4 Simulation Results
4.2 Experiment Results
We use different parameters such as tree level, cost value, and measure day to observe the effect of the simulation. Besides, for the dynamic location tracking strategy, we assume that the system performs computing the appropriate LA for users every three and five days.
First, we use different levels of the tree structure to compare their performance (See Figure 9 and Figure10), our strategy performs better than the basic and eager caching strategies. In this case, we can see that the eager caching performs worst when the CMR is smaller than 6. When the CMR is small, it implies that users may move to neighbor RAs frequently. When the user moves, the eager caching has to pay an extra cost to delete invalid bypass pointers. When the CMR is large, it implies that the user receives call frequently rather than moves. The eager caching has better performance than the basic because of using the bypass pointers to save the call setup time. Our strategy has the same performance in different tree structure. When the CMR is small, our strategy is better than the eager caching and basic, because it saves the cost of location registration. Our strategy uses the local anchor to keep the user’s location information.
6000
Figure 9: L=6, M=4, MT=100,000, D=3, FT=5, U=2, R=1, C=0.2
8000
Figure 10: L=7, M=3, MT=100,000, D=3, FT=5, U=2, R=1, C=0.2
When the CMR is large, our strategy still performs well because of the fact that the DLT also uses bypass pointers to efficiently reduce the cost of the call delivery operation.
Furthermore, the DLT uses the appropriate position deciding scheme to evaluate whether the position of the user’s LA should be adjusted. When the CMR is small, the LA will adjust its position to be close to the region where the user frequently moves. Thus the cost of location registration can be greatly saved . Besides, the position which the LA is adjusted to is also the region where the user receive calls frequently. When the CMR is large, the find events are much more than move events. The LA at a higher level of the tree is better than that at a lower level of the tree. Hence, the location of the LA is the same as before, it implies that the location of the LA won’t be changed.
Figure 10 and Figure 11 shows that the appropriate LA deciding scheme is invoked in three and five days respectively, and users rarely change their behavior. We can see if a user rarely changes his/her behavior, and then his/her LA will keep static according to our appropriate position deciding scheme. The system can extend the period for invoking the dynamic adjust procedure. In Figure 10, Figure 12, and Figure 13, we use different cost value to compare the performance. When the communication cost is more expensive than the cost of accessing databases, the eager caching does not perform well because of paying an addiction cost to delete invalid bypass pointers when the user leaves database t. If several
bypass pointers cache this location t, the cost of deleting invalid bypass pointers will increase dramatically. Our strategy doesn’t delete bypass pointers when the user leaves database t.
The bypass pointers are updated through the acknowledgement messages of the call delivery procedure.
Finally, we compare the difference between the DLT before invoking the appropriate position deciding scheme (DLTbefore) and the DLT after invoking the appropriate position deciding scheme (DLTafter) in Figure 14. When the CMR is small, the performance of DLTafter is better than DLTbefore. We see that the move events do influence the result of deciding which location is suitable for the user’s LA. As the CMR increases, DLTbefore and DLTafter will be the same because of the fact that call events are more than move events.
When the user’s call events are more than move events, the LA won’t be adjusted to a lower level of the tree.
10000 15000 20000 25000 30000 35000 40000
2 4 6 8 10 12 14 16 18 20
Total cost[1:10000]
CMR
Basic Eager-Caching DLT
Figure 11: L=7, M=3, MT=100,000, D=5, FT=5, U=2, R=1, C=0.2
10000 15000 20000 25000 30000 35000 40000
2 4 6 8 10 12 14 16 18 20
Total cost[1:10000]
CMR
Basic Eager-Caching DLT
Figure 12: L=7, M=3, MT=100,000, D=3, FT=5, U=2, R=1, C=1
8000 10000 12000 14000 16000 18000 20000 22000 24000
2 4 6 8 10 12 14 16 18 20
Total cost[1:10000]
CMR
Basic Eager-Caching DLT
Figure 13: L=7, M=3, MT=100,000, D=3, FT=5, U=0.4, R=0.2, C=1
0.4 0.6 0.8 1 1.2 1.4
2 4 6 8 10 12 14 16 18 20
Cost Ratio
CMR
DLTafter/DLTbefore
Figure 14: L=7, M=3, MT=100,000, D=3, FT=5, U=0.4, R=0.2, C=1
5 Conclusion
The main contribution of the paper is to propose the dynamic location tracking strategy, which dynamically adjusts the position of the local anchor according to the user’s behavior.
Our strategy is based on the caching strategy, but modifies the registration and call delivery procedures to get a great benefit in reducing communication and database access costs.
For registration procedures, we uses the LA to track a user’s current location, and it can greatly reduce the registration cost. For call delivery procedures, we make the forward bypass pointers cache each user’s LA, which can avoid cache miss when the user leaves t.
