1. Introduction

Light emitting diodes (LEDs) are taking over other illumination devices and becoming a popular lighting technology. LEDs are now being deployed in homes, offices, car head lights and street lights because of low cost, larger life times and low power consumption [1]. LEDs can also help in broadband optical wireless communication (OWC) and has an edge over radio frequency (RF) transmission as its spectrum is not regulated. For this reason, OWC can be used in medical applications where RF transmission systems cannot be used [2].

For luminance purpose two types of LEDs are commonly used: phosphorus-based type, containing a blue chip and a phosphor layer, and multicolor type, consisting of three or four independent chips. Because of simpler design and lower cost, phosphorus LEDs are preferred for illumination, while priority is given to multicolor LEDs for high speed data transmission applications as they allow wavelength division multiplexing (WDM).

In the past, we demonstrated visible light communication (VLC) using both types of white LEDs. We used Quadrature amplitude Modulation (QAM) and Discrete Multi-tone (DMT) to achieve data rates of 1 Gbit/s and 3.4 Gbit/s. We achieved 1 Gbit/s in line-of-sight transmission but at distance of only 10 cm (420 lx) using low power phosphorus LEDs [3]. VLC link at 50 cm was also demonstrated in the same experiment at just 10 lx with data rate of 640 Mbit/s. Later we also demonstrated 1.5 Gbit/s using single LED of a RGB chip in [4]. The aggregate values of 3.4 Gbit/s and 2.2 Gbit/s were achieved at distance of 10 cm (410 lx) and 30 cm (25 lx) respectively using multicolor LEDs, exploiting WDM. The illumination levels were below the standard limit for indoor environments (500 lx at desktop level [5]).

Furthermore, in [6] 1.25 Gbit/s data rate is demonstrated using a single RGB LED at 10 cm. MIMO OFDM is utilized in [7] to achieved data rate of 1.1 Gbit/s at 1 cm and with coverage area of 25 mm². They used beam shaping lens on LEDs to adjust the radiation pattern. Wang et al. in [8] achieved the aggregate data rate of 3.25 Gbit/s at 1 cm. They also reported 2.5 Gbit/s over 65 cm. But these working distances are still not sufficient for the indoor environments where at least 1 to 3 m distances is needed with high data rates and low complexity. Moreover, in all the described experiments only unidirectional transmission was reported.

In this paper, we demonstrate a sharp rise on bit rate for WDM VLC link based on an integrated four channel LED board at common indoor distance (1.5 to 4 m). It is for the first time that exploiting commercially available LEDs 5 Gbit/s aggregate value was exceeded using VLC. Moreover, in order to demonstrate full duplex high speed transmission we deployed IR channel for uplink transmission with data rate greater than 1.1 Gbit/s. We obtained maximum data rate of 5.6 Gbit/s for downlink transmission at 1.5 m while at the same distance uplink transmission of 1.5 Gbit/s was achieved. All channels were maintained at BER value lower than FEC limit (3.8·10⁻³ [9]). The light of the four LED boards generates an overall neutral white color and their combined luminance value varied from 128 lx to 720 lx at distance of 4 m to 1.5 m, respectively. These results were obtained after a deep investigation of the parameters and devices involved in the experiment such as LEDs emission wavelength optical filter spectra, working current, electrical signal power.

2. Experimental setup

We realized the experimental setup reported in Fig. 1. Four different boards were used: each of them was equipped with three low cost LED chips of a single color. Four different colors were used: red, green, blue and amber. In the single board, the three LED chips were connected in series. The four boards were closely (around 3 cm between one each other) mounted together, generating a total luminous flux of about 700 lm at 350 mA nominal driving current. The lambertian pattern of each board was reduced to 22 degrees from 120 degrees using proper low cost plastic lens, in order to maximize the transmission distance.

 

Fig. 1 Block diagram of the experimental setup of the VLC system. AWG: Arbitrary waveform generator; RTO: Real Time Oscilloscope.

