This paper proposes a novel indoor optical wireless positioning (IOWP) framework that aims to enhance localization precision and robustness through an advanced machine learning (ML)-driven fusion technique. Unlike traditional single-model approaches, the proposed framework uses received signal strength (RSS) data to intelligently combine multiple lightweight ML algorithms, including K-Nearest Neighbors (KNN), Random Forest (RF), and Gaussian Process Regression (GPR). In the training phase, our system utilizes a performance-optimized weight allocation strategy to identify the optimal weights, harnessing the complementary strengths of individual models while mitigating their limitations to achieve exceptional generalization in complex indoor environments. A comprehensive evaluation is conducted under a realistic ray-traced channel model that incorporates typical light distributions, high-order multipath reflections from walls and objects, and mixed diffuse-reflective surface interactions. Performance is assessed in terms of mean positioning error (MPE), 90th percentile (P90) error, and computational complexity. Results demonstrate that the proposed method achieves an MPE of 0.5 cm and a P90 error below 1 cm, offering a practical and scalable solution for next-generation IOWP applications in smart environments.

In this paper, two techniques including uplink positioning and multiple-input multiple output (MIMO) singular value decomposition (SVD) beamforming are proposed and utilized to provide a novel physical layer security (PLS) scheme for visible light communication (VLC) systems, which much enhance the integrated location-sensing and data security strength. The paper first develops a closed-form mathematical model for ubiquitous uplink positioning using a single light-emitting diode with the trilateration algorithm, considering the field-of-view constraints due to the user’s optical transmitter and access point receivers. SVD in MIMO VLC is subsequently introduced to exploit the acquired user’s location to form a pair of public and private keys for secure data transfer over the physical medium. The proposed system performance is evaluated in terms of the root mean square positioning error$\varepsilon _{\mathbb {X}}$and the achieved bit error rate (BER) for the uplink positioning and the downlink data transmission, respectively. Results show that the effective area of the photodiode significantly impacts$\varepsilon _{\mathbb {X}}$, with smaller areas resulting in higher$\varepsilon _{\mathbb {X}}$. It is also observed that the$\varepsilon _{\mathbb {X}}$directly relates to the strength of ambient light inducing noise in the system. However, the proposed uplink positioning is less prone to natural ambient light as compared to the traditional downlink visible light positioning. The BER analysis of the downlink data transmission shows that the proposed SVD-based PLS enhances the system’s confidentiality by significantly narrowing the communication zone area by over 90% in comparison to the zero-forcing beamforming scheme and achieves reasonable secrecy capacity.

In this work, a new indoor visible light positioning algorithm is proposed based on support vector machines (SVM) and polynomial regression. Two different multipath environments of an empty room and a furnished room are considered. The algorithm starts by addressing the received power distance relation, considering polynomial regression models fitted to the specific areas of the room. In the second stage, an SVM is used to classify the best-fitted polynomial, which is used with nonlinear least squares to estimate the position of the receiver. The results show that, in an empty room, the positioning accuracy improvement for the positioning error, εpof 2.5 cm are 36.1, 58.3, and 72.2 % for three different scenarios according to the regions’ distribution in the room. For the furnished room, a positioning relative accuracy improvement of 214, 170, and 100 % is observed for εpof 0.1, 0.2, and 0.3 m, respectively.

This paper addresses the effects of blocking and shadowing due to peoples’ movement on the performance of visible light positioning system based on received signal strength. We consider both line-of-sight and non-line-of-sight paths and using a non-sequential ray-tracing model, the channel is characterised with and without blocking and shadowing. A linear least square with a polynomial regression model is proposed for the position estimation. We show that, the VLP accuracy degrade by ~90 and ~67% due to blocking and shadowing, respectively. Besides, at a confidence interval of 90%, the positioning error εpof 1.47 m is recorded with human blockage as compared with 0.15 m for the non-blocking cases while shadowing introduces εpof 0.82 m compared with 0.27 m for the non-blocking case.