 denotes the effective photon counts due to the transmitted power, λ0 is the photon counts value caused by the background light and dark counts. Nm1 ∈ {0, 1, 2, . . . lt} is the number of 1’s and Ym1 is the sum of the received statistics photons number which corresponds to the indices of the 1’s in the m-sequence. Meanwhile, Ym0 is the 0’s operation as Ym1.3.2. The detection algorithm under dead time limit
The lognormal, gamma-gamma and negative exponential fading models are proposed under different turbulence conditions for the free space channel [17,18]. The higher signal to noise ratio is needed when the turbulence-induced fading becomes more severe. Hence, we do not consider the specific channel fading model in the following analysis.
The dead time limit is neglected in the most of the previous VLC systems about SPAD-based receivers. As a result, the output probability distribution of SPAD is always modeled as an ideal Poisson statistics with the probability mass function (PMF) [9–12]
In the presence of signal alone, the practical PMF PDT(yk, λsk|h) of outputting yk counts in the time interval Td under the fixed dead time τ limit is [15,16]
where . ψ(i, λyk) is a Poisson distribution function with . In addition, λsk is the average received photon counts before the SPAD with λsk = hnssk + nb, sk ∈ (0, 1), where h denotes the fading channel, ns denotes the transmitted photons of the LED indicator and nb is the photons that comes from background noise and dark count.Figure. 6 shows the probability distribution for SPAD with and without dead time limit under λ = 50. We could observe the obvious discrepancy between the practical and ideal output distribution of SAPD. Meanwhile, PDT(yk, λsk|h) tends to be a Poisson statistics when τ → 0.
Yet, it is in general difficult to obtain an analytical solution of (4). Whereas, according to our aforementioned experimental result yk < Kmax in subsection 2.2. An rational approximate expression for (4) can be rewritten as
and we can get the threshold ρk of the practical maximum likelihood symbol-by-symbol detection with the perfect channel state information (CSI) as where λ1 = nsh + nb = nsh + λ0 and λ0 = nb.In fact, the turbulence induced-fading channel h is a constant in a short time slot. To exploit the abundant temporal correlation of channel, the fast blind union detection (FBUD) scheme is proposed for the multiple consecutive received photon-counting symbols under dead time limit. Hence, the decision rule of the FBUD algorithm with channel distribution is
where ŝ is the estimated signal vector. After eliminating the irrelevant terms, we rewrite (7) as where Non ∈ 0, 1, 2, . . . ld is the number of 1’s and Yon is the sum of the received statistics photon counts which corresponds to the indices of the 1’s in the hypothesis vector, i.e. and Son ≜ {ki ∈ 1, 2, . . . ld : ski = 1}. The other positions of ŝ are 0’s.We take out the main term of (8) as . Then we make a derivative of Λ with respect to h and let the derivation equal to be zero. Then the channel estimation is calculated by
where Non − Yonδ > 0 as . Finally, we substitute ĥ into (8) and rewrite it asFrom (9) and (10), the FBUD scheme could be applied in the absence of CSI. Finally, we can observe that the conclusion of [18] is a special case for the practical photon-counting system under δ=0. Meanwhile, the complexities of the proposed FBUD scheme with dead time limit is equal to the ideal photon-counting Poisson MSD scheme in [18].
4. Experimental results and discussions
4.1. The BER performance under different detection algorithms
In this subsection, the BER performance of our proposed FBUD algorithm and the ideal photon-counting Poisson MSD scheme in [18] will be compared in experiment after different free space transmission distances as Fig. 2. The delaying interval index of PCPS is N =5. The illuminance of background light ranges from 3.92 lx to 6.98 lx. The data rates Rb are 1, 2, 10 and 20 Mbps, respectively. In addition, some low-pass filters are designed to observe the eye-diagrams. The −3 dB frequencies are 1, 2, 10 and 20 MHz, respectively.
