The impact of varying the values of w for mining approximate moving patterns is next investi-gated. Without loss of generality, we set the value of ε to be 12, that of mf to be 3, and the values of cf to be 1, 3 and 5. Both vertical_min_sup and match_min_sup are set to 20% , the value of δ is set to be 3 , and σ2 is set to be 0.25. With this setting, the experimental results are shown in Figure 4.3.
As can be seen from Figure 4.3, the precise ratio of AUMP increases as the value of w
60%
Figure 4.3: The performance of AUMP with the value of w varied.
0.83
Figure 4.4: The precise ratio of AUMP with vertical_min_sup and match_min_sup varied.
sequences considered in AUMP increases, AUMP is able to effectively extract more information from the log of call detail records. Note that with a given the value of w, the precise ratio of AUMP with a larger value of cf is bigger, showing that the log of data has more information when the value of cf increases. Clearly, for mobile users having high call frequencies, the value of w is able to set smaller in order to quickly obtain moving patterns. However, for mobile users having low call frequencies, the value of w should be set larger so as to increase the accuracy of moving patterns mined by AU M P .
0.7 0.75 0.8 0.85 0.9 0.95 1
0.1 0.25 0.75 1 1.5 2
variance threshold
precise ratio
A U M P w ith m atch_m in_sup=10%
A U M P w ith m atch_m in_sup=20%
A U M P w ith m atch_m in_sup=30%
Figure 4.5: The precise ratio of AUMP with the values of match_min_sup and the variance threshold varied.
Now, the experiments of varying the values of vertical_min_sup and match_min_sup for algorithm LS are conducted where we set the value of cf to be 5, that of mf to be 1, that of ε to be 12 and that of σ2 to be 0.25. The precise ratio of AU M P with various values of vertical_min_sup and match_min_sup are shown in Figure 4.4, where it can be seen that the precise ratio of AU M P with a given vertical_min_sup tends to increase as the value of match_min_sup increases. The reason is that increasing the match_min_sup is able to efficiently filter out call detail records that are viewed as noise data. As such, the precision of AU M P with higher match_min_sup is larger. In addition, with a given match_min_sup, the precise ratio of AU M P increases, as the value of vertical_min_sup increases. This is due to that as the value of vertical_min_sup increases, the more frequent set of base stations is determined from a set of call detail records. Thus, the frequent set of base stations, referring to areas that users more frequently travel, are very helpful to approximate user moving patterns.
As described before, the value of σ2 for algorithm TC affects the precision of time clustering in AU M P . To conduct the experiments to evaluate AU M P with the values of σ2 varied, we set
0.7 0.75 0.8 0.85 0.9 0.95 1
0.1 0.25 0.75 1 1.5 2
variance threshold
precise ratio
AUMP with vertical_min_sup=5%
AUMP with vertical_min_sup=10%
AUMP with vertical_min_sup=15%
Figure 4.6: The precise ratio of AUMP with the values of vertical_min_sup and variance thresh-old varied.
the value of ε to be 12, that of mf to be 3, that of vertical_min_sup to be 20%, and the values of match_min_sup to be 1, 3 and 5, respectively. Figure 4.5 shows the precise ratio of AU M P with the values of match_min_sup and variance threshold σ2 varied. As can been seen in Figure 4.5, the precise ratio of AU M P tends to increase as the value of variance threshold σ2 increases.
This is mainly due to the reason that with larger values of match_min_sup, algorithm LS is likely to identify those similar moving sequences and thus the time projection of the aggregate moving sequence is very related to real moving behaviors of mobile users. Therefore, the precise ratio of AU M P increases drastically when larger values of match_min_sup.
In addition, the hop counts of AU M P with various values of vertical_min_sup and variance threshold σ2 are shown in Figure 4.6, where the match_min_sup is set to be 20%. It can be seen that with a given value of variance threshold σ2, the larger the value of vertical_min_sup, the larger the precise ratio of AU M P .
Chapter 5
Conclusions
In this paper, without increasing the overhead of generating the moving log, we presented a new mining method to mine user moving patterns from the existing log of call detail records of mobile computing systems. Specifically, we proposed algorithm LS to capture similar moving sequences from the log of call detail records and then these similar moving sequences are merged into the aggregate moving sequence. Algorithm LS devised is able to accurately extract those similar moving sequences in the sense that those similar moving sequences are determined by two adjustable threshold values (i.e., vertical_min_sup and match_min_sup) when deriving the aggregate moving sequence from a set of call detail records. By exploring the feature of spatial-temporal locality, which refers to the feature that if the time intervals among consecutive calls of a mobile user is small, the mobile user is likely to move nearby, algorithm TC proposed is able to cluster those call detail records whose occurring time are very close from the aggregate moving sequence. For each cluster of the aggregate moving sequence, algorithm MF devised is employed to derive the estimated moving function, which is able to generate approximate user
moving patterns. Performance of the proposed algorithm was analyzed and sensitivity analysis on several design parameters was conducted. It is shown by our simulation results that the approximate user moving patterns achieved by our proposed algorithms are of very high quality and in fact very close to real user moving behaviors.
