A technique to improve MMSE performance of PACE in time-selective Rayleigh fading channels

The so-called peaky signaling has been shown to improve the performance of a communication system over noncoherent fading channels in the low signal-to-noise ratio (SNR) regime. We consider a signaling scheme which is equivalent to peaky signaling over a time-selective Rayleigh fading channel and make use of the unused degrees of freedom to improve the minimum mean square error (MMSE) performance of pilot-aided channel estimation (PACE) by a processing technique incorporating low pass filtering and downsampling. We assume infinite pilot symbols and optimal Wiener smoother at the receiver and give some analytical as well numerical results suggesting significant gains in MMSE performance of PACE.