In light of the prevailing limitations of existing indoor positioning methods, including high costs, inadequate positioning accuracy, and susceptibility to external environmental interferences, visible light communication (VLC) using white light LEDs has caught increasing attention as a sustainable and efficient communication method. Owing to the low cost, high efficiency, and extended lifespan of LEDs, indoor VLC positioning technology has emerged as a novel research field. In indoor VLC systems, the layout of light sources is closely related to indoor positioning accuracy. First, it is essential to optimize the layout of the light sources to ensure that the illumination in every corner of the room meets the requirements for both lighting and communication. Second, at the receiving end, it is also crucial to optimize the existing fingerprint positioning algorithms as much as possible and then minimize the average positioning error of the test surface and enhance positioning accuracy. By conducting spatial optimization, positioning LED light sources at appropriate emission locations not only meets the demands for illumination but also improves indoor positioning accuracy. By improving the existing fingerprint positioning algorithms at the receiver end, the average indoor positioning error is reduced. Therefore, in indoor visible light positioning (VLP) systems, the spatial optimization and algorithm improvement are significant for enhancing indoor positioning accuracy.

To address the aforementioned challenges, we introduce a novel indoor visible light positioning method based on spatial optimization. Initially, the Cramer-Rao bound (CRB) for the test surface is derived, and under the constraints of at least meeting indoor lighting requirements, the optimal layout of LED light sources is simulated by adopting an iterative algorithm. After establishing the optimal light source layout at the transmitter end, the K value associated with the minimum average positioning error is determined by comparing the average positioning errors of the weighted K-nearest neighbor (WKNN) algorithm and the K-nearest neighbor (KNN) algorithm across various numbers of nearest neighbors. To make the distance metric represented by received signal strength (RSS) closer to the actual distance, we should consider the relationship between the actual measurement target and the distance from the LED transmitter. Therefore, based on the received signal strength of the actual measurement target, different weights are assigned to make the RSS-based distance metric more consistent with the actual situation. Compared to the KNN algorithm and the original WKNN algorithm, the improved algorithm significantly enhances the positioning accuracy of indoor visible light positioning systems.

The initial step involves deriving the CRB for the surface to be tested, leading to the identification of the most efficient LED light source layout for optimal localization performance (Fig. 5). The accuracy of this theoretical approach is validated via an iterative algorithm, which compares the light source position coordinates at (1.3 m, 1.3 m) against (1.0 m, 1.0 m). This comparison supported by simulation confirms the correctness of our theoretical derivation (Fig. 6). Table 2 lists the specific parameters of the indoor VLC system. Meanwhile, we compare the average positioning errors of two algorithms at different KNN counts, determining that the WKNN algorithm exhibits the smallest average positioning error under the nearest neighbor number of three (Fig. 8). Subsequently, we compare the average positioning errors and the cumulative probability distributions of positioning errors of three algorithms under different signal-to-noise ratios. Simulation results indicate that the improved algorithm yields an average positioning error of 0.174 m (Fig. 9), representing an increase in average positioning accuracy of 51.12% and 23.34% compared to the KNN and WKNN algorithms respectively.

We initially explore the transmission characteristics of visible light signals in indoor environments and analyze the unique advantages demonstrated by indoor positioning technologies based on visible light communication compared to traditional techniques. The results indicate that the layout of LED light sources significantly influences indoor positioning accuracy. Under the premise of meeting indoor lighting requirements, we derive the CRB for the test surface in a simulated indoor visible light environment, thereby optimizing the layout of the LED light source transmitters. Additionally, the Gauss-Newton algorithm is employed for the iterative estimation of the proposed model. The precision of the theoretical derivations is confirmed by simulation involving two distinct arrangements of light sources and test points to demonstrate the model’s robustness and applicability in varied lighting scenarios. Additionally, we build upon existing location fingerprint algorithms by comparing the performance of the WKNN algorithm with the traditional KNN algorithm. The simulation results indicate that the WKNN algorithm significantly outperforms the KNN algorithm in terms of positioning accuracy when K is 3, thereby demonstrating the effectiveness of the WKNN approach in enhancing location determination accuracy. By making certain improvements and optimizations to the WKNN algorithm, different weights are assigned to different received signal strength differences based on the attenuation characteristics of visible light signals. Simulation results show that the improved algorithm reduces the average positioning error to 0.174 m, enhancing positioning accuracy by 51.12% and 23.34% compared to the original KNN and WKNN algorithms respectively. This significant improvement substantially increases the positioning accuracy of indoor visible light positioning systems.