Grid Scalable AdHoc Wireless Networking Douglas De Couto
Grid: Scalable Ad-Hoc Wireless Networking Douglas De Couto http: //pdos. lcs. mit. edu/grid
Goal: Networks out of Chaos B A C I D F E G H J
Ad hoc Applications • Temporary, fast setup • Emergencies & events • Rooftop networks • No wires, trenches, etc. • Developing communities • Cheap, incremental, automatic
Direct Contact Scales Badly A B C “Hello J!” I D F E G H J
Solution: Multi-hop Forwarding A B C “A to J: Hello!” I D F E G H J
Design Challenges • • • Finding routes Cope with mobile nodes Conserving battery power Coping with malicious/faulty nodes Scaling to large networks
Completed Research • Scalable routing: • Geographic forwarding • Distributed P 2 P location database • • Low-power forwarding Understanding capacity limits Avoiding malicious nodes Current research: link selection
System Status • Software distributions for • Linux, BSD • PC, i. Paq • Works with unmodified Internet software • Two Grid nets deployed • In-building network • Rooftop network
LCS Grid Net 5 6 5 5 6 6 5 6 • 17 static nodes on 5 th/6 th floors • A dozen i. Paq hand-helds 5 6 5 5 wired gateway
Roof-Top Grid Net 6 5 4 3 2 1 LCS
Geographic forwarding (GF) C’s radio range A C B D F G E • Packets addressed to id. G, location. G • Next hop is chosen from neighbors to move packet geographically closer to destination location • Per-node routing overhead constant as network size (nodes, area) grows • Requires location service, which adds overhead
Grid Location Service (GLS) overview E B H L D J G A F “D? ” I K C Each node has a few servers that know its location. 1. Node D sends location updates to its servers (B, H, K). 2. Node J sends a query for D to one of D’s close servers.
GLS’s Spatial Hierarchy level-0 level-1 level-2 level-3 All nodes agree on the global origin of the grid hierarchy
3 servers per node per level sibling level-0 squares sibling level-1 squares sibling level-2 squares s n s s s s • s is n’s successor in that square. (Successor is the node with “least ID greater than” n )
Queries search for destination’s successors s n s s s Each query step: visit n’s successor at increasing levels, until location server found s s 1 s 2 s s 3 location query path x
GF + GLS performs well Fraction of data packets delivered successfully Grid DSR Biggest network simulated: 600 nodes, 2900 x 2900 m (4 -level grid hierarchy) Number of nodes • Geographic forwarding is less fragile than source routing. • DSR queries use too much b/w with > 300 nodes.
GLS properties • • Spreads load evenly over all nodes Degrades gracefully as nodes fail Queries for nearby nodes stay local Per-node storage and communication costs grow slowly as the network size grows: O(log n), n nodes • More details: Li et al, Mobicom 2000
Does Grid Find Useful Paths? B A C I D F E G H J
Mistake: Shortest-Path Routes B A C I D F E A’s max range G H J
Link Quality Isn’t Bi-modal
Route metrics • How to select good routes? • Compare metrics • Good metric: expected total packet transmissions • Want to mimimize • Route metric = sum of link metrics • Fight strong bias towards shortest paths • While penalizing longer paths
Obstacles to Better Routing • Want to detect and avoid lossy links, but… • Loss rate masked by 802. 11 re-sends • Changes quickly with time, motion
How to find loss rate? • Signal strength?
Current Work • Trying to directly measure loss rates • Route broadcast packets • Long time constants • 802. 11 protocol beacons? • Requires driver integration
Grid Summary • Grid routing protocols are • Self-configuring • Easy to deploy • Scalable http: //www. pdos. lcs. mit. edu/grid
End Of Talk Demo
Application: Smart Devices Remote Control Print Share E-Mail Access Point Internet
Application: Rooftop Nets Game server School/Homework Server Internet Access
Application: Disaster Services • Disaster may have damaged phone system &c • Want to avoid N 2 plans for N services to communicate
Topology Distribution Scales Badly A B C 1. “C can reach A and B. ” 3. Data from F to B. D 2. “D can reach A, B, and C. ” F G
Geographic Forwarding Scales Well Latitude A B C “Send towards lat. G / lon. G. ” D F E G Longitude
Location Database Latitude A B DB 2. “Where is G? ” C 1. “G is at lat. G / lon. G” D F E G Longitude
Distributed Location Database • • Each node is DB for a few other nodes How to find a node’s location server(s)? Every node has an unchanging ID hash(ID) maps ID to position in unit square
G’s Location Server is a Point H x I hash(G) = 0. 1, 0. 9 G (0, 0)
Spatial Grid Hierarchy All nodes agree on the global origin of the Grid hierarchy
Multiple Servers per Node c b a G
Lookups Expand in Scope c b a ? A G
Protocol Overhead (packets per second) Grid Protocol Overhead Grows Slowly Number of nodes • Protocol packets include: Grid update, Grid query/reply.
- Slides: 38