Robust Precoding For MIMO OFDM With Insufficient CP
MIMO channels arising from the use of multiple antennas at both transmitter and receiver have received continuously attention because they provide an signicant increase in capacity over the single input single output (SISO) counterpart. Alternatively, techniques can also be applied to MIMO for enhancement of the link robustness.
However, MIMO channels are not only restricted to the multiple antenna scenario. The block transmission schemes over time dispersive or frequency selective channel, for example OFDM or single-carrier frequency domain equalization (SC-FDE), can be considered as signal models using MIMO channels. The equivalent channel in MIMO OFDM systems can be viewed as a generalized MIMO channel which consists of both spatial domain and frequency domain channel uses.
Each scalar data symbol in the vector input to the MIMO channel can be treated as a substream to a MIMO system. Assuming the signal constellations and coding schemes have been specied for each substream, it is viable to optimize the link quality of all the substreams by designing precoder at the transmitter and equalizer at the receiver. The joint optimization of precoder and equalizer is often intractable. One can nd suboptimal solutions by xing the receiver processing to be minimum mean square error (MSE) equalizer and design the precoder for specic requirement.
has shown that MIMO channel diagonalizing structure is optimal for a variety of minimization problems regarding the MSE of all the substreams, or the signal to noise plus interference ratio (SINR) or bit error rate (BER). The channel-diagonalizing techniques require CSI at the transmitter, which imposes stringent requirement on the channel estimation and CSI feedback. To ease this burden, statistically robust designs can be employed since the statistics of the channel is more likely to be long term static. When the mean and covariance CSI at the transmitter is available, robust designs are more involved. For example, designs a robust transceiver based a general cost function of the average MSEs, which takes advantage of the mean and covariance information at the transmitter.
