UWB Indoor Localization Using Deep Learning LSTM Networks

Abstract

Localization using ultra-wide band (UWB) signals gives accurate position results for indoor localization. The penetrating characteristics of UWB pulses reduce the multipath effects and identify the user position with precise accuracy. In UWB-based localization, the localization accuracy depends on the distance estimation between anchor nodes (ANs) and the UWB tag based on the time of arrival (TOA) of UWB pulses. The TOA errors in the UWB system, reduce the distance estimation accuracy from ANs to the UWB tag and adds the localization error to the system. The position accuracy of a UWB system also depends on the line of sight (LOS) conditions between the UWB anchors and tag, and the computational complexity of localization algorithms used in the UWB system. To overcome these UWB system challenges for indoor localization, we propose a deep learning approach for UWB localization. The proposed deep learning model uses a long short-term memory (LSTM) network for predicting the user position. The proposed LSTM model receives the distance values from TOA-distance model of the UWB system and predicts the current user position. The performance of the proposed LSTM model-based UWB localization system is analyzed in terms of learning rate, optimizer, loss function, batch size, number of hidden nodes, timesteps, and we also compared the mean localization accuracy of the system with different deep learning models and conventional UWB localization approaches. The simulation results show that the proposed UWB localization approach achieved a 7 cm mean localization error as compared to conventional UWB localization approaches.

Improving the Accuracy of Decawave’s UWB MDEK1001 Location System by Gaining Access to Multiple Ranges

“…An important contribution of the paper is that it demonstrates with two very different deployments how the same approach (same sensors, EKF parameters, and NLOS models) can be applied in totally different scenarios with the same rate of performance improvement. The need of learning or adaptation in the algorithms for different scenarios (common in many research papers [ 15 , 16 ]), is something that we were able to avoid, which is very convenient and practical in real life applications.…”

“…This kind of temporal in-range median-based solutions can circumvent the presence of sporadic outliers but fail when those errors are systematic. Other approaches that try to cancel outliers on the individual ranges, before the trilateration, are based on machine learning (ML) methods, such as k-nearest neighbors, Gaussian Processes, or Neural Networks [ 14 , 15 , 16 ]. However, methods based on learning are in many occasions invalid when changing the location site or if the conditions in the space change with time.…”

“…We will analyze the gain in accuracy obtained with the proposed multi-range (8-range) solution, and demonstrate how the same approach (same sensors, EKF parameters, and NLOS models) can be applied in totally different scenarios with the same rate of performance improvement, which is one of the contributions of the paper, apart from the proposal of a way to make possible the use of multiple ranges in one of the most used and successful pieces of UWB equipment in research labs and corporate solutions. The need of learning or adaptation in the algorithms for different scenarios (common in many research papers [ 14 , 15 , 16 ]) is something that we were able to avoid, which is very convenient and practical in real life applications.…”

Deep Gated Recurrent Unit-Based 3D Localization for UWB Systems

“…[34] An LSTM network-based localization method, which used distance between transmitter and receivers as input, was proposed. [35] ated with UWB, such as high cost, short range, and complex hardware installation.…”

“…Long short-term memory (LSTM) [29] and gated recurrent unit (GRU) [30] are two widely used RNN architectures because they can efficiently capture long term dependencies, as well as mitigate vanishing or exploding gradients in training [31]. Several RNN-based localization methods have been developed [32]- [35]. Compared with GRU, LSTM structure operates well with complicated time series data where multiple time steps must be considered, because it contains three gates (input, output and forget) with a cell state [31], [36].…”