(ICC 2002), New York, NY, pp.267-271, April 28-May 2, 2002.
Significance:
The computer simulation of realistic multiple-input multiple-output (MIMO)
fading channels is essential for evaluating the performance of practical
space-time modem techniques in multiantenna
wireless systems. In this contribution, we prove that existing Rayleigh/Ricianmultichannel
fading simulators can only generate MIMO channels for which the temporal
and spatial correlation statistics are separable and have the same
functional form. A more comprehensive simulator is thus needed for the
emulation of general MIMO stochastic fading channels. In [1], a multichannel generalization of the our AR simulator in
[3] (see below) is developed to synthesize vector Rayleigh
and Rician fading processes that possess
specified realizablejoint space-time cross-correlation statistics.The capability of the technique is
demonstrated by the accurate simulation of general space-time-selective
MIMO channel models that have recently appeared in the literature. Note
that such models could not be simulated by previous methods. As proof of
the usefulness of the proposed technique, we note that independent
researchers (journal paper published in IEEE Transactions on Vehicular
Technology, vol. 53, pp. 634-643, May 2004) have cited our work and adopted our multichannel
simulator design to investigate the outage capacity of MIMO channels with
receiver motion and nonisotropic scattering at
both ends of the link.
[3]
Baddour, K. E. and Beaulieu, N. C., “Autoregressive Modeling for Fading
Channel
Simulation,” to appear as a full paper in IEEE Transactions on Wireless
Communications, 2005 (in press).
Early work appeared in:
[4]
Baddour, K. E. and Beaulieu, N. C., “Autoregressive Models for Fading
Channel Simulation,” IEEE Global Telecommunications Conference (GLOBECOM
2001),
San Antonio, TX, pp. 1187-1192, Nov. 25-29, 2001.
Significance:
The emulation of mobile radio channels is of much theoretical and
practical interest to the wireless community, as the design of modern
communication systems cannot be carried out without computer simulations.
In this work, a general easy-to-use autoregressive (AR) stochastic
modelling approach is developed for the accurate computer simulation of
time-selective wireless fading channels. Prior to this work, previous
researchers had reported that AR models could not be used for accurate
correlated fading simulation due to stability problems. In [3], it is shown
that the deterministic nature of the bandlimited
Doppler fading process is the cause of the numerical problems faced by AR
simulators. A simple technique is suggested to resolve the stability issues
while maintaining the statistical accuracy of the generated fading process.
Theoretical analyses are carried out to verifythe validity of the proposed
method, and start-up methods are presented to eliminate the need to discard
many initial generated outputs due to transient distortion. Performance
comparisons are then made with popular fading generation techniques to
demonstrate the merits of the approach. The proposed simulator has
comparable accuracy yet provides much smaller run times than competing
sum-of-sinusoids simulators. Furthermore, the proposed method has the
attractive property that fading variates can be
generated as they are needed, which is a feature that is not possible by
the memory intensive inverse fast Fourier transform (FFT) based simulators.
Unlike competing approaches, the fading autocorrelation function is also
easily specified, which makes the proposed AR simulator especially suited
for the accurate emulation of nonisotropic/directional
fading scenarios encountered in practical wireless communication systems.
Doppler
Spread Estimation:
[5]
Baddour, K. E. and Beaulieu, N. C., “Nonparametric Doppler Spread
Estimation
for
Narrowband Wireless Channels,” accepted for publication as a full paper in IEEE
Transactions on Vehicular Technology, 2005.
Sections have appeared in:
[6]
Baddour, K. E. and Beaulieu, N. C., “Nonparametric Doppler Spread
Estimation
of
Flat Fading Channels,” IEEE Wireless Communications & Networking
Conf.
(WCNC 2003), New Orleans, Louisiana, pp. 953-958, Mar.16-20, 2003.
Significance:
A robust and computationally efficient nonparametric technique is proposed
to estimate the Doppler spread of a narrowband wireless channel. This
information is of significant value as it reveals the channel’s rate of
change and can be used to improve the performance of many wireless
communication subsystems. Prior to [5, 6], Doppler spread estimators were
designed under the theoretical assumption that the channel exhibits Rayleigh fading with isotropic scattering. However, in
a practical propagation environment these estimators can suffer from large
biases since the scattering is often directional and may contain a
line-of-sight component. In [5, 6] the proposed estimator is not based on
the assumption of a specific Doppler spectral shape. Instead, a simple periodogram-based estimator is proposed which can be
implemented using efficient FFT processing. We demonstrate that the
resolution afforded from short observation windows permits good
nonparametric estimates of the Doppler spread in a broad range of channels.
We study the finite-sample accuracy of our estimation algorithm and compare
it to that of the established techniques. Unlike the competing approaches,
we find that the proposed estimator is robust over a broad range of Doppler
rates, channel spectrum shapes and signal-to-noise ratio conditions.