SEEKER An Adaptive and Scalable Location Service for

SEEKER: An Adaptive and Scalable Location Service for Mobile Ad Hoc Networks Jehn-Ruey Jiang and Wei-Jiun Ling Presented by Jehn-Ruey Jiang National Central University

Outline • • • ICS 2006 Introduction Related Work The SEEKER Location Service Simulation Result Conclusion NCU ACN Lab. 2

Categories of Routing Protocols for MANETs • Topology-Based routing protocols: – AODV, DSR …etc More Scalable • Position-Based routing protocols: – Each node knows its location via a GPS or the like – A source node knows the destination location – Routing via greedy forwarding or the like ICS 2006 NCU ACN Lab. 3

Greedy Forwarding e b D c S a ICS 2006 NCU ACN Lab. 4

Location Service • Position-Based routing requires a location service to support : – location registration – location update – location query ICS 2006 NCU ACN Lab. 5

Problem • A good location service must have : – High scalability – High query accuracy and success rate – Low protocol overhead • Update and maintenance overhead • Query overhead • How to design a good location service? ICS 2006 NCU ACN Lab. 6

Outline • • • ICS 2006 Introduction Related work The SEEKER Location Service Simulation Result Conclusion NCU ACN Lab. 7

The Classification of Rendezvous-Based Location Services ICS 2006 NCU ACN Lab. 8

Quorum-Based (XYLS) [Stojmenovic 1999] Updates Intersection ICS 2006 Queries NCU ACN Lab. 9

Hierarchical-Based (GLS) [Li et al. @Mobi. Com 2000] 23 27 87 91 75 12 32 74 98 26, 31, 32, 74, 75, 79, 98 81 ICS 2006 23 23 23 2 62 26 31 23 28 NCU ACN Lab. 23 17 55 56 43 23 63 99 33 79 8 12 4 10

Hash-Based (SLURP) [S. -C. M 2001] Region Home Region B A A ICS 2006 NCU ACN Lab. 11

Outline • • • ICS 2006 Introduction Related Work The SEEKER Location Service Simulation Result Conclusion NCU ACN Lab. 12

SEEKER - Overview • Control overhead can be reduced by aggregate update • Adapt to mobility by adjusting the frequency of location updates to further reduce overhead: adaptive update • Goals: to achieve – High scalability – High query accuracy and success rate – Low protocol overhead ICS 2006 NCU ACN Lab. 13

SEEKER – Aggregate Update • Initial nodes send LOC packet (location update packet) periodically. • The initial node sets a virtual destination to deliver the LOC packet to. • The LOC packet is relayed by greedy forwarding until it reaches the terminal node. • The nodes in the path from the initial node to the terminal node serve as the location servers for all the nodes in the same row of regions. ICS 2006 NCU ACN Lab. 14

SEEKER – Location Update Terminal node Initial node : Self-selection by testing neighbors in the east and west Initial node Virtual destination ICS 2006 NCU ACN Lab. Ex. If no node is to the west of a node, it will be the initial node to send LOC to the east 15

SEEKER – Location Query Unicast Query. Request C B Broadcast Query Request Unicast Query. Reply A Broadcast Query Reply ICS 2006 NCU ACN Lab. 16

SEEKER - Adaptive Location Update • Observation: there is a tradeoff between update interval and query accuracy • Basic idea: let the terminal node calculate the average speed of nodes of each row • Procedures: 1. Calculate the average speed and transform it into the new update interval by the terminal node 2. Spread the new update interval to the terminal node’s neighbors 3. Spread the new update interval by LOC packets ICS 2006 NCU ACN Lab. 17

Transform Average Speed into Update Interval • Define speed levels, – ex: 2. 5 m/s, 5. 0 m/s, 7. 5 m/s, 10. 0 m/s, and 12. 5 m/s • Adopt ½ difference (1. 25 m/s) to be interval boundaries • Next update interval = Transmission range / Speed level 0 1. 25 200 (s) 2. 5 100 (s) Update interval ICS 2006 5. 0 50 (s) 7. 5 30 (s) 10. 0 25 (s) 12. 5 m/s 20 (s) Range of Speed levels NCU ACN Lab. 18

Procedure of Adaptive Location Update Hello packet A LOC Next update ICS 2006 NCU ACN Lab. 19

Outline • • • ICS 2006 Introduction Related Work The SEEKER Location Service Simulation Result Conclusion NCU ACN Lab. 20

Simulation Result • Simulation environment – – Simulate by ns 2 Node density within 100 nodes/km 2 Random waypoint with speed 0~10 m/s 300 seconds per simulation • Metrics – Control overhead (update + query) – QSR (Query Success Rate) – PDR (Packet Delivery Rate) ICS 2006 NCU ACN Lab. 21

Comparison • • ICS 2006 Up to 600 nodes Compared with GLS and HIGH-GRADE Bandwidth is 2 Mbps The performance of fixed update interval vs. adaptive update interval NCU ACN Lab. 22

Update Overhead ICS 2006 NCU ACN Lab. 23

Query Overhead (reply excluded) ICS 2006 NCU ACN Lab. 24

Query Success Rate (QSR) ICS 2006 NCU ACN Lab. 25

Packet Delivery Rate (PDR) ICS 2006 NCU ACN Lab. 26

Outline • • • ICS 2006 Introduction Related Work The SEEKER Location Service Simulation Result Conclusion NCU ACN Lab. 27

Conclusion • SEEKER achieves comparably low control overhead by using aggregate update • Control overhead (esp. update overhead) can be further reduced by adaptive update • SEEKER achieves – – ICS 2006 comparably high scalability (2000 nodes) comparably high query success rate comparably high query accuracy comparably low maintenance cost NCU ACN Lab. 28

Thank you! Questions and Comments? ICS 2006 NCU ACN Lab. 29