No abstract

Abstract-This paper considers the problem of mitigating fading and interference in wireless orthogonal frequency division multiplexing (OFDM) multiple access communication systems. Applications include cellular mobile radio, wireless local loop, and wireless local area networks. The effect of interchannel interference (ICI) arising from time-selective fading and frequency offsets and co-channel interference (CCI) is analyzed. A loop-timing method that enables a synchronous uplink between multiple mobile transceivers and a base-station is described. Adaptive antenna arrays are utilized at the base for uplink reception, and optimum array combining based on the maximum SINR criterion is used for each subchannel over slowly timevarying channels. For operation over fast time-varying channels, a novel two-stage adaptive array architecture that incorporates combined spatial diversity and constraint-based beamforming is presented. While ICI alone is most effectively overcome by spatial diversity, combined beamforming and diversity are most effective to combat CCI in the presence of fading. The overall method is suitable for real-time implementation and can be used in conjunction with traditional coding schemes to increase the link-margin.Index Terms-Adaptive arrays, interference suppression, OFDM.

Speech compredon is widely used in tel-dons, espedally in cellular telephony and voice over packet networks. Traditional network speech enhancement techniques, which operate on mded speech, q u i r e that the compresed speechbe demded. After perForming echo caudation and noise dudion, the speed^ is recylooded for M e r transport Due to the need for reencoding the speech, Such proceaing is coqutalionally iuhlsivq and gpeciany prone to additional quantization wise FurthermDre, additional delay is incurred. In this paper, a dim3 compresed domain signal p q approarh to speech enhancemnt is intt.oduced. Me4hods for mplmenting acoustic echo control and noise d u d i o n are discussed. The compmsed domain approach is athadive due to lpduced computational oasts, delay and quantization noise.

No abstract

No abstract