Andrei Szabo

WLAN-Based Pedestrian Tracking Using Particle Filters and Low-Cost MEMS Sensors

Indoor positioning systems based on wireless LAN (WLAN) are being widely investigated in academia and industry. Meanwhile, the emerging low-cost MEMS sensors can also be used as another independent positioning source. In this paper, we propose a pedestrian tracking framework based on particle filters, which extends the typical WLAN-based indoor positioning systems by integrating low-cost MEMS accelerometer and map information. Our simulation and real world experiments indicate a remarkable performance improvement by using this fusion framework.

Initialization and Online-Learning of RSS Maps for Indoor / Campus Localization

Common approaches for indoor positioning based on cellular communication systems use as measurements the received signal strength (RSS). In order to work properly, such a system often requires many calibration points before its start. This paper presents a two-fold approach achieving high indoor localization accuracies without requiring too many calibration points. The basic idea is to use an initial propagation model with few parameters, which can be adapted by a few measurements, e.g. mutual measurements of access points. Then the model is refined by incorporating additional parameters and using online learning. Investigations on the requirements and potentials of different approaches and results for DECT and WLAN setups are given. The first approach uses predefined paths that should be passed through by a service technician with measurement equipment. The second approach uses a Kohonen-like learning algorithm to adapt the model on-the-fly. For both approaches linear propagation models and more involved dominant path models incorporating map information are applied for the initialization.

Enhancing the map usage for indoor location-aware systems

Location-aware systems are receiving more and more interest in both academia and industry due to their promising prospective in a broad category of so-called Location-Based-Services (LBS). The map interface plays a crucial role in the location-aware systems, especially for indoor scenarios. This paper addresses the usage of map information in a Wireless LAN (WLAN)-based indoor navigation system. We describe the benefit of using map information in multiple algorithms of the system, including radio-map generation, tracking, semantic positioning and navigation. Then we discuss how to represent or model the indoor map to fulfill the requirements of intelligent algorithms. We believe that a vector-based multi-layer representation is the best choice for indoor location-aware system.

EM-based semi-blind channel estimation method for MIMO-OFDM communication systems

This paper presents a novel semi-blind method for channel estimation and symbol detection in multiple input multiple output-orthogonal frequency division multiplexing (MIMO-OFDM) digital communication systems over time-varying channels where the standard pilot-aided channel estimation (PACE) interpolation approaches are not directly applicable due to insufficient placement of pilot sequences over the frequency sub-channels. The presented channel estimation method utilizes the Expectation Maximization (EM) algorithm to exploit the information in the received signals and the statistical properties of the channel.

SPLL: Simultaneous Probabilistic Localization and Learning

Indoor localisation systems based on existent radio communication networks often make use of received signal strength (RSS) as measured feature. In order to achieve a good accuracy such systems have a huge payload in the called calibration phase, where many labelled measurements are collected and used to build a representative feature map. The present paper introduces a new algorithm based on previous works from the same authors, where the calibration phase is avoided by unsupervised online learning, during the operational phase of the system. Using probabilistic localisation and non-parametric density estimation, the new approach uses unlabelled measurements to learn a feature map, having as start only a rough initial model. Simulations with artificial generated data and with real measurements validate the introduced algorithm.

SLL: Relations to Kohonen SOMs

The simultaneous localization and learning (SLL) is an indoor localization technique based on existent radio communication networks. It originally takes received signal strength (RSS) as measured feature, used as input on an adaptive and iterative process based on Kohonen self organizing maps (SOMs) in order to learn and improve a feature map. The present paper points the main characteristics from both SLL and SOM, their differences and similarities. The somewhat generic formulation for SOMs acquire physical meanings with SLL that act as a constrainment, making the SLL a very particular case of SOM. The proofs for one dimensional SOMs are complemented by the proofs presented for the SLL by the authors in previous articles.

Adaptive Localization in Wireless Networks

1 Siemens AG, Automation and Drives, Advanced Technologies and Standards, Process Automation, A&D ATS 33, Oestliche Rheinbrueckenstr. 50, 76187 Karlsruhe, Germany henning.lenz@siemens.com 2 Siemens AG, Corporate Technology, Information and Communications, CT IC 4, Otto-Hahn-Ring 6, 81730 Munich, Germany betoni@gmail.com, {hui.wang.ext, andrei.szabo, joachim.bamberger}@siemens.com 3 Helmut-Schmidt-University / University of the Federal Armed Forces Hamburg, Department of Electrical Engineering, Institute for Control Engineering, Holstenhofweg 85, 22043 Hamburg, Germany joachim.horn@hsu-hh.de 4 Universitat Karlsruhe, Fakultat fur Informatik, Institut fur Technische Informatik, Kaiserstr. 12, 76128 Karlsruhe uwe.hanebeck@ieee.org

SLL: Statistical Conditions and Algebraic Properties

Common approaches for indoor positioning based on cellular communication systems use the received signal strength (RSS) as measurements. In order to work properly, such a system often requires many calibration points before its start. Applying simultaneous localization and learning (SLL) a self-calibrating RSS-based positioning system can be realized. Clearly, SLL avoids the requirement for manually obtained reference measurements. This paper explores the algebraic and statistical conditions required to perform the SLL approach. Firstly, as basis of the analysis a closed form of SLL is introduced. As main result of this paper the algebraic and statistical conditions are revealed that need to be satisfied such that SLL can successfully be utilized, leading to a self-calibration of RSS-based positioning systems. While the analysis is restricted to the one-dimensional case and although the extension of the analysis to higher dimensions is more complex, the results can straightforwardly be extended to the more-dimensional cases.

Unsupervised Learning of Propagation Time for Indoor Localization

Many of todays localization systems for indoor and outdoor positioning are based on propagation time measurements of radio signals. In order to achieve high positioning accuracy in presence of non line of sight (NLOS) propagation, these systems require either an expensive manual calibration or additional information. In this paper, we present a novel approach for a channel impulse response (CIR) based fingerprint system, which reduces the calibration and measurement effort and simultaneously improves localization results. The basic idea is to initialize a simple model, which is improved by an online learning procedure using unlabeled measurements. This unsupervised learning algorithm is composed of two independent components, which exploit the similarity of neighboring CIRs as well as the temporal relation of measurements. Our tests indicate a significant improvement compared to traditional methods in case of time difference of arrival (TDoA) measurements. The algorithm can straightforwardly be adapted to arbitrary propagation time measurements.

Indoor online learning of feature maps using SPLL

Many indoor localisation systems based on existent radio communication networks use the received signal strength (RSS) as measured feature. The accuracy of such systems is directly related to the amount of labelled data, gathered during a calibration phase. This paper explores the algorithm based on previous works from the same authors, where an explicit calibration phase is avoided applying un-supervised online learning, while the system is already operational. Using probabilistic localisation and non-parametric density estimation, this approach uses unlabelled measurements to automatically learn a feature map with the probabilistic distribution of the measurements, starting only with a rough initial model, based on plausible physical properties. A real example in a highly structured office environment validates the introduced algorithm, covering discontinuities on the feature map and the imposed multimodal distributions.