| This dissertation includes two parts: robust receiver design for wireless radio frequency (RF) communication, and robust receiver design for underwater acoustic (UWA) communication. In the first part, four aspects including fading channel modeling, Doppler spread estimation, channel estimation, and (turbo) equalization, are investigated for a common digital receiver. The general methodology of study is theoretical analysis and numerical simulation. In the second part, different receiver designs are proposed for underwater acoustic communication, with their robustness tested by experimental data collected at several undersea field trials.;An equivalent discrete-time channel model is the basis for robust receiver design in any digital communication system. The work on channel modeling highlights the loss of separability property of the discrete-time channel scattering function, a result that is commonly overlooked. The conditions for the validity of the separability property is provided, under which efficient channel simulators can be used.;Doppler spread contains information about fading channels, thus is an important system parameter guiding robust receive design. A Doppler spread estimation algorithm is proposed, for the prevailing orthogonal frequency division multiplexing (OFDM) systems undergoing Rayleigh or Rician fading. The estimator belongs to the class of "method of moment", and it estimates the Doppler spread with the auto-covariance function of the received signal power.;Channel estimation is critical for equalization and detection. A pilot-aided channel estimation algorithm is presented for OFDM systems with non-ideal effects including both carrier frequency offset (CFO) and phase noise. It estimates and compensates the CFO first, and then performs joint channel estimation and phase noise suppression with the maximum a posteriori (MAP) criterion. The CFO compensation and phase noise suppression improves the channel estimation accuracy.;The work on equalization focuses on turbo (iterative) equalization, and discusses two turbo equalization schemes for multiple-input, multiple-output (MIMO) systems. The first scheme enhances existing linear minimum mean square error (LMMSE) turbo equalizer, by utilizing not only the a priori information at the input of the equalizer but also the a posteriori information obtained when the equalization progresses. With the new equalization mechanism, the detection ordering matters as to the equalization performance, and has been exploited to further enhance the detection performance. The second scheme develops a non-linear block decision-feedback equalizer (BDFE) with reliability-based successive soft interference cancelation (SSIC). The symbol reliability information is calculated directly with the a priori input, incurring minimum extra cost.;UWA communication is much more challenging than RF communication due to its limited available bandwidth, long delay spread, fast temporal variation and significant Doppler effect. Signal detection under such scenarios, is thus very difficult. The remaining part of the dissertation discusses robust receiver design for UWA communication. Two design schemes have been proposed. The first scheme adopts MIMO linear equalizer (LE) working with a group-wise phase compensator. It was tested with experimental data collected at Kauai, Hawaii, in September 2005, and Saint Margarets Bay, Nova Scotia, Canada, in May 2006. Experimental results show that it was robust under different communication environments. The second scheme adopts the developed MIMO turbo BDFE mentioned above, with its robustness and performance tested by experimental data collected at the coast of Martha's Vineyard, Edgartown, MA, in October 2008, and Gulf of Mexico in July 2008, respectively. |
