摘要
分析了移动轨迹预测的已有方案及各方案存在的问题,提出了一种全新的移动设备位置预测方法,即基于模式挖掘与模式匹配的移动用户移动轨迹预测(Mpp)方法。在若干个实际WLAN用户的移动跟踪数据集上对Markov预测器和新预测器的预测精度进行了比较。实验结果表明:该方法能够达到比较理想的预测效果,与二阶Markov预测器的预测效果基本持平。同时,该方法能够实现增量挖掘,预测精度和可靠性有了进一步提高,具有较高的实用价值。
The shortcomings of some existed mobile path prediction schemes are analyzed in this paper. Then a new prediction method based on pattern mining and matching is proposed, which we call MPP method. In this study we compare the performance of the MPP with that of Order-k Markov predictors using a trace of the mobility patterns of 1,200 users on real Wi-Fi wireless network. Results show that using the MPP method we can achieve ideal predicting effect, which is equally good as or a little better than that using Order-2 Markov predictor, which has the best performance among Markov Order-k predictors. Also based on MPP method we can realize incremental mining that further improves the predicting accuracy and the reliability.
出处
《吉林大学学报（工学版）》
EI
CAS
CSCD
北大核心
2008年第5期1125-1130,共6页
Journal of Jilin University:Engineering and Technology Edition
基金
国家自然科学基金项目(60573128)
关键词
计算机系统结构
移动轨迹预测
模式挖掘
增量挖掘
模式匹配
computer systems organization
mobile path prediction
pattern mining
the Increment of mining
pattern matching
