ION GNSS 2012
Session F2: Urban & Indoor Alternatives: Wireless

Title: Physical-Statistical Channel Model for Joint GNSS and Mobile Radio Based Positioning
Author(s): W. Wang, T. Jost, A. Lehner, and U-C. Fiebig, C. Gentner, German Aerospace Center, Germany
Date/Time: Wednesday, September 19, 2012, 1:50 p.m.
Room: Grand Ballroom Center (Renaissance)

Global navigation satellite systems (GNSSs) provide high position accuracy as long as pure line-of-sight (LoS) conditions between the satellites and the receiver exist. However, in critical environments like urban canyons, the position accuracy by GNSSs very much deteriorates due to shadowing, diffraction, and reflection of satellite signals. Augmenting GNSS based positioning by mobile radio signals of opportunity very much helps in these environments and improves the position accuracy compared to a GNSS-only solution.
To assess receiver performance for combined GNSS and mobile radio based positioning, coherent channel models considering both GNSS and ground mobile radio propagation effects are of high importance.

The DLR land-mobile satellite (LMS) channel model is such a coherent channel model which provides the channel impulse response as a function of time for a mobile receiver. This channel model is widely used for the simulation of position accuracy of mobile satellite navigation receivers. It consists of a combination of a deterministic and statistical modeling approach which enables a fast and accurate generation of the non-stationary channel impulse response (CIR). The model creates an artificial urban canyon scenario including house fronts, lamppost and trees. The direct path component is determined using physical deterministic methods and considers diffractions caused by a house front or a lamppost. Shadowing caused by trees is incorporated by a tree top model attenuating the signal´s amplitude proportional to the path length through the canopy additional to a stochastic fading process. The multipath components are generated by a geometric based stochastic channel model (GSCM) approach where each path is represented by a virtual reflector in space. Statistics for these multipath components are obtained from measurement data and encompass location, variable number of echoes, life span, and power fading characteristics. For a moving receiver the delay variation of each path is geometrically determined and the complex amplitude fading is generated by a stochastic process. During movement of the receiver, the reflectors are removed from the scenario based on their corresponding life span and new reflectors are generated according to the given statistics. The channel model has been standardized within the ITU-R P.681-7.

There are a number of accurate channel models for mobile radio propagation. However, they are not suitable to model all characteristics required to simulate the accuracy for mobile radio based positioning. As one of the most accurate channel models for mobile radio communications, the WINNER channel model lacks of two features which are important for the positioning application. Firstly, the WINNER channel model does not consider the absolute propagation delay between transmitter and receiver. Therefore, in non LoS (NLoS) situations, this model is unable to predict the delay bias of the first detectable propagation path with respect to the geometric LoS (GLoS) path, defined as the NLoS bias in this abstract. In contrast to most communication applications, the delay bias is essential for the evaluation of time-based range measurements. The second feature which is lacking in the current implementation of the WINNER model is the time evolution of the CIR which is insufficiently incorporated for positioning applications currently. The time evolution of the channel is an important feature to test tracking algorithms. Moreover, the coherent combination of the mobile radio channel model and the LMS channel model is missing so far.

In this paper we have two contributions: We proposed an extended WINNER channel model which fulfills above mentioned requirements and we proposed a coherent combination of the DLR LMS channel model with the extended WINNER channel model.

A GSCM approach incorporating equivalent reflectors is used to extend the current WINNER model. The statistics characterizing the equivalent reflectors are derived from the original WINNER channel model. Using the concept of equivalent reflectors allows the evolution of a realistic time-variant impulse response. A life time for the virtual reflectors has been incorporated. Furthermore, we introduce the application of a virtual environment, where the NLoS bias is generated in a physical deterministic way. A validation based on the power excess delay profile indicates that the proposed extended model is statistically consistent with the original WINNER model.

Thus, we have two channel models which allow for stand-alone simulations of the positioning accurate of mobile receivers: either for satellite based positioning or for mobile radio based positioning.

The remaining task is to coherently combine the DLR LMS and the extended WINNER channel models for simulations of joint GNSS and mobile radio based positioning. This is done by using the same artificial scenery for both models. In this scenery the satellite is placed according to its location in the sky and the base stations are placed on the roof top of buildings. The local environment around the mobile receiver is identical for both channel models. Therefore, it is possible to combine the LMS and the mobile radio channel by considering the propagation effects caused by the common local environment: we calculate the direct path in a physical deterministic way taking into account the same scenery, and obtain a coherent combination of both channels models.

Return to Session F2 Next Abstract