WPR.1 - Modeling, calibration, and validation of multi-dispersive, multi-link channels

Objectives

The main objective of this work package is to refine and extend existing standard models of the radio channel, like the 3GPP (Winner) Spatial Channel Model and to derive new models for new applications, like body-area networks and vehicular communications. The sought models account for all features in the radio channel that affect the behaviour and performance of advanced wireless communication systems or wireless networks operating in it. Under the term “model” we understand here a generic name encompassing any type of models, e.g. purely stochastic, deterministic, or hybrid deterministic-stochastic.

Different sub-goals have been identified, which address specific aspects of the sought models. Collaborative activities have been defined to reach these sub-goals. These activities are organized in tasks, corresponding each to one of the identified sub-goals.

• Task TR1.1 Design of multi-dispersive models. Multi-dispersive models for radio links will be developed, which reproduce dispersion in all dispersion dimensions, i.e. delay, direction of departure, direction of arrival, Doppler frequency, and polarization. Models describing the dynamic fluctuations of radio channels, like the vehicular channel, will be also designed.
• Task TR1.2 Design of multi-link models. Multi-channel or multi-link models will be designed, which account in a realistic manner for the statistical dependence between the responses of the channel links in a cooperative or relaying network. This aspect is particular important in body-area networks and in cooperative networks.
• Task TR1.3 Adaptive channel modelling for flexible radio. Based on the information sensed by the receiver (band, environment), an adaptive modelling procedure will be developed taking into account “at best” the information at hand in order for the flexible radio to adapt its power and transmission rate to the statistical model.
• Task TR1.4 Experimental and theoretical model calibration and validation. Channel measurement data will be either made available or collected for calibration and validation of the derived models. Due to the costs of collecting a sufficiently large amount of measurement data to achieve significant statistical results, ray tracing simulations will also be performed for the same purpose. Robust, efficient estimators of the channel parameters and metrics characterizing the proposed models will be derived and their performance will be assessed.

Description of work

Task TR1.1: Design of multi-dispersive channel models

The more dispersion dimensions are exploited in a wireless communication system, the more sensitive to the detailed characteristics of dispersion in these dimensions the performance of this system is. Advanced present and future communication systems will be wideband or ultra-wideband, are likely to be equipped with dual-polarized multiple-element transmit and receive antennas (dual-polarized MIMO -multiple-input, multiple-output-), and will be requested to operate in fast time-variant environments. To design and optimize these systems and assess their performance as well, accurate channel models are required that incorporate dispersion in delay, in direction of departure (at the Tx), in direction of arrival (at the Rx), in Doppler frequency, and in polarization. A further characteristic of relevance for scheduling is the dynamic (long-term) fluctuation of the time-varying channel, e.g. the vehicular channel.

The main objective of Task TR1.1 is to derive models of the radio channel accounting for the above mentioned-effects.

The following specific aspects will be investigated in this task:

• Identification of the kinds of multi-dispersive propagation/channel models required by today’s system design, deployment and optimization (empirical vs. physical; stochastic vs. deterministic).
• Modelling of the short-term and long-term temporal fluctuations
• Investigation of channel prediction methods and theoretical assessment of maximal prediction horizon
• Antenna characterization and optimization in multi-dispersive channels
• Definition of the appropriate metrics for characterizing the features of the radio channel critical to communication systems operating across this channel.

 Main players: BILKENT, CNIT, IST/TUL, CNRS, AAU, UCL, FTW

Task TR1.2: Design of multi-link channel models for cooperative and relaying networks

The development and realistic performance assessment of relay and cooperative networks require stochastic models that jointly characterize the different channel links in these networks. The traditionally postulated assumption of statistical independence of these links is likely to deviate significantly from reality and so the theoretical and simulation results obtained based on this assumption are.

The main objective of this task is to assess the statistical dependence, e.g. in terms of correlation, between the responses of the channel links in a wireless network, and to design realistic models reflecting this dependence. As a special case the statistical dependence between the up-link and down-link will be addressed as well.

The following specific aspects will be investigated in this task:

• Derivation of models describing the statistical dependence between the responses of channel links in a wireless network.
• Realistic description of macro-diversity
• Efficient, i.e. reduced complexity modelling of physical-layer processes for link-level simulations
• Characterization of body-area channels

 Main players: BILKENT, IST/TUL, CNRS, RWTH, AAU, UCL, FTW

Task TR1.3: Adaptive channel modelling for flexible radio

Next generation wireless systems will be based on adaptive and flexible coding and resource allocation schemes depending on the user's environment. In particular, provided the information of the user's location (dense, urban, indoor, outdoor, measured statistics), consistent channel models will be developed taking into account all the information at hand. The procedure based on the principle of maximum entropy (which has already been proved to be a successful methodology) will be extended to cases where the user has deterministic information as well as time-variable information (as until recently, only statistical knowledge has been done). Note that this approach can provide important gains in terms of capacity as the model (which is less time-variant then the channel realization) will be taken into account to optimize adaptively the code structure. In particular, the following specific aspect will be investigated in this task:

• Development of a mathematical framework to incorporate information on the environment in the model
• Study of the performance of the procedure compared to other simple classification model based methods.

 Main players: CNRS, FTW

Task TR1.4: Experimental and theoretical model calibration and validation

A prerequisite to the design of accurate channel models is calibration and validation by means of experimental data and/or realistic ray-tracing-based simulations. In the calibration process the critical parameters and metrics characterizing the models are estimated from possibly large sets of data obtained either from measurement or from Monte Carlo simulations using ray tracing. The obtained estimates are then used to determine the empirical probability distributions of these parameters and metrics. These distributions are utilized to generate realistic realizations of the channel response.

Experimental data will be either made available from past measurement campaigns or collected in campaigns to be carried out within the NoE NEWCOM++.

The investigations to be performed in this task are targeted to the multi-dispersive channel models to be designed in Task TR 1.1 and the multi-channel models to be developed in Task TR1.2. Two important issues in experimental channel calibration are (1) the robustness and accuracy of the parameter and metric estimates, and (2) the mitigation of the impairing effects due to the inherent imperfections of the channel sounders, like phase noise. Finding efficient solutions to both issues is another major goal of this task.

The following specific aspects will be investigated in this task:

• Development of experimental and simulation (ray-tracing-based) methods that enable to gather multi-link channel data.
• Gathering of experimental data from previous measurement campaigns or new campaigns
• Improvement of the performance of channel sounders (sounding mode optimization, mitigation of the effect of phase noise, adaptive sounding waveform optimization, antenna design).
• Derivation of robust, computationally efficient, and accurate estimators of the channel parameters and metrics.
• Theoretical performance assessment of these estimators.

 Main players: BILKENT, CNIT, CNRS, AAU, UCL, FTW