230 likes | 332 Views
Cache Invalidation and Replacement Strategies for Location-Dependent Data in Mobile Environments. Speaker : Hsin-Chin Mao Fu Jen Catholic University Computer Science and Information Engineering Department High Speed Networks Lab 2003/10/28. Outline. Introduction The System Model
E N D
Cache Invalidation and Replacement Strategies for Location-Dependent Data in Mobile Environments Speaker : Hsin-Chin Mao Fu Jen Catholic University Computer Science and Information Engineering Department High Speed Networks Lab 2003/10/28
Outline • Introduction • The System Model • Location-Dependent Invalidation Strategies • Location-Dependent Cache Replacement Policies • Simulation Model • Performance Evaluation • Conclusion • References
two common issues in client cache management cache invalidation scheme cache replacement policy location-dependent data location-dependent cache invalidation valid scopes We first introduce two basic location-dependent invalidation schemes Polygonal Endpoints (PE) Approximate Circle (AC) a generic method Cache-Efficiency Based scheme (CEB) Introduction
The System Model • two distinct sets of entities • mobile clients • fixed hosts ( mobile support stations (MSSs)) • data item value from data item • Mobile clients can identify their locations using systems such as the Global Positioning System (GPS) • cache data values on its local disk or in any storage system; fixed sizes and read-only
Location-Dependent Invalidation Strategies • The advantages of the idea that attach complete/partial invalidation information • two situations where validity checking is necessary • cache replacement policies • The Polygonal Endpoints (PE) Scheme • a straightforward way • The Approximate Circle (AC) Scheme • the overhead of this scheme can be minimized • 56 bytes => 12 bytes
Location-Dependent Invalidation Strategies • The Caching-Efficiency-Based (CEB) Method
Location-Dependent Cache Replacement Policies • Data Distance • the distance between the current location of a mobile client and the valid scope of a data value • Valid Scope Area • the geometric area of the valid scope of a data value proposed PA and PAID policies • Probability Area (PA) • Probability Area Inverse Distance (PAID)
Simulation Model • System Execution Model • 110 points randomly distributed in a square Euclidean space • the locations of 185 hospitals in the Southern California area • Server Execution Model • Client Execution Model
Performance Evaluation • Evaluation of Location-Dependent Invalidation Schemes • Evaluation of Cache Replacement Policies • uniform access (θ=0), skewed access(θ=0.5) • Effect of Changing Query Interval • Effect of Changing Moving Interval • Effect of Cache Size • Effect of Combining Different Invalidation and Replacement Schemes
Effect of Combining Different Invalidation andReplacement Schemes
Conclusions • explored cache invalidation and replacement issues for location-dependent data under a geometric location model • PE, AC, CEB • proposed two cache replacement policies • PA, PAID • future work • location-dependent queries
References • Baihua Zheng, Jianliang Xu, Dik Lun Lee: Cache Invalidation and Replacement Strategies for Location-Dependent Data in Mobile Environments. IEEE Transactions on Computers 51(10): 1141-1153 (2002) • Q. Ren and M.H. Dunham, “Using Semantic Caching to Manage Location Dependent Data in Mobile Computing,” Proc. Sixth Ann. ACM/IEEE Int’l Conf. Mobile Computing and Networking (MobiCom 2000), pp. 210-221, Aug. 2000. • G.K. Zipf, Human Behaviour and the Principle of Least Effort.Addison-Wesley, 1949.