High-accuracy positioning has gained significant interest for many use-cases across various domains such as industrial internet of things (IIoT), healthcare and entertainment. Radio frequency (RF) measurements are widely utilized for user localization. However, challenging radio conditions such as non-line-of-sight (NLOS) and multipath propagation can deteriorate the positioning accuracy. Machine learning (ML)-based estimators have been proposed to overcome these challenges. RF measurements can be utilized for positioning in multiple ways resulting in time-based, angle-based and fingerprinting-based methods. Different methods, however, impose different implementation requirements to the system, and may perform differently in terms of accuracy for a given setting. In this paper, we use artificial neural networks (ANNs) to realize time-of-arrival (ToA)-based and channel impulse response (CIR) fingerprinting-based positioning. We compare their performance for different indoor environments based on real-world ultra-wideband (UWB) measurements. We first show that using ML techniques helps to improve the estimation accuracy compared to conventional techniques for time-based positioning. When comparing time-based and fingerprinting schemes using ANNs, we show that the favorable method in terms of positioning accuracy is different for different environments, where the accuracy is affected not only by the radio propagation conditions but also on the density and distribution of reference user locations used for fingerprinting.

Wireless high-accuracy positioning has recently attracted growing research interest due to diversified nature of applications such as industrial asset tracking, autonomous driving, process automation, and many more. However, obtaining a highly accurate location information is hampered by challenges due to the radio environment. A major source of error for time-based positioning methods is inaccurate time-of-arrival (ToA) or range estimation. Machine leaning (ML) techniques emerged as potential solutions to mitigate ToA-related errors. However, existing ML-based solutions either employ a set of features representing channel measurements only to a limited extent, or account for only device-specific proprietary methods of ToA estimation. In this paper, we propose a convolutional neural network (CNN) to estimate and mitigate the errors of a variety of ToA estimation methods utilizing channel impulse responses (CIRs). Based on real-world measurements from two independent campaigns, the proposed method yields significant improvements in ranging accuracy (up to 37%) of conventional ToA estimators, often eliminating the need of optimizing the underlying conventional methods.

High-accuracy positioning enables various applications such as industrial asset tracking, autonomous driving and process automation. Accurate location information relies on accurate time-of-arrival (ToA) or time-difference-of-arrival (TDoA) estimation in widely utilized time-based ranging. In this paper, we propose artificial neural network (ANN) based methods either to estimate T(D)oA directly or to mitigate the error of the conventional estimators. Based on real-world channel measurements, we show that the proposed direct ANN estimator outperforms the conventional estimators at least by approximately 37% and 24% in the 90th percentile ranging error, derived from ToA and TDoA estimations, respectively. Additionally, the proposed T(D)oA error-mitigating ANNs outperform the benchmark error mitigation methods with a gain varying between 17–43% in the 90th percentile ranging error, depending on the underlying conventional estimator. The ranging accuracy delivered by the direct estimation and error mitigation methods using ANNs are similar. Furthermore, the ANN estimators yield a more robust performance than the conventional estimators when the carrier frequency of the positioning signal is varied. ANN-based ToA estimation yields a marginally better ranging accuracy than ANN-based TDoA estimation. This advantage comes at the expense of a larger communication latency, while avoiding the need for synchronization among the positioning anchors.

Location information is essential for a wide range of applications requiring, for example, highly accurate positioning information such as industrial automation, autonomous driving, inbound logistics and augmented reality. One of the major error sources in positioning is non-line-of-sight (NLOS) propagation while a line-of-sight (LOS) propagation is anticipated. Therefore, classifying positioning measurements as LOS or NLOS plays a key role to enable high accuracy positioning use cases.Existing solutions for the classification task either rely on availability of label or reference measurement data or do not consider effects of different scenarios or channel models. In this paper, we propose an unsupervised method for the classification task through a channel feature selection process to select only useful channel features to be used for the classification. Then, we show based on real-world data from several measurement campaigns that the proposed method outperforms the existing solutions both in classification performance and ranging accuracy.

Mobile communication systems need to adapt to temporally and spatially changing mobile network traffic, due to dynamic characteristics of mobile users, in order to provide high quality of service. Since these changes are not purely random, one can extract the deterministic portion and patterns from the observed network traffic to predict the future network traffic status. Such prediction can be utilized for a series of proactive network management procedures including coordinated beam management, beam activation/deactivation and load balancing. To this end, in this paper, an intelligent predictor using artificial neural networks is proposed and compared with a baseline scheme that uses linear prediction. It is shown that the neural network scheme outperforms the baseline scheme for relatively balanced data traffic between highly random and deterministic mobility patterns. For highly random or deterministic mobility patterns, the performance of the two considered schemes is similar to each other.