Overhead-efficient algorithms for acquisition, channel estimation and tracking in mimo-ofdm systems

Orthogonal Frequency Division Multiplexing (OFDM) has become a leading modulation technique for wideband wireless communications. The logical next step in the evolution of the technology is to boost its bandwidth efficiency by extending it to operate over Multiple-Input Multiple-Output (MIMO) wireless links. The result—MIMO-OFDM—promises a quantum leap in the achievable throughput over wireless media. The path toward commercially attractive packet-switched MIMO-OFDM systems, however, is not devoid of technical obstacles. For instance, extending conventional techniques of initial channel estimation to MIMO-OFDM is costly in terms of training overhead and spoils much of the throughput gain. Similarly, the increased MIMO receiver complexity results in large processing latencies in hardware implementations, which compromises the bandwidth and stability of conventional tracking algorithms. These kinds of problems motivate the development of entirely new acquisition and tracking techniques, and constitute the main fields of contribution of this dissertation. The work presented herein proposes a complete set of new algorithms for channel estimation, acquisition and tracking in MIMO-OFDM. The algorithms result from theoretical derivations of optimal parameter estimators in each case. The results provide valuable insight into how transmission overhead shall be allocated and used with maximum efficiency. For acquisition, the proposed method uses a carefully designed multi-transmitter preamble sequence that allows for joint channel estimation, frequency offset acquisition, and symbol timing estimation at the MIMO-OFDM receiver. The acquisition sequence length is controlled by two competing parameters, which allow for an explicit and fine trade-off between the preamble overhead and the quality of the resulting channel estimates. Compared to the acquisition sequence used in the IEEE 802.11a standard, the proposed technique requires about 8 times shorter training sequences in order to attain acquisition results of similar quality. The proposed methods for tracking carrier and sampling clock frequency offset are based on maximum likelihood estimation theory applied to observations of received pilot subcarriers at the output of the FFTs of the receiver. Simulation results show that larger MIMO configurations benefit from lower estimator variances, providing increased tracking accuracy at low signal-to-noise ratios, or allowing for a reduction of the number of pilot subcarriers compared to conventional OFDM systems.