Under high mobility, the orthogonality between sub-carriers in an OFDM symbol is destroyed resulting in severe inter-carrier interference (ICI). We present a novel algorithm to estimate the channel and ICI coefficients by exploiting the channel's time and frequency correlations and the (approximately) banded structure of the frequency-domain channel matrix. In addition, we invoke the asymptotic equivalence of Toeplitz and circulant matrices to reduce the dimensionality of the channel estimation problem by retaining the dominant terms only in an offline eigen-decomposition. Furthermore, we show that the asymptotically MMSE-optimum pilot design consists of identical equally-spaced frequency-domain clusters whose size is determined by the channel Doppler spread. Comparisons of our proposed algorithm with a widely-cited recent algorithm demonstrate a significant performance advantage at a comparable real-time complexity.

We consider comb-type OFDM transmission over doubly-selective channels. Given a fixed number and total power of the pilot subcarriers, we show that the MMSE-optimum pilot design consists of identical equally-spaced clusters where each cluster is zero-correlation-zone sequence.

We propose two novel algorithms based on compressed-sensing theory to estimate and cancel narrow band interference (NBI) in orthogonal frequency division multiplexing (OFDM) systems. Simulation results demonstrate the effectiveness of our proposed algorithms in estimating the NBI frequency support and approaching the performance with no NBI.

We design a novel class of space-time codes, called linear diversity-embedding space-time block codes (LDE-STBC) where a high-rate STBC is linearly superimposed on a highdiversity STBC without requiring channel knowledge at the transmitter. In applying this scheme to multimedia wireless communications, each traffic type constitutes a transmission layer that operates at a suitable rate-diversity tradeoff point according to its quality-of-service requirements. This, in turn, provides an unequal-error-protection (UEP) capability to the different information traffic types and allows a form of wireless communications where the high-rate STBC opportunistically takes advantage of good channel realizations while the embedded high-diversity STBC ensures that at least part of the information is decoded reliably. We investigate transceiver design issues specific to LDE-STBC including reduced-complexity coherent decoding and effective schemes to vary the coding gain to further enhance UEP capabilities of the code. Furthermore, we investigate the application of LDE-STBC to wireless multicasting and demonstrate its performance advantage over conventional equal-error-protection STBC.

The reliability of mobile packet data transmission in multi-input multi-output (MIMO) communication systems is degraded by time-varying multi-path channel fading. This so-called Doppler-effect can be mitigated by using a shorter packet length and/or inserting more pilot training symbols and frequently tracking the channel at the receiver at the cost of additional rate overhead. A judicious choice of the packet length according to channel conditions aims at achieving a tradeoff between reliability and overhead to maximize the achievable throughput. We present an analytical framework for optimizing the packet length in mobile MIMO systems to maximize throughput taking into account channel estimation and Doppler effects. As a case study, we consider the Alamouti space-time block code (STBC) for 2 transmit antennas over Rayleigh flat-fading channels and determine the optimum packet length as a function of input SNR and Doppler rate. Furthermore, we investigate the effect of various system design parameters on the optimum packet length and achievable throughput such as signal constellation size, number of receive antennas, and receiver detection algorithm.