Download Full Size | PDF

The DMT signals were generated by a personal computer and fed into an arbitrary waveform generator (AWG). The signal consisted of N = 512 subcarriers within a baseband bandwidth of B = 220 MHz (429 kHz subcarrier spacing). The bandwidth was chosen to fully exploit the available bandwidths and slow frequency roll-off of the LEDs in order to increase the throughput. We preliminary estimated the signal-to-noise ratio (SNR) on each subcarrier from the Error Vector Magnitude (EVM) [10] of the received constellations of a DMT probe signal having equally power BPSK modulated subcarriers. Bit and power loading was then applied on N-1 subcarriers, in order to adapt the individual carrier loading to the estimated SNR. The gross data rate includes 12% overhead which comprises of cyclic prefix (3%), training sequence (2%) and FEC (7%). The DMT signal generated by positive output port 1 of the AWG was used to modulate the channel under test. The other three LEDs were modulated using Port 1 negative output and two outputs of port 2 (positive and negative). An analogous DMT signal with different bit stream was loaded on port 2.

Four similar power amplifiers (Minicircuits, 25 dB gain, 29 dBm minimum output power at 1 dB compression, 130 MHz 3-dB bandwidth) were used to amplify the four AWG outputs. LEDs were driven using bias-T which superimposed amplifiers outputs on dc bias currents. At the downlink receiver (Rx_D), a lens (Thorlabs, 50 mm diameter) was used to focus light onto an ac-coupled avalanche photodiode module (Hamamatsu APD, 0.42 A/W responsivity at 620 nm and gain = 1) having 3.14 mm² active area and an integrated transimpedance amplifier (280 MHz 3-dB bandwidth). Optical dichroic filter was mounted in front of a photo-detector at the Rx_D (the spectral emissions of the LEDs and the respective bandpass filters are shown in Fig. 2) to receive each color independently. As can be see, the broad emission of the LEDs (especially the green) falls in the adjacent filters, generating a cross-talk. However, this phenomenon was negligible in most cases. The only notable overlap is the one where the green emission pass through the amber filter. We will see that this overlap will cause a small degradation into the system performance. Finally, offline processing was performed on the received signal from the APD by sampling it using real-time oscilloscope (LeCroy, 2 GSa/s sampling rate).

 

Fig. 2 Normalized optical spectra of the LEDs used for the downlink experiment with the corresponding optical filters (dashed lines). The percentages values represent the transmittance of the filters.

Download Full Size | PDF

The uplink channel exploited an IR-LED as transmitting source (Tx_u) emitting at 850nm (0.6 W optical power at 160 mA bias current) with 130 degrees lambertian emission. The assumption of an IR-LED for the uplink transmitter is common, as low-power infrared light could be placed very close to the user, provided that the eye safety conditions are fulfilled. Unlike downlink, no optical lens was used with uplink source. Same APD of downlink was adopted for uplink signal reception by placing high pass filter (805 nm edge wavelength) and optical lens in front of uplink receiver (Rx_u). The output acquired from the Rx_u was then stored in RTO for offline demodulation.

The adopted IR-LED was compliant with the eye-safety standards at its maximum bias current (1 A). An IR-LED source is completely exempt from risk if the device does not pose an IR radiation hazard for either the cornea nor for the retina, according to clause 4.3 of IEC-62471 [11]. Using the specifications provided by the datasheet at 1 A bias current, considering a realistic source-eye distance of 0.2 m, the cornea exposure is 30 Wm⁻², which is lower than the exposure limit fixed to 100 Wm⁻² for long exposure (>1000 s). We also calculated that the retinal exposure is 42 mW/mm²sr, which is lower than the retinal exposure limit (at λ_IR = 850 nm) of 60 mW/mm²sr for weak visual stimulus (at λ_IR the pupil is fully opened). These values were obtained assuming 1 A bias current; however, in our experiment, the bias current used for the communication was 160 mA, which is much lower than the one used for the eye-safety calculations. We thus conclude that the link emission is exempt from hazard both to cornea and retina, also for long exposures.

3. System characterization

The bipolar DMT signal is very sensitive to nonlinear distortions because of its large dynamic range. The LEDs exhibit nonlinearity because of current-voltage exponential relationship and due to nonlinear characteristics of the output power as a function of the input current. The overall distortion levels can be controlled by adjusting input power and bias current of the LEDs. In Fig. 3(a)-3(e) we present the achievable bitrate (at fixed BER value) as a function of the electrical power of the DMT signal for different values of applied bias current. The circle highlights the selected value of bias current and signal power. The LED’s (especially the green) exhibited optimal working bias over a wide range. In this case, the bias current was selected in order to reduce the color tendencies in the overall white.

