Simo Ali-Löytty

A comparative survey of WLAN location fingerprinting methods

The term “location fingerprinting” covers a wide variety of methods for determining receiver position using databases of radio signal strength measurements from different sources. In this work we present a survey of location fingerprinting methods, including deterministic and probabilistic methods for static estimation, as well as filtering methods based on Bayesian filter and Kalman filter. We present a unified mathematical formulation of radio map database and location estimation, point out the equivalence of some methods from the literature, and present some new variants. A set of tests in an indoor positioning scenario using WLAN signal strengths is performed to determine the influence of different calibration and location method parameters. In the tests, the probabilistic method with the kernel function approximation of signal strength histograms was the best static positioning method. Moreover, all filters improved the results significantly over the static methods.

Particle filter and smoother for indoor localization

We present a real-time particle filter for 2D and 3D hybrid indoor positioning. It uses wireless local area network (WLAN) based position measurements, step and turn detection from a hand-held inertial sensor unit, floor plan restrictions, altitude change measurements from barometer and possibly other measurements such as occasional GNSS fixes. We also present a particle smoother, which uses future measurements to improve the position estimate for non-real-time applications. A light-weight fallback filter is run in the background for initialization, divergence monitoring and possibly re-initialization. In real-data tests the particle filter is more accurate and consistent than the methods that do not use floor plans. An example is shown on how smoothing helps to improve the filter estimate. Moreover, a floor change case is presented, in which the filter is capable of detecting the floor change and improving the 2D accuracy using the floor change information.

Statistical path loss parameter estimation and positioning using RSS measurements in indoor wireless networks

A Bayesian method for dynamical off-line estimation of the position and path loss model parameters of a WLAN access point is presented. Two versions of three different on-line positioning methods are tested using real data. The tests show that the methods that use the estimated path loss parameter distributions with finite precisions outperform the methods that only use point estimates for the path loss parameters. They also outperform the coverage area based positioning method and are comparable in accuracy with the fingerprinting method. Taking the uncertainties into account is computationally demanding, but the Gauss-Newton optimization method is shown to provide a good approximation with computational load that is reasonable for many real-time solutions.

Fingerprint Kalman Filter in indoor positioning applications

In this paper, we present a new filter, the Fingerprint Kalman Filter (FKF), and apply it to indoor positioning. FKF enables sequential position estimation using WLAN RSSI measurements and fingerprint data. Fingerprints that are collected beforehand in a calibration phase contain samples of the radio map in certain points, namely, calibration points. This means that FKF does not need an analytic formula of the measurement equation like conventional Kalman-type filters do (e.g. the Extended Kalman Filter (EKF) and the Unscented Kalman Filter (UKF)). Like EKF and UKF, FKF is based on the recursive computation of the Best Linear Unbiased Estimator (BLUE) and has small computational and memory requirements. An often-used Kalman-type filter for this problem is so-called Position Kalman Filter (PKF) that uses static position solutions as measurements for the conventional Kalman filter. In the test part of the paper, we compare FKF, PKF and different static location estimation methods, namely, the Nearest Neighbor (NN) and the Kernel method. The test data is real WLAN RSSI measurement data. The results indicate that the filters give more accurate position estimates than the static methods. FKF performs better than PKF with NN as the static estimator, and the computational load of FKF is smaller than PKF with the Kernel method.

Statistical path loss parameter estimation and positioning using RSS measurements

An efficient Bayesian method for off-line estimation of the position and the path loss model parameters of a base station is presented. Two versions of three different on-line positioning methods are tested using real data collected from a cellular network. The tests confirm the superiority of the methods that use the estimated path loss parameter distributions compared to the conventional methods that only use point estimates for the path loss parameters. Taking the uncertainties into account is computationally demanding, but the Gauss-Newton optimization methods is shown to provide a good approximation with computational load that is reasonable for many real-time solutions.

Positioning with coverage area estimates generated from location fingerprints

We present a method to estimate the coverage areas of transmitters, and a method to use a database of such coverage areas for personal positioning. The coverage areas are modelled as ellipsoids, and their location and shape parameters are computed from reception samples (fingerprints) using Bayesian estimation. The position is computed as a weighted average of the ellipsoid centers, with weights determined by the ellipsoid shape parameters. The methods are tested using a subset of a prototype wireless sensor network (TUTWSN) consisting of 30 nodes deployed indoors on one floor of a university building. The network nodes use low power commercial off-the-shelf components including a 2.4GHz radio.

Received signal strength models for WLAN and BLE-based indoor positioning in multi-floor buildings

This paper investigates the similarities and differences of the signal strength fluctuations and positioning accuracy in indoor scenarios for three types of wireless area networks: two Wireless Local Area Networks (WLANs) at 2.4 GHz and 5 GHz frequency, respectively, and one Wireless Personal Area Network (WPAN), namely the Bluetooth Low Energy (BLE). Two path-loss models based on weighted centroids and non-negative least squares estimation are presented: one including a floor loss factor, and the other one ignoring the floor losses, and the three signal types are compared in terms of the path-loss parameters, channel fluctuations and positioning accuracy, namely the distance errors and floor detection probabilities. The comparison is done based on real-field measurement data collected from a university building in Tampere, Finland. It is shown that all these three signal types have similar shadowing variances and close path-loss parameter values, and that a path-loss model considering floor losses gives the best floor detection probability, but not necessarily the smallest distance error.

Mobile Tracking in Mixed Line-of-Sight/Non-Line-of-Sight Conditions: Algorithm and Theoretical Lower Bound

The paper investigates the problem of mobile tracking in mixed line-of-sight (LOS)/non-line-of-sight (NLOS) conditions. The motion of mobile station is modeled by a dynamic white noise acceleration model, while the measurements are time of arrival (TOA). A first-order Markov model is employed to describe the dynamic transition of LOS/NLOS conditions. An improved Rao-Blackwellized particle filter (RBPF) is proposed, in which the LOS/NLOS sight conditions are estimated by particle filtering using the optimal trial distribution, and the mobile state is computed by applying approximated analytical methods. The theoretical error lower bound is further studied in the described problem. A new method is presented to compute the posterior Cramer-Rao lower bound (CRLB): the mobile state is first estimated by decentralized extended Kalman filter (EKF) method, then sigma point set and unscented transformation are applied to calculate Fisher information matrix (FIM). Simulation results show that the improved RBPF is more accurate than current methods, and its performance approaches to the theoretical bound.

Efficient Gaussian mixture filter for hybrid positioning

This paper presents a new way to apply Gaussian mixture filter (GMF) to hybrid positioning. The idea of this new GMF (efficient Gaussian mixture filter, EGMF) is to split the state space into pieces using parallel planes and approximate posterior in every piece as Gaussian. EGMF outperforms the traditional single-component positioning filters, for example the extended Kalman filter and the unscented Kalman filter, in nonlinear hybrid positioning. Furthermore, EGMF has some advantages with respect to other GMF variants, for example EGMF gives the same or better performance than the sigma point Gaussian mixture (SPGM) [1] with a smaller number of mixture components, i.e. smaller computational and memory requirements. If we consider only one time step, EGMF gives optimal results in the linear case, in the sense of mean and covariance, whereas other GMFs gives suboptimal results.

Benchmarking Nonlinear Filters

Algorithm developers need relevant and practical criteria to evaluate and compare the performance of different discrete-time filters or filter variants. This paper discusses some pit-falls in different approaches and proposes a combination of criteria on which to base comparisons. A comparison of eight filters for a class of hybrid personal positioning problems is presented as an example.