A Deep Learning Approach to Position Estimation from Channel Impulse Responses

Niitsoo, Arne; Edelhäußer, Thorsten; Eberlein, Ernst; Hadaschik, Niels; Mutschler, Christopher

doi:10.3390/s19051064

Cited by 49 publications

(50 citation statements)

References 76 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…II shows the gain of the proposed "Phase Branch" over traditional fingerprinting approaches, resulting in an improvement in measurements of 29 % by only adding 22000 weights. It can be seen that for the simulated data and measured data the NN achieves (with a total of only 440.000 weights) the same accuracy as in [19] for a similar area and propagation environment (cf. Fig.…”

Section: Proposed Neural Network Architecturementioning

confidence: 78%

See 1 more Smart Citation

Towards Practical Indoor Positioning Based on Massive MIMO Systems

Widmaier

Arnold

Dörner

et al. 2019

2019 IEEE 90th Vehicular Technology Conference (VTC2019-Fall)

View full text Add to dashboard Cite

We showcase the practicability of an indoor positioning system (IPS) solely based on Neural Networks (NNs) and the channel state information (CSI) of a (Massive) multiple-input multiple-output (MIMO) communication system, i.e., only build on the basis of data that is already existent in today's systems. As such our IPS system promises both, a good accuracy without the need of any additional protocol/signaling overhead for the user localization task. In particular, we propose a tailored NN structure with an additional phase branch as feature extractor and (compared to previous results) a significantly reduced amount of trainable parameters, leading to a minimization of the amount of required training data. We provide actual measurements for indoor scenarios with up to 64 antennas covering a large area of 80 m 2 . In the second part, several robustness investigations for real-measurements are conducted, i.e., once trained, we analyze the recall accuracy over a time-period of several days. Further, we analyze the impact of pedestrians walking in-between the measurements and show that finetuning and pre-training of the NN helps to mitigate effects of hardware drifts and alterations in the propagation environment over time. This reduces the amount of required training samples at equal precision and, thereby, decreases the effort of the costly training data acquisition.

show abstract

Section: Proposed Neural Network Architecturementioning

confidence: 78%

“…Tab. I gives an overview of the used datasets and their corresponding coverage areas, where the area dimensions match those in [19]. In the "LoS Weekdays Measurements" each day was measured with the same meander-like path structure, resulting in a sample distance of around 1 cm.…”

Section: B Measured Channelsmentioning

confidence: 99%

Towards Practical Indoor Positioning Based on Massive MIMO Systems

Widmaier

Arnold

Dörner

et al. 2019

2019 IEEE 90th Vehicular Technology Conference (VTC2019-Fall)

View full text Add to dashboard Cite

show abstract

“…To the best of our knowledge, all previous work use CIRs in combination with absolute time information to estimate a position with DL approaches. For example, the authors of [25][26][27] used convolutional neural networks (CNN) or other deep learning approaches and showed promising results, albeit with a very large amount of data and a complex network structure (at the cost of computational expense). A feature-based approach for discrete positioning, based on propagation models, has been proposed in [1].…”

Section: Literature Reviewmentioning

confidence: 99%

“…While the simplified models introduced in Section 3 only cover a part of the actual propagation effects, they show that, apart of the ToA (which, usually, corresponds to the time of the first correlation peak above a certain threshold), a lot of additional spatial information related to the environment is contained in CIRs. While this information has been used by deep neural networks and using complete CIRs [26,27], this article proposes a more compact and comprehensible representation of the information by feature extraction. In the following, some approaches to represent this information are identified.…”

Section: Deduction Of Relevant Featuresmentioning

confidence: 99%

UWB Channel Impulse Responses for Positioning in Complex Environments: A Detailed Feature Analysis

