Using silicon photovoltaic cells and machine learning and neural network algorithms for visible-light positioning systems

Indoor positioning systems based on visible light communication (VLC) using white light-emitting diodes (WLEDs) have been widely studied in the literature. In this paper, we present an indoor visible-light positioning (VLP) system based on red–green–blue (RGB) LEDs and a frequency division multiplexing (FDM) scheme. This system combines the functions of an FDM scheme at the transmitters (RGB LEDs) and a received signal strength (RSS) technique to estimate the receiver position. The contribution of this work is two-fold. First, a new VLP system with RGB LEDs is proposed for a multi-cell network. Here, the RGB LEDs allow the exploitation of the chromatic space to transmit the VLP information. In addition, the VLC receiver leverages the responsivity of a single photodiode for estimating the FDM signals in RGB lighting channels. A second contribution is the derivation of an expression to calculate the optical power received by the photodiode for each incident RGB light. To this end, we consider a VLC channel model that includes both line-of-sight (LOS) and non-line-of-sight (NLOS) components. The fast Fourier transform (FFT) estimates the powers and frequencies of the received FDM signal. The receiver uses these optical signal powers in the RSS-based localization application to calculate the Euclidean distances and the frequencies for the RGB LED position. Subsequently, the receiver’s location is estimated using the Euclidean distances and RGB LED positions via a trilateration algorithm. Finally, Monte Carlo simulations are performed to evaluate the error performance of the proposed VLP system in a multi-cell scenario. The results show a high positioning accuracy performance for different color points. The average positioning error for all chromatic points was less than 2.2 cm. These results suggest that the analyzed VLP system could be used in application scenarios where white light balance or luminaire color planning are also the goals.

With the increasing demand for indoor positioning, visible light positioning (VLP) systems have become an optics technologies research hotspot due to their high accuracy and avoidance of electromagnetic interference. However, current photodiode (PD)-based and image sensor (IS)-based VLP systems still suffer from high costs at the receiver, and the problem of the trade-off between field of view (FoV) and positioning accuracy. Therefore, we propose a smartphone-based low-cost, and high-precision VLP system, Acpro. The system combines the smartphone’s embedded ambient light sensor and IS as the receiver for the positioning. First, for the low sampling rate of the ambient light sensor, we use the frequency aliasing characteristic to identify the LED’s beacon through the dynamic undersampling principle. Then, to increase the FoV and positioning accuracy, we use the standard distance information corrected by the IS to improve the noise distance calculated by the ambient light sensor. Further, the corrected distance information is used for positioning. Extensive results show that our system can achieve sub-meter positioning accuracy.

As a promising approach to implement indoor positioning, visible light positioning (VLP) based on optical camera communication (OCC) image sensor has attracted substantial attention. However, the decoding schemes of existing VLP systems still face many challenges. First, the transmission channel between transmitters and receivers can be easily affected by environmental changes, resulting in poor thresholding performance. Second, the inherently unsynchronized air transmission channel issue remains a big obstacle for decoding data. The above two problems limit the application of VLP systems, where various mobile devices are used as receivers and the properties of transmission channel are constantly changing with the movement of receivers. In this paper, a universal and effective decoding scheme named pixel-to-bit calculation (PBC) decoding algorithm for VLP systems is proposed and experimentally demonstrated. It includes a Staged Threshold Scheme which provides excellent thresholding performance for different transmission channel conditions, as well as a Synchronous Decoding Operation to automatically synchronize the clock between transmitters and receivers. A decoding rate of 95.62% at the height of 2.73 m is realized in a practical Robotic-based VLP system embedded with our proposed PBC decoding scheme. In addition, experimental results show that the average decoding rate of the proposed PBC decoding scheme reaches 99.9% when applying different transmitters and receivers.

With the booming development of green lighting technologies, ubiquitous lighting infrastructures make visible light positioning (VLP) become a promising technology for indoor localization and navigation. However, most existing VLP systems use modulated LEDs as anchors for target localization. LEDs with its customized hardware and modulated information greatly prevent VLP systems from large-scale deployment and limit their widespread applications. Moreover, some VLP systems have the disadvantage of imposing strong restrictions on smartphone such as fixed phone orientation, which is not practical for real-life use. In this paper, we propose a ubiquitous visible light localization (LiLoc) algorithm using ambient lights based on augmented particle filter algorithm. LiLoc directly uses the light intensity fingerprint of ambient light sources (e.g., fluorescent, incandescent, and LEDs) to estimate target location without obtaining modulation information of light sources in buildings. To improve localization accuracy, we introduce a light-intensity trajectory fingerprint matching scheme to calibrate target position. We also develop a beacon identification algorithm based on the inherent visual features of light sources, which are determined by specific manufacturing variations. We combine beacon identification algorithm with PDR to estimate target position in terrible illumination environment (e.g., sunlight interference, beacons sparse deployment, individual beacon damaged/closed, or training fingerprint missing).To reduce power consumption, we design adaptive scheduling mechanism to invoke corresponding algorithms based on the detection result of sunlight interference. The experimental results demonstrate that LiLoc can achieve medium localization errors of 2.2 m in a complex lighting environment with various light sources and sunlight interference.