Besides, the appropriate position deciding scheme is invoked in our strategy to decide which position is better for a user’s LA according to his/her mobility and calling patterns. Then we adjust his/her LA to the best position. We can get a great benefit in location registration through this dynamic adjusting LA procedure. Besides, while the position of the LA has been changed from t to t, several forward bypass pointers which cache the previous position t may need to be updated at the same time. In our strategy, these bypass pointers will be updated through the acknowledgement messages of the call delivery procedure. In other words, the bypass pointer will be updated through the acknowledgement message only when it is used by a user to perform the call delivery procedure. Thus we do not need to update several bypass pointers at once.
With a small CMR, the frequency of successive move operations is high. Our strategy greatly reduces registration cost by using a LA to keep a user’s current location. With a large CMR, the frequency of successive find operations is high. Our strategy still greatly reduces call delivery cost by using bypass pointers to locate mobile users. In our strategy, each location database which serves as a LA must store a user’s information and provide the computing function for adjusting LA dynamically. Besides, each user’s LA will not locate at the same level; thus all users’ profiles are efficiently distributed in the tree structure.
We compare the performance of different strategies in the simulation. The results clearly indicate that our strategy has better performance than the basic and caching strategies.
References
[1] EIA/TIA,”Cellular radio-telecommunicaiton intersystem operations,”Tech. Rep. IS-41 Reversion B, EIA/TIA, 1991.
[2] I.F. Akyildiz, J. McNair, J. Ho, H. Uzunalioglu and W. Wang, ”Mobility management in next-generation wireless systems,” Proc. IEEE, vol. 87 , no. 8 , pp. 1347 -1384, Aug.
1999.
[3] C. Eynard, M. Lenti, A. Lombardo, O. Marengo, and S. Palazzo, “A methodology for the performance evaluation of data query strategies in universal mobile telecommunication systems (UMTS),” IEEE Journal Selected Areas in Communications, vol. 13, no. 5, pp.
893-907, June 1995.
[4] J. S. M. Ho and I. F. Akyildiz, ”Local anchor schmem for reducing signaling costs in personal communications networks,” IEEE/ACM Trans. on Networking, vol. 4, no. 5, Oct. 1996,pp. 709-725
[5] J.S.M. Ho and I.F. Akyildiz, “Dynamic hierarchical database architecture for location management in PCS networks,” IEEE/ACM Transactions on Networking, vol. 5, no. 5, pp. 646-660, Oct. 1997.
[6] H.-C. Lin and S.L. Lee, “A Presetting Location Strategy for Personal Communication Using Hierarchical Location Database,” Proceeding of International Conference on Par-allel and Distributed System, pp. 349-356, 2000.
[7] Y.-B. Lin and W.-N. Tsai, “Location tracking with distributed HLR’s and pointer for-warding,” IEEE Transactions on Vehicular Technology, vol. 47, no. 1, pp. 58-64, Feb.
1998.
[8] R. Jain, Y.-B. Lin, C. Lo, and S. Mohan, “A caching strategy to reduce network impacts of PCS,” IEEE Journal Selected Areas in Communications, vol. 12, no. 8, pp. 1434-1444, Oct. 1994.
[9] R. Jain and Y.-B. Lin, “An auxiliary user location strategy employing forwarding point-ers to reduce network impacts of PCS,” IEEE International Conference in Communi-cations, vol. 2, pp. 740-744, 1995.
[10] R. Jain, “Reducing traffic impacts of PCS using hierarchical user location databases,”
IEEE International Conference in Communications, vol. 2, pp. 1153-1157, 1996.
[11] R. Jain and F. Anjum, “Caching in hierarchical user location databases for PCS,” IEEE International Conference on Personal Wireless Communication, pp. 496-500, 1999.
[12] S. Mohan and R. Jain, “Two user location strategies for personal communications ser-vices,” IEEE Personal Communications, vol. 1, no. 1, pp. 42-50, First Quarter 1994.
[13] M. Mouly and M. B. Pautet, ”The GSM system for personal communication services,”
M. Mouly, 49 rue Louise Bruneau, Palaiseau, France, 1992.
[14] E. Pitoura, G. Samaras,”Locating objects in mobile computing,” IEEE Transactions on Knowledge and Data Engineering, vol. 13 , no. 4 , pp. 571-592, July-Aug. 2001.
[15] K.-L. Sue and C.-C. Tseng “One-step pointer forwarding strategy for location tracking in distributed HLR environment,” IEEE Journal Selected Areas in Communications, vol. 15, no. 8, pp. 1455-1466, Oct. 1997.
[16] N. Shivakumar and J. Widom, ”User profile replication for faster location lookup in mobile environments,” in Proc. ACM MOBICOM’95, Nov. 1995, pp. 161-169.
[17] S. Tabbane, “Location Management Methods for Third-Generation Mobile System,”
IEEE Personal Communications., pp. 72-84, Aug. 1997.
[18] J.Z. Wang, “A fully distributed location registration strategy for universal personal communication systems,” IEEE Journal Selected Areas in Communications, vol. 11, no.
6, pp. 850-860, Aug. 1993.