Kram

Stahlke

Feigl

et al. 2019

Sensors

View full text Add to dashboard Cite

Radio signal-based positioning in environments with complex propagation paths is a challenging task for classical positioning methods. For example, in a typical industrial environment, objects such as machines and workpieces cause reflections, diffractions, and absorptions, which are not taken into account by classical lateration methods and may lead to erroneous positions. Only a few data-driven methods developed in recent years can deal with these irregularities in the propagation paths or use them as additional information for positioning. These methods exploit the channel impulse responses (CIR) that are detected by ultra-wideband radio systems for positioning. These CIRs embed the signal properties of the underlying propagation paths that represent the environment. This article describes a feature-based localization approach that exploits machine-learning to derive characteristic information of the CIR signal for positioning. The approach is complete without highly time-synchronized receiver or arrival times. Various features were investigated based on signal propagation models for complex environments. These features were then assessed qualitatively based on their spatial relationship to objects and their contribution to a more accurate position estimation. Three datasets collected in environments of varying degrees of complexity were analyzed. The evaluation of the experiments showed that a clear relationship between the features and the environment indicates that features in complex propagation environments improve positional accuracy. A quantitative assessment of the features was made based on a hierarchical classification of stratified regions within the environment. Classification accuracies of over 90% could be achieved for region sizes of about 0.1 m 2 . An application-driven evaluation was made to distinguish between different screwing processes on a car door based on CIR measures. While in a static environment, even with a single infrastructure tag, nearly error-free classification could be achieved, the accuracy of changes in the environment decreases rapidly. To adapt to changes in the environment, the models were retrained with a small amount of CIR data. This increased performance considerably. The proposed approach results in highly accurate classification, even with a reduced infrastructure of one or two tags, and is easily adaptable to new environments. In addition, the approach does not require calibration or synchronization of the positioning system or the installation of a reference system.

show abstract

“…For instance, Stefano Maran [17] evaluated the resulting performance based on various UWB propagation features through Monte Carlo simulations, and proposed an algorithm for identifying NLOS conditions and reducing the ranging error caused by NLOS conditions. The authors in articles [18,19] attempted to use the channel impulse response (CIR) from the antennas, together with ground truth locational data (derived by a robot equipped with an optical reference system), to train a deep convolutional neural network (CNN). Then, they used the CNN model to identify and mitigate the NLoS propagation.…”

mentioning

confidence: 99%

A Robust PDR/UWB Integrated Indoor Localization Approach for Pedestrians in Harsh Environments

Tong

Ning

et al. 2019

Sensors

View full text Add to dashboard Cite

Wireless sensor networks (WSNs) and the Internet of Things (IoT) have been widely used in industrial, construction, and other fields. In recent years, demands for pedestrian localization have been increasing rapidly. In most cases, these applications work in harsh indoor environments, which have posed many challenges in achieving high-precision localization. Ultra-wide band (UWB)-based localization systems and pedestrian dead reckoning (PDR) algorithms are popular. However, both have their own advantages and disadvantages, and both exhibit a poor performance in harsh environments. UWB-based localization algorithms can be seriously interfered by non-line-of-sight (NLoS) propagation, and PDR algorithms display a cumulative error. For ensuring the accuracy of indoor localization in harsh environments, a hybrid localization approach is proposed in this paper. Firstly, UWB signals cannot penetrate obstacles in most cases, and traditional algorithms for improving the accuracy by NLoS identification and mitigation cannot work in this situation. Therefore, in this study, we focus on integrating a PDR and UWB-based localization algorithm according to the UWB communication status. Secondly, we propose an adaptive PDR algorithm. UWB technology can provide high-precision location results in line-of-sight (LoS) propagation. Based on these, we can train the parameters of the PDR algorithm for every pedestrian, to improve the accuracy. Finally, we implement this hybrid localization approach in a hardware platform and experiment with it in an environment similar to industry or construction. The experimental results show a better accuracy than traditional UWB and PDR approaches in harsh environments.Sensors 2020, 20, 193 2 of 20 most important feature of harsh environments. Indoor localization technology is still very challenging under NLoS propagation in harsh environments.The current mainstream indoor localization systems are radio frequency (RF)-based. This kind of system basically requires beacon nodes, or beacon nodes (BNs), with known coordinates in the localization system. Each target to be located has an electronic tag, or tag node (TN). Specific features generated from the communication between the TN and BN are used for localization. These features mainly include the radio signal strength (RSS), time, signal angle, etc. The challenge of indoor localization systems in harsh environments is that these features can be severely disturbed [1] by NLoS propagation or other interference. As a result, the localization accuracy will be greatly reduced.In harsh environments, interference of an RF-based localization system can be mainly divided into three types. Firstly, multipath refraction is a common type of interference in harsh environments. It means that an RF signal will be transmitted to the receiver through multiple paths, which causes the transmission time to be unreliable and unpredictable RSS superposition. Additionally, it apparently has a greater impact on the signal angle. Secondly, non-line-of-sight (NLoS) propagati...

show abstract

A Deep Learning Approach to Position Estimation from Channel Impulse Responses

Cited by 49 publications

References 76 publications

Towards Practical Indoor Positioning Based on Massive MIMO Systems

Towards Practical Indoor Positioning Based on Massive MIMO Systems

UWB Channel Impulse Responses for Positioning in Complex Environments: A Detailed Feature Analysis

A Robust PDR/UWB Integrated Indoor Localization Approach for Pedestrians in Harsh Environments

Contact Info

Product

Resources

About