Visible light positioning (VLP) has been studied widely due to its high accuracy and low cost in the field of location-based services (LBS). However, many existing VLP systems have the requirements that the receiver should be placed horizontally and more than three LED lamps should be used, which are difficult to meet in practical scenarios. Therefore, it is necessary to develop a novel VLP algorithm for tilted conditions. An effective and simple VLP system while the receiver is tilted based on double-LED lamps is proposed in this paper. The vertical position can be determined by combining the information from angle sensors with geometric information. Through analyzing the imaging characteristics of the tilted state, we can utilize the relationship of similarity to calculate the location of the mobile receiver. Experimental results show that the positioning accuracy of our proposed algorithm can reach 5.48 cm.

Angle-of-arrival (AOA) estimation of optical signals is a key technique in visible light positioning (VLP) systems, which can be realized using photodiode (PD) arrays. However, most of existing works do not properly design the angles of the PDs, which hampers the accuracy of VLP systems. In this work, we optimize the PD array to achieve the highest AOA estimation accuracy. Specifically, we achieve the optimal angle combination of the PDs in order to minimize the power of the additive Gaussian noise imposed on the exact incident vector. Closed-form expressions are derived for the optimal angles of the PD array. Our analytical results show that improper placement of PDs can cause significant accuracy loss. Both simulation and experimental results show that the optimized PD array can suppress the angle estimation noise. The theoretical results provide a benchmark for the design of PD arrays, and lay a foundation for further studies on error analysis of VLP.

The accuracy of the visible light positioning (VLP) system is improved by using regression-based and interpolation-based approaches. These approaches are mixed with the received signal strength in both linear and nonlinear least squares techniques. The simplicity of the model is the key point, where a highly accurate VLP system with only one equation “interpolation” is used instead of many complex equations. Gaussian noise and reflections up to three bounces with high reflectivity are the source of noise. In addition, the strength of the methods is studied by changing some affecting parameters. In a previous work, we introduced the regression-based approach in VLP-based systems with one reflection. Here, we extend the work with three reflections, leading to an increase in noise to examine system robustness. The approaches are compared to the traditional ones, and show a remarkable enhancement, e.g., the average error has been improved by 67%.

In this letter, a quasi-gapless integrated visible light communication and positioning (VLCP) system is experimentally demonstrated for the first time, based on filter bank multicarrier- based subcarrier multiplexing (FBMC-SCM). Compared with orthogonal frequency division multiplexing-based SCM (OFDM- SCM), FBMC-SCM has lower out-of-band interference (OOBI) and therefore requires much smaller guard band (GB) spacing. A low-complexity transmit diversity transmission scheme is adopted here, where all the light emitting diode (LED) lamps transmit the same communication signal.Moreover, phase difference of arrival (PDOA) is applied for positioning. The experimental results show that, in a coverage area of 1.2×1.2 m2 with a height of 2.1 m, the mean position errors using OFDM–SCM and FBMC-SCM for a GB of 0.7 MHz are 10.91 and 6.08 cm, respectively. Moreover, a comparable bit error rate (BER) performance can be achieved for both OFDM–SCM and FBMC-SCM. Due to the negligible GBs when using FBMC-SCM, the effective bandwidth utilization ratio of the integrated VLCP system is improved from 72% to 98% when OFDM-SCM is replaced by FBMC-SCM.

Train position information is the key factor to ensure the safe and efficient operation of trains for CBTC (Communications Based Train Control) systems. Aiming at the unique characteristics of tunnel environment and train operation for urban rail transit, a train positioning method using VLC(visible light communication)and RSSI (Received Signal Strength Indication) has been proposed, and the visible light communication train positioning system is established. Firstly, based on the classical illumination model of Lambert, a train positioning model with LED lamps for CBTC systems has been established. Secondly, The visible light communication positioning system is designed. Thirdly, the linear distance between the LED lamps and the train on the tunnel wall is calculated by using the RSSI value, and the train positioning results are fitted by using the trilateral positioning principle and the least square method. Finally, according to the real line data and equipment parameters from Chengdu Metro Line 1, the feasibility and effectiveness of the proposed method has been verified. The simulation results show that the maximum positioning error is 9. 4cm, and the average positioning error is less than 10cm,all of these meet the CBTC train positioning requirements. VLC-based positioning can be used as a new choice of train positioning method.

In visible light positioning (VLP) systems, factors like multipath reflection and noise interference degrade positioning accuracy. To address this challenge, this letter innovatively applies a Transformer-Encoder model to a VLP system with a dual-LED and multi-photodiode (PD) architecture. The proposed Transformer-Encoder model captures the spatial distribution information of the PDs through a Positional Encoding module. Its core Multi-Head Attention mechanism, incorporating positional information, enables the model to focus on critical channel features. This significantly enhances the model's robustness against multipath interference and noise while strengthening its capability to characterize channel features. Simulation results demonstrate that within a 4 m × 4 m × 3 m space, the Transformer-Encoder model achieves an average positioning error of 0.75 cm, with 90% of errors below 1.27 cm. Comparative analysis with other positioning models confirms the high precision of the proposed method.

