High-Precision Indoor Visible Light Positioning Using Deep Neural Network Based on the Bayesian Regularization With Sparse Training Point

Abstract

In this letter, we propose an indoor visible light positioning technique that combines deep neural network based on the Bayesian Regularization (BR-DNN) with sparse diagonal training data set. Unlike other neural networks, which require a large number of training data points to locate accurately, we realize the high precision positioning with only 20 training points in a 1.8 m × 1.8 m × 2.1 m location area. Furthermore, we test a new optimization method of training data set, which is the diagonal set. To verify our ideas, we experimentally demonstrate three different training data acquisition methods that contain the common choice of training points (even set), arbitrary selection (arbitrary set), and diagonal selection (diagonal set). Experimental results show that the average localization accuracy optimized by the BR-DNN is 3.40 cm with the diagonal set, while the average localization accuracy is 4.35 cm for the arbitrary set and 4.58 cm for the even set. In addition, the training time and positioning time are only 11.25 and 8.66 ms due to a significant reduction of the sparse training data. All of the aforementioned experimental results show that the algorithm and training data optimization we proposed provide a new solution for real-time and high-accuracy positioning with the neural network.

Experimental Evaluation of Machine Learning Methods for Robust Received Signal Strength-Based Visible Light Positioning

“…This is where data-driven approaches can offer advantages. One of these models, used in the context of VLP, is the MLP model [21,22], of which the structure is shown in Figure 1. An MLP consists of at least three layers: an input layer (represented by vector z 0 ) containing the features (relative intensities), one or more hidden layers (represented by vector z i ), and an output layer (represented by vector z o ) (predicting the position).…”

“…The classical VLP schemes have limitations, as often they require close to ideal behavior of the propagation model and its parameters to perform well. Recently, however, it was demonstrated that Machine Learning (ML) techniques like for example Artificial Neural Networks (ANNs) are capable of delivering accurate localization results in the context of VLP [21][22][23][24][25], as they can handle noise naturally and are not bounded by ideal physical behavior. Furthermore, linear interpolation techniques can be used to reduce the amount of required measured data [26].…”

A Closed-Form Location Algorithm Without Auxiliary Variables for Moving Target in Noncoherent Multiple-Input and Multiple-Output Radar System

“…The indirect method could be also characterized in different types of parameters that are applied in localization such as TDOA [31]- [34], FDOA [32] and AOA [33]. Except for distance or angle based positioning methods mentioned above, there are feature based or neural network based indoor positioning methods [35]- [37]. Usually, in neural network based method, measurements of RSSI (Received Signal Strength Indicator) are used to establish a database and then the database can be applied for the learning of neural network to locate the target.…”

High-Efficient and Low-Cost Biased Multilevel Modulation Technique for IM/DD-Based VLP Systems

“…Through an in-depth analysis in this subsection, it was confirmed that most of the results were affected by the LED placement. Meanwhile, some nodes (10,15,20,24), which were judged to have a relatively insignificant impact, showed much better performance than other nodes. This indicates that although the results of this experiment are excellent, if the structure of the transmitter is supplemented with a completely symmetrical LED arrangement or a single LED that is capable of controlling the intensity is used, better results can be obtained.…”

“…For example, in [17], [18], high accuracy three-dimensional positioning was studied by applying an artificial neural network (ANN), but due to a long operation time and a large amount of computation, these methods are difficult to apply to industry. In [19], [20], the speed of ANN was improved by efficiently reducing the data or optimizing the structure, however, the burden of collecting training data and adjusting detailed parameters, which is a problem of pre-learning-based machine learning, is still required. In [21], Cai et al proposed the positioning system based on modified particle swarm optimization (PSO) algorithm that shows high localization accuracy and significantly lower algorithm complexity.…”