Pareto Optimal Prediction for Moving Objects

This paper addresses the optimal prediction for moving objects according to the Pareto optimal model. There are many factors need to be taken into account when predicting moving objects, e.g. current position, history trajectory, weather, and so on, and these factors are often independence. Existing techniques for moving objects prediction always use a synthesized factor, which is calculated by a score function contain many factors, as the single metric to select optimal results. Unfortunately, in many applications, these factors often can't be integrated as one score-function, and users often need select the most possible prediction result by themselves. This paper firstly defines a prediction model using the Pareto optimal model, which selects optimal results by multiple independence score-functions, and then presents the Pareto Optimal Prediction algorithm (POP algorithm) for moving objects, which selects Pareto optimal prediction results for moving objects. By this algorithm, users can select the most possible prediction result by themselves.