The BER performance comparison and some eye-diagram measurements are presented in Fig. 7 and 8. To observe the communication distance obviously under different detection algorithms, we set a BER baseline with 1 × 10−3 in Fig. 7. The achievable maximum communication distance with BER=1 × 10−3 is shown in Table 3. It is obvious that our proposed FBUD scheme under dead time limit outperforms the ideal Poisson MSD algorithm significantly. The BER gaps between the two schemes will become larger evidently when the data rate Rb increases and the dead time ratio δ increases. Especially, when the data rate is Rb=20 Mbps and the distance is L=45.640 m, the BER of our proposed FBUD algorithm is 1.29 × 10−3, which is far better than the traditional dead time-free MSD scheme with BER of 9.62 × 10−2.
Table 3. Communication distances with BER=1 × 10−3 under different detection algorithms.
From the conclusion of practical maximum photon counts in subsection 2.2, the practical probability of (4) and the proposed detection rule of (10), all of them are related to the data rate Rb because of the variable dead time ratio δ. However, the traditional MSD criterion neglects the effect of dead time and only considers the ideal Poisson photon-counting model. Hence, the proposed FBUD algorithm shows a better performance than the traditional MSD algorithm.
4.2. The communication distances under the outdoor condition
In this subsection, we will measure the communication distances under outdoor condition with our proposed FBUD algorithm as Fig. 2. The weather condition is good, meanwhile the background noise and interference are relatively small at nightfall. The delaying interval of PCPS is N =10. The data rates Rb are 10 Kbps, 100 Kbps, 500 Kbps and 1 Mbps, respectively.
The results of BER performance and the distance under outdoor condition is presented in Fig. 9. Some eye-diagram measurements with Rb=100 Kbps under different distances are shown in Fig. 10. We can observe that the communication distance is nearly 270 m with Rb=500 Kbps. Meanwhile, the maximum communication distance is 333.362 m with Rb=100 Kbps and BER of 1.22 × 10−3.
In addition, we also compare the above mentioned FBUD scheme and the ideal Poisson MSD algorithm with Rb=10 Kbps in Fig. 9(a). The BER curve of our proposed scheme is slightly better than the ideal Poisson detector and can be almost overlapped, which is the result that the output of SPAD tends to be a ideal Poisson distribution when δ → 0. This conclusion corresponds to our previous analysis about Fig. 6.
In this experiment, after adjusting the focal length with the eyepiece, the imaging area in the eyepiece shows the LED indicator and the nearby luminous objects. As a result, the equivalent noises and interference of the proposed long distance SPAD-based VLC system mainly come from the transmitter, which are different from traditional noises that come from receiver. Meanwhile, the optical system of the telescope could filter the light noise near the receiver. In addition, optical alignment is the most important experimental procedure. Finally, by analyzing the optical gains of the multiple condensers in the telescope, the signals can be detected effectively by the SPAD device. Yet, the strict calculation of optical system gains is out of this paper.
5. Conclusion
In this paper, for the practical SPAD-based VLC system under dead time limit, we first put forward the synchronization scheme and the FBUD theory. And then, we experimentally realized a long distance SPAD-based VLC system with an LED status indicator in the practical circuit board. The experimental results showed that the proposed FBUD scheme outperforms the ideal dead time-free Poisson MSD algorithm without compromising calculation complexity. Especially, when the data rate and the dead time ratio increase, the performance gaps of the two algorithms would be more larger.
In the end, the above proposed achievements can be beneficial for the wireless isolation and information transmission in Smart Grid systems and high voltage electric iron tower industry.
Funding
National Natural Science Foundation (NSFC) of China (61671477); Major Scientific and Technological Project of Guangdong Province, China (2015B010112001); Major Scientific and Technological Project of Henan Province, China (161100210200).
