CHARACTERIZATION AND ESTIMATION OF HIGH FREQUENCY CHANNEL WITH MULTIPLE ANTENNA SYSTEM

摘要

This thesis presents the development of an efficient and adaptable simulation model forcharacterization and estimation of MIMO-HF channel subjected to impairments, such asmultipath fading, system non-linearity and non-Gaussian noise. Under channelcharacterization and modelling, this thesis proposes the extension of conventionalWatterson model for SISO to MIMO-HF channel incorporating the associated spatialcorrelation and the non-linearity of the system. The novelty of the developed model liesin the capability of its impulse response to emulate nearly the practical HF channel byincorporating the cited adverse channel impairments. This thesis proposes the modellingof MIMO-HF channel through the computationally efficient IIR /AR filter approachinstead of conventional FIR filter. The generic and the versatile features of IIR filterbasedapproach for modelling the HF channel impairments have been demonstrated by itsapplication to both SISO and MIMO-HF systems.Within the purview of channel estimation for improved reliability and enhanced data rateof the MIMO-HF communication link, this thesis proposes a novel PF based channelestimation technique for HF communication links subjected to multipath fading andsystem non-linearity. The PF based channel estimation algorithm proposed in thesis forMIMO-HF is shown to closely approximate the impulse response of the channel inducedby the channel impairments. The improved channel estimation facilitates the effectiveutilization of system resources to ensure enhanced capacity and reliability of HF links.Although one can conceive an idea of invoking the PF concept devoid of EKF, this thesisattempts to adopt a unified approach wherein PF and EKF schemes are combined torealize better posterior density functions, thereby improving the accuracy in channelestimation. The advantageous and desirable features derived by invoking the proposed PFformulation over the conventional RLS have also been addressed. This thesis alsoaddresses the effects of spatial correlation, system non-linearity and the non-Gaussiannoise on the proposed PF based channel estimation algorithm. The expected improvementin the receiver performance in lieu of improved channel estimation algorithms, as well asreplacement of conventional SISO with MIMO, is validated through the performanceparameters such as capacity and reliability.