Effects of channel estimation errors on discrete multitone modulation (DMT) systems in a Rayleigh fading environment

This paper analyzes the degradation in signal-to-noise ratio, and thus probability of error, due to channel estimation errors for a discrete multitone modulation (DMT) system in a Rayleigh fading environment. A scalar channel model similar to the well known Jakes model with additive white Gaussian noise is assumed. Analytic expressions for the multitone SNR as a function of mean-square-error in channel estimation are derived. Various prediction techniques are analyzed, and a least mean squares (LMS) algorithm is proposed to track the channel; the weights obtained are then used to predict the next snapshot of the channel. Finally, the link between channel estimate update frequency and maximum Doppler shift is derived.