1. Adaptive Location Management Model for Manet (ALM) Navid NIKAEIN Christian BONNET http://www.eurecom.fr/~nikaeinn

2. Outline  Introduction  Motivation  Intuition and Basic Ideas  Adaptive Location Management Model  Conclusion

3. Introduction  Routing: How to achieve scalability? Topology-based  Hierarchical Architecture Position-based  Location Management  Location Management  Location Directory Location Update Location Search  Design choice of Location Directory Reactive: e.g. LAR[Ko&Vaidya]  Flooding Proactive: e.g. DREAM[Basagni]  Flooding Hybrid: e.g. GLS[Li]

4. Flooding Overhead Number of nodes Avg. packets transmitted per node per second Ref. GLS[Li]

5. Motivation  Scalability in routing for Manet  Flooding-based routing protocols fail to achieve scalability as the: Frequency of end-to-end connections increases Frequency of topology changes increases Total number of nodes increases

6. Intuition & Basic Ideas I Provide multiple location servers replicated at several geographical positions Hierarchical Grid: GLS, SLURP, SLALoM, DLM Graph-based: Archimedean Spiral, Concentric circles, Epi Spiral: ALM

7. Grid vs. Graph-based Architecture L 0 L 1 L 2 Hierarchical GridGraph-based L 3

8. Archimedean Spiral: -id can be the origin or origin=Hash(id) -R=aθ, where θ= θ + / 2 Distribution of Location Servers Location Sever 4a4a 2a2a 6a6a Y X

9. Intuition & Basic Ideas II Mobility RateDistance ALM Location update / Search interval Location update /search Zone Inputs Outputs Our location update/search procedures employ an Adaptive time-based scheme on node granularity

10. Spread load evenly among servers Degrade gracefully as servers fail The communication overhead is optimized on the distribution Location Update LU, A> 9 m/s 3a m speeddistance ALM 26 s5 # timezone LU Updated Servers

11. Location Search I LS Queries for nearby nodes remain local Avoid the overhead of servers’ search LR , A> 17 m/s 2a m speeddistance ALM 15 s3 # timezone, B> b

12. Location Search II, B> LS Queries for far away nodes remain in the neighborhood Load balancing among servers LR , A> 0 Searched Servers

13. Conclusion We present a graph-based model to distribute location servers We also present an adaptive time- based scheme on per-user basis  Fuzzy logic process is used to deal with imprecise and uncertain information

14. Future Work Consider call arrival rate for location update procedure Study the performance of different distributions (graphs) Evaluate the performance of our approach with the related works

15. Location Search III, B> LS Location servers remains active for location search time interval, here 15 seconds, A> 0

16. Location Search III, B>, A>

17. Intuition & Basic Ideas III  As the mobility rate increases: the location update/search message have to be sent more frequently to the small number of location servers.  As the distance increases: the location update/search message have to be sent less frequently to the large number of location servers

18. Node Addressing  An address represents: Position is a location dependent address and reflects the current Geo-spatial position of nodes  A Position represents: id is a location independent identifier which is unique and well-known in the network