Multiple antennas employed both at the transmitter and receiver have a
huge potential for future wireless systems (Foschini and Gans, 1998) which has the
ability to increase their capacity and reliability. Orthogonal Frequency Division
Multiplexing (OFDM) is well-known for efficient high speed transmission and robustness
to frequency selective channels (Anibal, 2000). Hence, the integration of these
two technologies has the potential to meet the ever growing demands of
future communication systems (Stuber et al., 2004). To protect the transmitted data
against random channel impairments, it is desirable to employ Link Adaptation (LA). The
basic idea behind employing LA techniques is to operate a link as efficiently as
possible (Jungnickel et al., 2003). In single antenna systems, LA is usually based on the
received Signal-to-Noise Ratio (SNR). But in case of multiple antenna systems, it is not
so straightforward. Even if the SNR is high, individual sub channels might
interfere significantly with each other, making it difficult even for the optimal receiver
to separate them and decode the packet successfully.
Hence, more advanced channel quality indicators (instantaneous SNR, Shannon capacity and Log
Likelihood Ratio (LLR)) (Zhang et al.,
2004) must be considered such that it takes care of the multidimensional
channel into account.
This is the prime motivation of the work for both single user and multiuser
MIMO-OFDM (Hojin et al., 2006).
In order to use the system efficiently with LA, channel information is required
at the transmitter for adaptive Modulation and Coding Scheme (MCS). Therefore,
an accurate channel estimation plays a key role in data detection, especially in
MIMO-OFDM system where the number of channel coefficients is M × N (M and N are the number of transmitted and received antennas respectively) times more
than Single-Input Single-Output (SISO) system.
Technically, there are two types of channel estimation
approaches (Reza et al., 2007): training-based and non-training based channel
estimations. However, training-based channel estimation has a lower computational
complexity since the statistical properties of the receiving data is not required. In this method, training symbols or pilot
tones which are known to the receiver are multiplexed along with the data stream
and transmitted using a wireless link. However, one drawback of
the training-based channel estimation is the huge overhead of the transmitted block. This problem can be
solved using adaptive algorithm. In adaptive algorithm,
the channel is estimated using pilot data at the start of transmission, then
it can be tracked using the data recovered from previous blocks. |
