Visible light communication (VLC) uses light-emitting diodes (LEDs) to provide illumination and transmit data through the same optical channel. For industrial automation engineers and system integrators, the approach is relevant wherever optical wireless links are of interest. A practical challenge is that white LEDs designed for lighting may not switch rapidly enough for data transmission. One transmitter design addresses this by keeping the illumination LED continuously on and assigning data modulation to a separate LED.
Why white LEDs can limit visible light communication
White LEDs commonly produce white light in one of two ways. Some combine red, green, and blue emissions. Others use a blue-emitting chip with a phosphor layer that converts part of the blue light into longer-wavelength light, which combines with the remaining blue to appear white.
Both arrangements can create communication constraints. In a multicolour LED, the high drive current needed for the lighting source can make rapid switching problematic. In a phosphor-based LED, the phosphor may continue emitting light briefly after the blue excitation is switched off. This luminescence decay extends the effective symbol duration and can interfere with rapid data transmission.
The proposed arrangement separates the lighting and communication functions. The white LED remains at a constant operating condition, while a monochromatic communication LED is switched to encode the signal. Because the communication LED does not rely on phosphor conversion, its light generation is determined by the semiconductor material rather than the decay of a phosphor layer.
How spectral suppression and reconstruction work
The method first uses a notch filter to suppress a selected portion of the white LED’s emitted spectrum. A second LED is chosen to emit light in the suppressed spectral region. When the filtered white light and communication LED output are combined, they can approximate the original white spectrum.
This arrangement has two objectives: maintain the perceived lighting characteristics while enabling the communication channel to be modulated separately. The illumination LED is not switched on and off with the data. Instead, only the replacement LED carries the changing signal.
In the reported measurement setup, researchers combined the light beams using a 50:50 beamsplitter, a mirror, and a diffuser. The beamsplitter arrangement discarded half of the optical power, a limitation noted in the paper. The diffuser helped mix the beams, but its attenuation also contributed to the short test distance.
Evaluating the reconstructed white light
The team measured the original white LED and then repeated the measurements with the notch filter and communication LED in place. A spectrometer and integration sphere were used to assess the spectrum, colour coordinates, and correlated colour temperature (CCT). The communication LED’s forward current was adjusted in 10 mA steps, from 100 mA to a maximum of 350 mA, to find a setting that best matched the original light.
The original source had CIE colour coordinates of x = 0.3064 and y = 0.3107, with a CCT of 7070.7 K. The closest reported reconstruction was achieved with the communication LED at 280 mA. Its measured coordinates were x = 0.3087 and y = 0.3108, and its CCT was 6915.7 K. The authors reported that the colour-coordinate changes were below 1% and that the CCT changed by 2.2%.
These measurements indicate that, in this setup, the added communication light could replace the filtered spectral portion without a large change in the measured colour coordinates. The evaluation used colour measurements as a way to compare the reconstructed light with the original; it does not establish performance for every LED, filter, or installation.
Communication performance and EVM
The transmitter’s data performance was evaluated using quadrature amplitude modulation (QAM) and Error Vector Magnitude (EVM). EVM compares the received symbols with their ideal positions in the modulation constellation. A larger error indicates a greater difference between the received and intended signal, making EVM a useful measure of digitally modulated link quality.
Tests used 4QAM and 8QAM at carrier frequencies of 3 MHz and 4 MHz. The symbol rate was increased until communication broke down. Bitrate was calculated from symbol rate and the number of possible modulation states: 4QAM carries two bits per symbol, while 8QAM carries three.
The highest reported bitrate was 4.4 Mb/s, achieved with 4QAM at a 4 MHz carrier frequency. The 4QAM test at 3 MHz reached 3.8 Mb/s. Both 8QAM tests reached 3.6 Mb/s before communication broke down. The transmitter and PIN photodetector were separated by 1 m, so these results describe the specific laboratory arrangement rather than a general operating range.
Practical implications for automation applications
The central engineering idea is functional separation: preserve steady illumination with one source and use another source to carry the modulated signal. This avoids switching the white lighting LED and, for the tested design, avoids relying on a phosphor-based source for the data modulation.
Any implementation would need to consider both the spectral match and the optical arrangement. The communication LED’s output must replace the filtered portion closely enough to meet lighting requirements, while the transmitter geometry affects how much light reaches the receiver. The reported beamsplitter power loss and diffuser attenuation identify areas for design improvement; the authors specifically point to improving transmitter geometry as a way to address the short test distance.
Conclusion
Spectral suppression and reconstruction offer one way to reconcile white-light output with LED-based data transmission. The measured prototype closely matched the original colour coordinates and demonstrated QAM communication, with a highest reported bitrate of 4.4 Mb/s under a 1 m laboratory test. For technical decision-makers, the results highlight a promising design principle, alongside practical considerations: validate light quality, measure EVM and link behaviour, and account for optical losses and installation geometry before applying the approach to an operational system.