 

Fig. 3 (a-e) Achievable bitrates as a function of the electrical signal power to the LEDs at different bias currents for all the WDM channels. (f-j) optimal bit and power loading for downlink and uplink measured at 1.5 m (720 lx) for all the WDM channels. All the results are taken considering a BER ≤ 3.8·10⁻³.

Download Full Size | PDF

This analysis provided us the optimum working conditions for red (170 mA, 21.7 dBm), green (240 mA, 22.7 dBm), blue (150 mA, 23.7 dBm), amber (40 mA, 22.7 dBm), and infrared (160 mA, 24.7 dBm).

We also then measured the cross-talk measurements in order to characterize the WDM system and to analyze reduction in bitrate when other channels were turned on. Cross talk was noticed only for amber channel, where 3% decrease in bitrate was observed, whereas the rest of the channels had no degradation lower than 1%. The cross talk values observed justified the adjacent channel interference resulted from overlap of optical spectra as depicted in Fig. 2. In our setup, 2% of green light pass into the amber filter and degradate the amber channel performance. The other cross-talk were lower then 1%.

4. Transmission experiment

After characterization and optimization we demonstrated the bidirectional transmission. The system performance was analyzed in terms of bit rate guaranteed with BER ≤ 3.8·10⁻³ at different distances, ranging from 1.5 m to 4 m, with illumination levels in the range (approximately) from 720 lx to 120 lx respectively. The distributions of bit and power loading for all the LEDs used in the experiments are reported in Fig. 3(f-j), considering a distance of 1.5 m. In Fig. 4 we report 5 constellation diagrams for the carriers having highest modulation format in each WDM channels. The constellation can be clearly recognized.

 

Fig. 4 Received constellation diagrams of the highest modulation format loaded for different channels: red, green, amber, blue and infrared.

Download Full Size | PDF

In Fig. 5(a) we report the maximum transmission speed of individual channel (color) as a function of the distance between Tx and Rx. The corresponding illuminance levels are also reported in the secondary X axis. The four channels had quite different performance because of receiver responsivity, which is higher for red and lower for blue light. Moreover, the degraded behavior of green and amber was due to higher losses resulted from corresponding band pass filters.

 

Fig. 5 a) Achievable downlink bit rates as a function of the Tx-Rx distance for the four WDM channels. b) Achievable total bit rates for downlink (left axis) and uplink (right axis) as a function of the Tx-Rx distance. All the bit rates are taken at BER ≤ 3.8·10⁻³.

Download Full Size | PDF

In Fig. 5(b) we report aggregated downlink (black squares) and uplink (red diamonds) bit rates as a function of distance. A total downlink data rate ≥ 5 Gbit/s is achieved for distance ≤ 3.5 m (168 lx), with a maximum of 5.6 Gbit/s at 1.5 m (720 lx). While on the other end, the uplink data varies from 1.1 to 1.5 Gbit/s at distances of 4 to 1.5 m. Although, the transmitter source in the uplink was not equipped with a lens in order to restrict the beam divergence, yet same performance was observed as compared to one WDM downlink channel. This was due to reason that receiver had better responsivity at this wavelength.

We finally characterized the behavior of the downlink system as a function of the radial distance from the vertical axis connecting Tx to the Rx plane. We estimated the achievable bit rate measuring the illuminance levels along the radial distance, up to 55 cm. Then, from the data reported in Fig. 5(a), we extracted the bit rate for a particular illuminance value, interpolating the measured values. The results of this estimation for each WDM channel are reported in Fig. 6 where we can see that for an area with a ray of 55 cm (~1 m²), the system still can achieve an aggregate bit rate around 5 Gbit/s.

 

Fig. 6 Maximum downlink bit rates versus the radial distance for each WDM channel.

Download Full Size | PDF

5. Conclusions

In this paper, we proposed and experimentally demonstrated a bi-directional high-speed VLC system based on WDM and DMT modulation (with offline processing) of a custom RGBY LED source. For the first time, record capacities of 5.6 Gbit/s (downlink) and 1.5 Gbit/s (uplink) were achieved maintaining illuminance levels within the standards of working environment, greatly increasing the previous bitrate-illuminance ratio. The resultant bit error ratios (BERs) in all the channels were below the forward error correction (FEC) limit of 3.8·10⁻³. We also outline that this result was obtained at common indoor distances (1.5 ÷ 4 m).