This paper analyzes the effect of fast fading Doppler channel on user scheduling algorithms. Considering a millimeter wave (mmWave) downlink massive multiple-input multiple-output (M-MIMO) system, we adopt a two-stage precoding method to cancel the interference from users. Improved K-mean clustering method with weighted likelihood (WLD) similarity measure is employed to group the active users within a given region. After grouping the users, different scheduling methods, semi-orthogonal user selection (SUS) algorithm and greedy algorithm, are implemented. Initially, grouping and scheduling are performed based on the static channels. Then, the scheduled users are simultaneously served for the next time frames without rescheduling while the users are moving. With this approach, this paper reveals how static and time-varying user grouping and scheduling methods perform in terms of system sum-rate when the fast moving users experience the Doppler effect. Different scenarios of number of active users and time frames are simulated. It is observed that rescheduling is required when greedy algorithm is used in order to maintain high sum-rate, while SUS algorithm does not achieve high sum-rate compared to greedy algorithm.