Modeling of Reverberation Effects for Radio Localization and Communications

Publikation: ForskningPh.d.-afhandling

Abstrakt

For decades the terrestrial radio channel has been characterized and modeled for communication purpose only, e.g. to design wireless systems and/or to assess their performance by means of Monte Carlo simulations. The recent emergence of localization capabilities in terrestrial wireless systems demand for novel channel models that, in addition, accurately emulate the location-dependent features of real channels.

In this thesis we address and provide answers to the central questions of the cause, the effect and the modeling of the diffuse component observed in delay power spectra measured in indoor environments. We show that this component carries a significant portion of the total received power. Thus, the accurate modeling of it is of prime importance for both communication and positioning.

To clarify the cause of the diffuse component we first experimentally investigate the spread in delay and direction of multipath components in an indoor environment. The results indicate small per-path-component spreads. This finding suggests that the diffuse component is contributed by a multitude of (weak) specular multipath components. Others have also experimentally confirmed this conclusion using an ultra-wideband (UWB) set-up: measured UWB delay power spectra exhibit a transition from early specular components to a diffuse component. We name this transition the ``avalanche effect''. We propose a propagation graph that mimics propagation conditions in an environment: transmitters, receivers, and scatterers are represented as vertices and propagation conditions between these vertices as (labeled) edges. Due to its recursive structure, the graph model is capable of reproducing the avalanche effect and the diffuse component, even though propagation along the edges is assumed specular.

Experimental evidence also shows that the diffuse component of delay power spectra measured in the same room decays exponentially with nearly the same decay rate and magnitude regardless of the transmitter-receiver distance. Only its onset varies with distance. We incorporate these empirical features into a distance dependent model of the delay power spectrum, which we then validate experimentally. From this model we derive secondary models that predict the received power, the mean delay, the rms delay spread and the kurtosis versus distance.

The behavior of the diffuse component versus distance in indoor environment is linked to reverberation effects analog to reverberation effects observed in room acoustics and electromagnetic reverberation chambers. Reverberation models of room acoustics relate the decay rate of the diffuse component to the room geometry and an average absorption coefficient. Following a recently proposed approach, we transcribe these models to electromagnetics and validate them experimentally following a systematic procedure. These transcribed models provide accurate predictions of the delay power spectrum in a typical office environment. Furthermore, they can predict changes in the diffuse power due to opening windows, the presence of people or by changing the size of the room.

As an example of the benefits of modeling the diffuse component in indoor environments we present a study of the performance of a positioning estimator that jointly utilizes the secondary models of received power and mean delay versus distance. The results indicate that the mean delay carries relevant location-dependent information that can be exploited to enhance the localization accuracy.

Luk

Detaljer

For decades the terrestrial radio channel has been characterized and modeled for communication purpose only, e.g. to design wireless systems and/or to assess their performance by means of Monte Carlo simulations. The recent emergence of localization capabilities in terrestrial wireless systems demand for novel channel models that, in addition, accurately emulate the location-dependent features of real channels.

In this thesis we address and provide answers to the central questions of the cause, the effect and the modeling of the diffuse component observed in delay power spectra measured in indoor environments. We show that this component carries a significant portion of the total received power. Thus, the accurate modeling of it is of prime importance for both communication and positioning.

To clarify the cause of the diffuse component we first experimentally investigate the spread in delay and direction of multipath components in an indoor environment. The results indicate small per-path-component spreads. This finding suggests that the diffuse component is contributed by a multitude of (weak) specular multipath components. Others have also experimentally confirmed this conclusion using an ultra-wideband (UWB) set-up: measured UWB delay power spectra exhibit a transition from early specular components to a diffuse component. We name this transition the ``avalanche effect''. We propose a propagation graph that mimics propagation conditions in an environment: transmitters, receivers, and scatterers are represented as vertices and propagation conditions between these vertices as (labeled) edges. Due to its recursive structure, the graph model is capable of reproducing the avalanche effect and the diffuse component, even though propagation along the edges is assumed specular.

Experimental evidence also shows that the diffuse component of delay power spectra measured in the same room decays exponentially with nearly the same decay rate and magnitude regardless of the transmitter-receiver distance. Only its onset varies with distance. We incorporate these empirical features into a distance dependent model of the delay power spectrum, which we then validate experimentally. From this model we derive secondary models that predict the received power, the mean delay, the rms delay spread and the kurtosis versus distance.

The behavior of the diffuse component versus distance in indoor environment is linked to reverberation effects analog to reverberation effects observed in room acoustics and electromagnetic reverberation chambers. Reverberation models of room acoustics relate the decay rate of the diffuse component to the room geometry and an average absorption coefficient. Following a recently proposed approach, we transcribe these models to electromagnetics and validate them experimentally following a systematic procedure. These transcribed models provide accurate predictions of the delay power spectrum in a typical office environment. Furthermore, they can predict changes in the diffuse power due to opening windows, the presence of people or by changing the size of the room.

As an example of the benefits of modeling the diffuse component in indoor environments we present a study of the performance of a positioning estimator that jointly utilizes the secondary models of received power and mean delay versus distance. The results indicate that the mean delay carries relevant location-dependent information that can be exploited to enhance the localization accuracy.

OriginalsprogEngelsk
Antal sider240
ISBN (elektronisk)978-87-7152-024-8
StatusUdgivet - 2013

Download-statistik

Ingen data tilgængelig
ID: 208328824