 Abstract | The use of multiple antennas at the transmitter as well as at the receiver can greatly
improve the capacity of a wireless link when operating in a rich scattering environment.
In such an arrangement all transmitting antennas radiate in the same frequency band
so the overall spectral efciency becomes very high. Such a multiple antenna scheme,
popularly known as Multiple Input Multiple Output (MIMO) has potential application
in wireless local area networks (WLAN) and cellular micro-cells. One reason is that the
WLANs and other short range wireless systems often operate in an indoor environment,
which offers rich scattering. The other reason is the demand for higher data rates in
cellular and WLAN systems to cater for multimedia services. Recently researchers have
proposed di erent architectures for materializing the potential of the MIMO scheme.
VBLAST (Vertical-Bell Labs Space Time) is a popular architecture that will play an
important role in future standardizations. Furthermore, different decoding methods
have been proposed for VBLAST. The SVD (Singular Value Decomposition) based
system is envisioned as a highly effective MIMO technique in a TDD (Time Division
Duplex) framework. Such a system operates by adapting the constellation size across
different subchannels.
In this work we study the VBLAST and SVD architectures and compare the perfor-
mance and computing power requirement of these architectures. Also in this study a
new effcient decoding method for the VBLAST architecture is proposed. The original
VBLAST decoding method relies on the repetitive computation of the pseudoinverse
of the channel matrix. Alternatively, there are methods based on the QR decomposi-
tion, the matrix square root etc. Our new decoding method is based on a relatively
less known matrix decomposition, the Polar Decomposition. The new method requires
less computation and has several other advantages like the possibility of incremental
updates, channel rank tracking, etc. We consider three different types of channels:
IID random, slow fading and measured channels. The entire work is simulated in the
MATLAB environment.
The main contribution of this work includes: a comparative study and a head to
head comparison of the VBLAST and SVD-based MIMO systems. The application of
adaptive modulation in the SVD-based system and the introduction of a new effcient
decoding method for the VBLAST system are also included. Simulation results are
reported with comments and conclusions. |
