For the simplification for the original trajectory sequence of the moving object collected from the mobile devices, this paper defined a kind of backtracking based simplification framework, which used the linear prediction and length of simplification queue to dominate the time of simplification, and simplified the original trajectory sequence between the present moment and starting time of a retrospective historical trajectory adopting the method of temporal distance as the error metric. In the backtracking based simplification framework, this paper first utilized the new reduced points to construct several vectors and predicted the velocity, which could narrow the gap between the prediction and actual velocity in the future. This paper then utilized the point sets to store the edges in the directed acyclic graph needed in the access to reduce time complexity of the algorithm. The first experiment shows that the reduction rate using the optimal velocity prediction is greater than that of the original velocity prediction with the high fluctuant trajectory data. It suggests that the predicted velocity is closer to the actual velocity in the future moving direction than that in the tangent direction. The second experiment shows that the time performances of the optimized simplification algorithm are improved. This study shows that the reduction of the visits of the edges can decrease the time overhead of the algorithm.