用于移动路径预测的多步Markov模型

针对k阶Markov预测器在状态空间上的高复杂性,提出了用二步Markov预测器进行WLAN环境下的移动路径预测。通过分别对二步和k阶Markov预测器条件熵的计算,在理论上了证明了二步Markov预测器的可行性。同时在1200个WLAN用户的移动跟踪数据集上对两者的预测精度的比较分析结果表明:二步Markov预测器在不同长度的移动跟踪数据集上比一阶Markov预测器有更好的稳定性,在大幅度地降低了k阶Markov预测器状态空间复杂性的同时,与二阶Markov预测器预测精度大致相当,并且比二阶以上Markov预测器的预测精度更高,具有比较大的实用价值。

Markov model used for path prediction over WLAN

In order to reduce the complexity of the state space in k-order Markov predictor,a 2-step Markov predictor was proposed to make path prediction over WLAN.Its feasibility was proved by calculation conditional entropy of 2-step and k-order Markov predictor.In addition,the analysis and comparison of the predictor accuracy for two kinds of Markov predictor using a two-year trace of the mobility pattern of 1200 users on Dartmouth's campus-wide Wi-Fi wireless network were conducted.The results show that the proposed Markov predictor is more stable than one-order Markov predictor in different length trace files and can also dramatically reduce Markov state space complexity with almost same prediction accuracy of two-order Markov predictor and higher accuracy than one-order Markov predictor.