Undersampled Pulse Width Modulation for Optical Camera Communications Pengfei Luo 1 , Tong Jiang 1 , Paul Anthony Haigh 2 , Zabih Ghassemlooy 3,3a , Stanislav Zvanovec 4 1 Research Department of HiSilicon, Huawei Technologies Co., Ltd, Beijing, China E-mail: {oliver.luo, toni.jiang}@hisilicon.com 2 Department of Electronic and Electrical Engineering, University College London, London, UK Email: p.haigh@ucl.ac.uk 3 Optical Communications Research Group, NCRLab, Faculty of Engineering and Environment, Northumbria University, Newcastle-upon-Tyne, UK 3a QIEM, Haixi Institutes, Chinese Academy of Sciences, Quanzhou, China Email: z.ghassemlooy@northumbria.ac.uk 4 Department of Electromagnetic Field, Faculty of Electrical Engineering, Czech Technical University in Prague, 2 Technicka, 16627 Prague, Czech Republic Email: xzvanove@fel.cvut.cz Abstract—An undersampled pulse width modulation (UPWM) scheme is proposed to enable users to establish a non-flickering optical camera communications (OCC) link. With UPWM, only a digital light emitting diode (LED) driver is needed to send signals using a higher order modulation. Similar to other undersample- based modulation schemes for OCC, a dedicated preamble is required to assist the receiver to indicate the phase error introduced during the undersampling process, and to compensate for nonlinear distortion caused by the in-built gamma correction function of the camera. To test the performance of the UPWM- based OCC system, an experimental test-bed is developed. The experimental results show that the proposed system is able to achieve a data rate of 150 bps (spectrum efficiency of 5 bits/symbol) at a bit error rate of 6.76×10 -4 , which is well below the forward error correction limit of 3.8×10 −3 , over a link span of 1 m using a Huawei Nexus 6P smartphone with a frame rate of 30 fps. Keywords—Optical camera communications, undersampled pulse width modulation, visible light communications I. INTRODUCTION As superior imaging sensors (ISs) and faster smartphone processors are becoming commonplace, the latest smartphones are as powerful as professional digital single-lens reflex (DSLR) cameras on the photography side. In Dec. 2017, the top 3 ranked mobiles from DxOMark are Google Pixel 2, Apple iPhone X, and Huawei Mate 10 Pro with an overall score of 98, 97, and 97, respectively [1]. According their product specifications, each of these devices are capable of 720p or 1080p video capture at up to 240 frames per second (fps). Such an ability to capture high speed video streams by a smartphone paves the way for future optical camera communication (OCC) technology, mostly for low speed applications such as vehicular communications and indoor positioning [2]. However, most of the cameras embedded in smartphones used today are low frame rate (LFR) based cameras, and the most common frame rates (FRs) are 24, 30, 50 and 60 fps. According to the Nyquist sampling theorem, if these FRs are adopted for sampling, the transmitted symbol rate R s must be lower than half the sampling rate. However, this will clearly lead to light flickering due to the response time of the human eye. Therefore, a number of techniques have been proposed to support non-flickering OCC using low speed cameras (e.g., ≤ 60 fps). More precisely, there are three main modulation categories for LFR-based OCC using both global shutter (GS) and rolling shutter (RS) digital cameras: i) display-based [3], ii) oversampled-based [4, 5], and iii) undersampled-based [6] modulation schemes. In the display-based OCC system, the data is transmitted via a screen, which displays either observable or unobservable video frames. For example in [3], the useful information is first encoded into 2D visual code, which is then displayed on the screen. The user receives the information by scanning the 2D visual code using a smartphone’s camera and a code scanner application software. Under the oversampled-based OCC system, we have i) polarization-based modulation, which uses a polarized light generator to emit polarized light with two different polarization states to represent the binary data [4]. Since the change of light polarization state will not be observed, flickering is avoided; and ii) the RS-based OCC system [5], where the signal is transmitted at a very high R s (e.g., several kBd, lower than the RS scanning rate) to eliminate flickering. At the receiver (Rx), a RS-based camera is used to capture the light signal. With the RS receiving mechanism, dark and bright lines, from which the transmitted signal can be extracted, are captured above a normal image frame. Unlike the oversampled method, in the undersampled-based OCC system, the subcarrier multiplexing scheme is adopted, where the subcarrier frequency higher than the critical flicker frequency (CFF) f cf of the human eye (i.e., ~100 Hz [7]) is used to provide constant illumination. Once the FR of the Rx is equivalent to R s of the transmitter (Tx), a camera is able to undersample the transmitted signal correctly. Many undersampled modulation schemes, along with their dedicated