Channel sounding and signal design for wireless MIMO channels.

摘要

In this thesis we focus on the problem of Channel estimation for MIMO systems, using OFDM sounding (or Pilot) signals, a critical component in many modern wireless communication systems. There are two basic components to this problem that we consider: Estimation of the channel parameters based upon observing the received sounding signal after passage through the channel; and optimum design of the sounding signal itself. Applications include channel estimation for demodulating signals in data transmission, as well as for purposes of characterizing the outdoor channel in various locales. Channel estimation is by no means new, and the novel aspect of this thesis is in the use of prior statistical channel information to aid in minimum mean square estimation (MMSE) of channel parameters for the MIMO channel. This kind of MMSE channel estimation has been previously used for the Single Input Single Output (SISO) case. In this thesis we extend this basic idea to the MIMO case, which presents some new issues to be dealt with.;In this thesis, we derive optimal sounding signal characteristics for MMSE channel estimation assuming that second order channel statistics are known, for the SISO and MIMO cases. We develop the mathematical formulation of the Minimum Mean Square Estimation (MMSE) for the SISO case and compare the MMSE channel estimator to the standard correlator estimator under different Doppler bandwidth scenarios. We also extend consideration to the MIMO case and present two ideas for MMSE channel estimation in this case. First, we investigate the possibility of using orthogonal signals on the different transmit antennas by using different sets of OFDM subcarriers, in order to simultaneously estimate all the constituent channels in the MIMO system. Next, we consider the idea of sequentially transmitting on the different transmit antennas and using MMSE interpolation to estimate all the constituent channels at the same time instant.;We also study optimal sounding signal design in the (SISO) case, under the restricted condition fdT = 0 and find that the optimum sounding signal should be flat over frequency, or impulsive in time. We also extend the study to optimal MIMO sounding signal design with fdT = 0 and find the required conditions on the OFDM symbol amplitudes. Finally we present some simulation results for the MMSE channel estimator when it uses estimated channel statistics, using the 3GPP Spatial Channel Model (SCM) and find that a limited number of measurements can greatly improve channel estimation relative to the simple correlator.