References and links
1. A. Jovicic, J. Li, and T. Richardson, “Visible Light Communication: Opportunities, Challenges and the Path to Market,” IEEE Commun. Mag. 51(12), 26–32 (2013). [CrossRef]
2. C. W. Chow, C. Y. Chen, and S. H. Chen, “Visible light communication using mobile-phone camera with data rate higher than frame rate,” Opt. Express 23(20), 26080 (2015). [CrossRef] [PubMed]
3. W. C. Wang, C. W. Chow, L. Y. Wei, Y. Liu, and C. H. Yeh, “Long distance non-line-of-sight (NLOS) visible light signal detection based on rolling-shutterpatterning,” Opt. Express 25(9), 10103 (2017). [CrossRef] [PubMed]
4. V. P. Rachim, Y. B. Jiang, H. S. Lee, and W. Y. Chung, “Demonstration of long-distance hazard-free wearable EEG monitoring system using mobile phone visible light communication,” Opt. Express 25(2), 713 (2017). [CrossRef] [PubMed]
5. Y. J. Zhu, W. F. Liang, J. K. Zhang, and Y. Y. Zhang, “Space-Collaborative constellation designs for MIMO indoor visible light communications,” IEEE Photon. Technol. Lett. 27(15), 1667–1670 (2015). [CrossRef]
6. D. F. Zhang, Y. J. Zhu, and Y. Y. Zhang, “Multi-LED phase-shifted OOK modulation based visible light communication systems,” IEEE Photon. Technol. Lett. 25(23), 2251–2254 (2013). [CrossRef]
7. D. Tsonev, H. Chun, S. Rajbhandari, J. J. D. McKendry, S. Videv, E. Gu, M. Haji, S. Watson, A. E. Kelly, G. Faulkner, M. D. Dawson, H. Haas, and D. O’Brien, “A 3-Gb/s Single-LED OFDM-Based Wireless VLC Link Using a Gallium Nitride μLED,” IEEE Photon. Technol. Lett. 26(7), 637–640 (2014). [CrossRef]
8. R. X. G. Ferreira, E. Xie, J. J. D. McKendry, S. Rajbhandari, H. Chun, G. Faulkner, S. Watson, A. E. Kelly, E. Gu, R. V. Penty, I. H. White, D. O’Brien, and M. D. Dawson, “High Bandwidth GaN-Based Micro-LEDs for Multi-Gb/s Visible Light Communications,” IEEE Photon. Technol. Lett. 28(19), 2023–2026 (2016). [CrossRef]
9. Y. C. Li, M. Safari, R. Henderson, and H. Haas, “Optical OFDM With Single-Photon Avalanche Diode,” IEEE Photon. Technol. Lett. 27(9), 943–946 (2015). [CrossRef]
10. T. Q. Mao, Z. C. Wang, and Q. Wang, “Receiver design for SPAD-based VLC systems under Poisson-Gaussian mixed noise model,” Opt. Express 25(2), 799–809 (2017). [CrossRef] [PubMed]
11. D. Chitnis and S. Collins, “A SPAD-Based photon detecting system for optical communications,” J. Lightwave Technol. 32(10), 2028–2034 (2014). [CrossRef]
12. C. Wang, H. Y. Yu, and Y. J. Zhu, “A long distance underwater visible light communication system with single photon avalanche diode,” IEEE Photon. J. 8(5), 7906311 (2016). [CrossRef]
13. Z. Q. Zeng, S. Fu, H. H. Zhang, Y. H. Dong, and J. L. Cheng, “A survey of underwater optical wireless communications,” IEEE Commun. Surveys Tuts. 19(1), 204–238 (2016). [CrossRef]
14. X. N. Liu, C. Gong, S. B. Li, and Z. Y. Xu, “Signal Characterization and Receiver Design for Visible Light Communication Under Weak Illuminance,” IEEE Commun. lett. 20(7), 1349–1352 (2016).
15. M. C. Teich and B. I. Cantor, “Information, error, and imaging in deadtime-perturbed doubly stochastic Poisson counting systems,” IEEE J. Quantum Electron. 14(12), 993–1003 (1978). [CrossRef]
16. E. Sarbazi, M. Safari, and H. Haas, “On the Information Transfer Rate of SPAD Receivers for Optical Wireless Communications,” in Proceedings of IEEE Global Commun. Conf., (IEEE, 2016), pp. 1–6.
17. M. L. B. Riediger, R. Schober, and L. Lampe, “Multiple-symbol detection for photon-counting MIMO free-space optical communications,” IEEE Trans. Wireless Commun. 7(12), 5369–5378 (2008). [CrossRef]
18. 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]
