[attachment=11322]
Abstract
Recent advances in information theory show thatemploying multiple antennas at both sides of a wireless channelpromises enormous capacity potential. With perfect knowledgeof channel state information (CSI) at the transmitter, eigenbeamformingis the optimal coding scheme to exploit this potential.However, in non-stationary wireless environments, highcomplexity on multiple-input-multiple-output (MIMO) channeltracking and large amounts of CSI feedback render such anapproach impractical. In this letter, by exploiting the wirelessmultipath channel structure characterized by the path delaysand the path directions-of-departure/arrival, a new space-timetransmit scheme which employs a structure-based water-fillingalgorithm is proposed
I. INTRODUCTION
Employing multiple antennas at both sides of a wirelesschannel is considered an effective way to improve the performanceof a wireless communication system. Informationtheory of this perspective studied in [1] provides measuresof significant increase in system capacity. Several space-timearchitectures [2], [3] were proposed to exploit this potential.With channel state information (CSI) at the transmitter, thespace-time eigen-beamforming approach adopted in [3] isclaimed to be the optimum coding scheme in terms of systemcapacity. However, eigenbeams are usually fast time-varying.High complexity on multiple-input-multiple-output (MIMO)channel tracking and large amounts of CSI feedback rendersuch an approach impractical.Alternatively, coding schemes looking into space-time channelstructure should be more practical and robust. In [4],[5], spatially correlated fading channels are considered ininvestigating the effects of the spatial channel structure onthe capacity of a wireless MIMO communication system. Onthe other hand, the frequency-selective channels in most ofthe research works on space-time coding so far are modeledas independently-fading time taps with no channel structure.Current and upcoming indoor wireless applications usuallyadopt higher carrier frequencies, e.g., 5GHz for wirelessLAN 802.11a and even up to 60GHz in the future wirelessstandards, for obtaining more channel bandwidth and higherdata throughputs. In such indoor wireless environments, measurementresults [6], [7] clearly demonstrate strong multipathchannel structures, which are highly dependent on signal propagationenvironments. Illustrated in Fig. 1 are the descriptionsof the channel structures in two different environments: from1Category: 7.7 Array processing for wireless communicationsone room to another room, and from the corridor to a room[7]. For instance, in Fig. 1(a), the power polar plot at thereceiver in Lab