To estimate speed of a moving user transmitting data over a Single Input Single Output (SISO) Rayleigh fading channels, many Autocorrelation function (ACF) based schemes have been presented in literature [2-7]. Recently, it was shown that in a Multiple Input Multiple Output (MIMO) channel with N antennas on each side, the velocity of a moving mobile can be obtained by applying the ACF based velocity estimation schemes to each MIMO sub-channel and averaging the resulting N 2 velocity estimates [1]. To successfully deliver bandwidth-intensive services to consumers, upcoming 4G consumer devices and base stations will be equipped with a large number of antennas. As the number of antennas (and thus the number of sub-channels) in MIMO systems increase, this scheme will become too complex to implement.In this paper, for correlated MIMO channels with Uniform Linear Arrays on both the transmitted and receiver, we propose a subset of four intelligently selected MIMO sub-channels, over which, if velocity estimates are obtained using ACF based schemes, then the average of these estimates will give an accurate MIMO velocity estimate. The estimates obtained by the proposed scheme are more accurate while at the same time, the computational complexity is reduced from N 2 reported in [1] to a constant, irrespective of N. The usage of this scheme leads to improved system performance and increased battery life in consumer devices.