This article reports a novel visible light positioning (VLP) system and the associated experimental results. The developed VLP system is completely passive as it does not require a tracked object to carry any active device or tag. At the same time, it does not require any modification to the existing lighting infrastructure. The positioning system, termed Watchers on the Wall (WoW), localizes a target based on measuring the change it creates in the received signal strength (RSS) of the ambient light recorded at an array of light-sensors embedded in the wall. A prototype system has been implemented and tested to investigate the performance of the proposed approach with regard to localization and tracking. The experimental results show that the median and 90-percentile localization errors of 7 and 21 cm, respectively, can be achieved for a 2 m × 3.6 m testbed. The effect of various parameters like the height of placement and the number of light-sensors, as well as the size of the fingerprint database, has also been studied. The impact of various distance metrics on the localization performance of the weighted K-nearest neighbor (WKNN) classifier has been investigated. It has been found that two distance metrics outperform the commonly employed Euclidean metric. The experimental results also demonstrated that the developed system could track a mobile target along multiple routes with a median error of 12 cm.

In recent years, several indoor positioning systems have been extensively studied for environments where a satellite signal is typically degraded and interrupted. Among them, visible light positioning (VLP), which has several advantages such as the absence of electromagnetic interference, energy efficiency, and high bandwidth availability, has received considerable attention. However, most VLP systems are established based on two-dimensional (2D) positioning methods in which the height of the receiver is fixed or restricted and the tilt of the receiver is not considered. To solve these problems, we propose a new positioning method that can be applied to three-dimensional (3D) space. We first formulate a new mathematical model for a 3D VLC positioning system and derive the channel gain as a function of source and receiver location in Cartesian coordinates. By employing the cost function developed using the induced channel gain, we demonstrate that the proposed 3D method is more accurate compared to the 2D method in experiments. The quantitative results show an average error of 7.95 cm, in a 2.5 m × 2.5 m × 3 m region.

ABSTRACTUsing deep learning to improve the accuracy of indoor visible light positioning (VLP) systems has gradually become a widely used research strategy in the field. However, current deep learning‐based indoor visible light localization algorithms have not been able to effectively mine the deep temporal and spatial sequence features in signals, resulting in complex network construction and low localization accuracy. To address this issue, the text proposes a deep learning framework that utilizes an attention mechanism to train a small number of randomly continuously sampled spatial received signals to predict the coordinates of the received signals and encode the spatiotemporal sequence attributes of the received signals as a feature into the data, constructed a highly reliable spatiotemporal sequence attention mechanism for indoor visible light localization method. Combined with Convolutional Neural Networks (CNN), the localization accuracy is further improved. Through simulation experiments, it has been verified that the neural network structure designed in this paper has better positioning accuracy compared to advanced algorithms, and can still achieve centimeter‐level (9.886cm) average positioning error under low signal‐to‐noise ratio (SNR) conditions. It is proved that the method proposed in this paper is reliable in the indoor VLP system.

In this paper, a positioning algorithm based on the combination of K-means clustering and deep neural networks (DNNs) is first presented for multiple light emitting diodes (LEDs) integrated with visible light positioning (VLP) systems. We extracted the maximum value from the collected optical power of LEDs, utilizing the ratio of each optical power to this maximum optical power as the input training data. The experimental results demonstrate that the proposed algorithm outperformed the conventional DNN algorithm in terms of anti-jamming capability and positioning accuracy. In addition, the positioning accuracy of the proposed system reached a millimeter level, which is the highest experimental VLP accuracy, to the best of our knowledge.

BackgroundDespite the interest in machine learning (ML) algorithms for analyzing real-world data (RWD) in healthcare, the use of ML in predicting time-to-event data, a common scenario in clinical practice, is less explored. ML models are capable of algorithmically learning from large, complex datasets and can offer advantages in predicting time-to-event data. We reviewed the recent applications of ML for survival analysis using RWD in healthcare.MethodsPUBMED and EMBASE were searched from database inception through March 2023 to identify peer-reviewed English-language studies of ML models for predicting time-to-event outcomes using the RWD. Two reviewers extracted information on the data source, patient population, survival outcome, ML algorithms, and the Area Under the Curve (AUC).ResultsOf 257 citations, 28 publications were included. Random survival forests (N = 16, 57%) and neural networks (N = 11, 39%) were the most popular ML algorithms. There was variability across AUC for these ML models (median 0.789, range 0.6–0.950). ML algorithms were predominately considered for predicting overall survival in oncology (N = 12, 43%). ML survival models were often used to predict disease prognosis or clinical events (N = 27, 96%) in the oncology, while less were used for treatment outcomes (N = 1, 4%).ConclusionsThe ML algorithms, random survival forests and neural networks, are mainly used for RWD to predict survival outcomes such as disease prognosis or clinical events in the oncology. This review shows that more opportunities remain to apply these ML algorithms to inform treatment decision-making in clinical practice. More methodological work is also needed to ensure the utility and applicability of ML models in survival outcomes.