Abstract: The Lattice Orthogonal Frequency Division Multiplexing (LOFDM) system can achieve excellent performance in combating time-varying and multi-path fading in fast-moving environment via the special time-frequency Lattice (TFL) structure and the property of lager Euclidean distance. The maximum Doppler spread, or equivalently, the mobile speed, is a measure of the spectral dispersion of mobile fading channel. Accurate estimation of the mobile speed is of importance in LOFDM systems which require the knowledge of the rate of channel variations to achieve its transmission signal designing and adaptive strategy. In this paper, aiming at the special structure of LOFDM signals and fast time-varying characteristics of doubly-dispersive channels, the discrete Fourier transform based basis expansion models (DFT-BEM) is used to approach the real fast time-varying channel, and the Doppler domain averaging channel frequency response (CFR) is estimated by exploiting the comb-type pilots and the block transmission strategy. On the base of the time auto-correlation function of the CFR estimated value, the F-norm based maximum Doppler spread estimation is realized. On the other hand, the subspace based maximum Doppler spread estimation algorithm is proposed, which reduces the effect of noise on the estimation and effectively improves the performance of the algorithm at low signal-to-noise ratio (SNR).