CCS CONCEPTS
Passive visible light communication (VLC) takes advantage of the pervasive nature of ambient light in our environment for wireless transmissions. The design of transmitters in passive VLC predominately uses liquid crystal displays (LCDs). While LCDs are an economical choice with low power consumption, they lack some key properties that are desirable for passive VLC. For example, LCDs absorb more than half of the incident light, leaving only a small portion to be used for communication. In addition, since the direction of ambient can change over time, the relative positions of the LCDs and receivers have to be changed constantly to maintain the correct alignment.
To overcome these shortcomings, we propose the use of a novel transmitter with integrated optical fibres and digital micro-mirror devices (DMDs). DMDs are able to reflect up to 97% of the incident light, while the accompanying optical fibres aim to capture ambient light from various angles and guide them to the DMDs in a fixed direction. This design is a first step towards the goal of decoupling the direction of ambient light from the direction of the optical link, while achieving the same communication characteristics as LCDs with a much smaller device. We also design an App to allow users Copyright 2021 for this paper by its authors. Use permitted under Creative Commons License Attribution 4.0 International (CC BY 4.0). to easily interact with the system and our evaluation shows that the link can achieve a data rate of 1bps at a distance of 30cm. • Hardware → Wireless devices; • Computer systems organization → Embedded systems. Visible Light Communication (VLC) is an emerging technology for wireless communication that has gained traction from both academia and industry in recent years. Compared to traditional radio frequency wireless communication, VLC has several advantages such as an unregulated wide bandwidth and high security. VLC can be further divided into two main areas: active and passive. In both areas, the intensity of the light is modulated at a speed that is invisible to the human eye, but can be received and decoded by optical receivers. In active VLC, the driver circuitry is modified to modulate message signals by varying the driving currents of an LED. In passive VLC, an external surface is used to modulate message signals by changing the characteristics of the light passing through or reflecting from the surface. Compared to active VLC, passive VLC has the advantage of exploiting the ambient light in our environment, without having the need to directly control the light source.
The majority of passive VLC systems proposed in the literature
rely on the use of liquid crystal displays (LCDs) to modulate light
[
In our work, we are trying to implement a communication link
between a DMD and a smartphone. Smartphones have been used
before as VLC receivers but mainly using active lights sources as
transmitters (LEDs) [
As the reflection of a DMD is primarily specular, the light source and the receiver have to be precisely aligned to establish an operational link. When sunlight is used as the light source in passive VLC, as the position of the sun changes in the sky throughout the day, its direction with respect to the DMD also changes. This causes the reflected light to be misaligned to the receiver, as shown in Figure 4, where the signal-to-noise ratio (SNR) can significantly deteriorate. To overcome this problem, we propose the integration of optical ifbers and lenses into a passive VLC system, as shown in Figure 5. The sunlight is "collected" using a convex plano lens connected to multiple optical fibers. In this manner, the collected light is guided through the optical fibers and emitted directly onto the DMD. This allows the incident angle of the light on the DMD to remain the same regardless of the location of the sun. The design is shown in Figure 6, and the other end of the optical fibers, illuminating the DMD at a fixed angle, is shown in Figure 7.
For a proof of concept design, we choose the DLP2000 DMD from Texas Instrument. The DLP2000 DMD has an aperture size of 4.84mm by 3.26mm, as shown in figure Figure 8. In our design, all pixels of the DMD have the same state, and the entire DMD device acts as a single pixel with on and of states. A simple pulse width modulation (PWM) was chosen to transmit the data. When a logic one is sent, the light is reflected into the receiver for a certain time period and then not reflected for the same amount of time. When a logic zero is sent, the time that the light is not reflected into the receiver is doubled. This modulation scheme allows ones and zeros to be easily distinguish, in the expense of unequal transmit times for diferent symbols. To demonstrate our design, ASCII texts are sent over the optical link. The non-extended ASCII table contains characters that are all 8-bit long and start with a zero. Because there isn’t a character of all ones a preamble is chosen containing eight ones and then a zero. Note that using mostly ones in the preamble is faster because zeros require more transmission time. The preamble is sent multiple times to make sure the receiver will be able to see it. After the preambles, the ASCII characters are sent. There is no limit to the number of characters that can be sent. The format of the data frame is shown in Figure 9. Images from the smartphone camera are captured and processed using an Android App. After selecting a region of interest, every captured frame is sent to an image processing pipeline, as shown in Figure 10. The region of interest makes the processing easier, as only a small part of the captured image needs to be used. An OpenCV tracker is used to track the DMD, so when the user moves the hand a little, the region of interest will still be on the DMD chip.
To describe the decoding process, let us denote R as the region of interest at time (i.e. frame ). We calculate the average pixel value of each region of interest , denoted as Rc . During the preamble transmission, when the DMD is "on", the DMD will be reflecting light and the average reaches its maximum value Rˆ . When the DMD is "of", we obtain the lowest average value Rˆ . Later, when ˆ ˆ ˆ ˆ the ASCII characters are transmitted, if |R − R | < |R − R |, the DMD is decoded as "on", else as "of". Every time the state of the DMD changes, from "on" to "of" or vice-versa, we calculate the number of frames that the DMD spent in that state. If the "on" state has approximately the same length as the subsequent "of" state, a one is decoded, otherwise it is a zero. As soon as the bits are decoded, they are converted back to ASCII characters and displayed on the screen. 5
Currently a bit rate of 1bps can be achieved. One of the main reasons for the low bit rate is the tracker used in the app. There is significant room for improvement. For example, putting the tracker in a diferent thread, so the rest of the program doesn’t have to wait for the tracker to finish. A communication distance of up to 30cm is possible with our prototype. There are also some opportunities to increase the range. The image recognition step can be improved to look for smaller regions of interest. If we can capture only the DMD, the noise introduced by the surrounding areas will be eliminated and the range could increase significantly. At the moment the region of interest is set to a fixed size. 6
This work has been funded in part by the European Union’s Horizon 2020 research and innovation programme under the Marie Skłodowska Curie grant agreement ENLIGHTEM No. 814215; and by the LuxSenz project, a TOP-Grant, Module 1, Physical Sciences with project number 612.001.854, which is financed by the Dutch Research Council (NWO).