Bibliography
[1] M.-S. Chen, J.-S. Park, and P. S.Yu. Efficient Data Mining for Path Traversal Patterns.
IEEE Transactions on Knowledge and Data Engineering, 10(2):209—221, April 1998.
[2] B. Z. W.-C. L. Dik Lun Lee, Jianliang Xu. Data Management in Location-Dependent Information Services. In Proceeding of IEEE Pervasive Computing, pages 65—72, 2002.
[3] M. H. Dunham. Mobile Computing and Databases. Tutorial of International Conference on Data Engineering, February 1998.
[4] EIA/TIA. Cellular Radio Telecommunication Intersystem Operations. 1991.
[5] A. Elmagarmid, J. Jain, and T. Furukawa. Wireless Client/Server Computing for Personal Information Services and Applications. ACM SIGMOD RECORD, 24(4):16—21, December 1995.
[6] J. Han, G. Dong, and Y. Yin. Efficient Mining of Partial Periodic Patterns in Time Series Database. In Proceeding of the 15th International Conference on Data Engineering, March 1999.
[7] J.-T. H. Hsiao-Kuang Wu, Ming-Hui Jin and C.-Y. Ke. Personal Paging Area Design Based On Mobile’s Moving Behaviors. In Proceeding of IEEE INFOCOM, 2001.
[8] S. J.Leon. Linear Algebra with Application. Prentice-Hall International Inc., 2002.
[9] N. Krishnakumar and R. Jain. Escrow Techniques for Mobile Sales and Inventory Applica-tions. ACM Journal of Wireless Network, 3(3):235—246, July 1997.
[10] D. Lam, D. C. Cox, and J. Widom. Teletrafic Modeling for Personal Communication Ser-vices. IEEE Communications, 35(2):79—87, February 1997.
[11] D. L. Lee. Data Management in a Wireless Environment. Tutorial of International Confer-ence on Database System for Advance Applications, April 1999.
[12] Y.-B. Lin. GSM Network Signaling. ACM Mobile Computing and Communications, 1(2):11—
16, 1997.
[13] Y.-B. Lin. Modeling Techniques for Large-Scale PCS Networks. IEEE Communications Magazine, 35(2):102—107, February 1997.
[14] J.-S. Park, M.-S. Chen, and P. S. Yu. Using a Hash-Based Method with Transaction Trim-ming for Mining Association Rules. IEEE Transactions on Knowledge and Data Engineering,
[15] W.-C. Peng and M.-S. Chen. Developing Data Allocation Schemes by Incremental Mining of User Moving Patterns in a Mobile Computing System. In Proceeding of IEEE Transactions on Knowledge and Data Engineering, Volume 15, pages 70—85, 2003.
[16] E. Pitoura and G. Samaras. Locating Objects in Mobile Computing. IEEE Transactions on Knowledge and Data Engineering, 13(4):571—592, July/August 2001.
[17] N. Shivakumar, J. Jannink, and J. Widom. Per-User Profile Replication in Mobile Envi-ronments: Algorithms, Analysis and Simulation Results. ACM Journal of Mobile Networks and Applications, (2):129—140, 1997.
[18] A. P. Sistla, O. Wolfson, S. Chamberlain, and S. Dao. Modeling and Querying Moving Objects. In Proceeding of the 13th International Conference on Data Engineering, pages 422—432, April 1997.
[19] R. L. R. Thomas H. Cormen, Charles E. Leiserson and C. Stein. Introduction to Algorithms.
MIT Press, 2001.
[20] S. Tseng and W. Chan. Efficiently Mining Complete User Moving Path in Mobile Systems.
In Proceeding of International Workshop on Database and Software Engineering (held with ICS), 2002.
[21] S. M. Tseng and C. F. Chiu. An Efficient Method for Mining Associated Service Patterns in Mobile Web Environments. In Proceeding of ACM Symposium on Applied Computing, March 2003.
[22] U. Varshney and R. Vetter. Emerging Mobile and Wireless Networks. Communication of the ACM, 43(6):73—81, June 2000.
[23] R. V.Hogg and E. A. Tanis. Probability and Statistical Inference. Prentice-Hall International Inc., 1997.