Скачать презентацию Routing in Large Scale Ad Hoc and Sensor Скачать презентацию Routing in Large Scale Ad Hoc and Sensor

d35342c0958d262d4bcfce9eab01a90b.ppt

  • Количество слайдов: 67

Routing in Large Scale Ad Hoc and Sensor Networks Ten H. Lai Ohio State Routing in Large Scale Ad Hoc and Sensor Networks Ten H. Lai Ohio State University

Two Approaches n Traditional routing algorithms adapted to ad hoc networks n Geographical routing Two Approaches n Traditional routing algorithms adapted to ad hoc networks n Geographical routing 2

Review of Routing n Next-hop routing n Source routing n Flooding 3 Review of Routing n Next-hop routing n Source routing n Flooding 3

Next-Hop Routing destination x y. . . next hop a c cost 3 5 Next-Hop Routing destination x y. . . next hop a c cost 3 5 X a ? y c Which neighbor (next hop)? 4

Source Routing destination x y. . . path (a, b, c) a cost b Source Routing destination x y. . . path (a, b, c) a cost b c X Which path? 5

Link-State Routing n Each node periodically broadcasts the link states of its outgoing links Link-State Routing n Each node periodically broadcasts the link states of its outgoing links to the entire network (by flooding). n As a node receives this information, it updates its view of the network topology and routing table. 2 5 3 4 1 4 3 6

Distance-Vector Routing least-cost(A, B) = min {cost(A, x) + least-cost(x, B): for all neighbors, Distance-Vector Routing least-cost(A, B) = min {cost(A, x) + least-cost(x, B): for all neighbors, x, of A} n Neighbors exchange distance vectors n Destination A B C D Distance 0 10 … x A E F G B C 7

Routing in MANETs n Every node works as a router 8 Routing in MANETs n Every node works as a router 8

Challenges n Quick topology changes n Scalability 9 Challenges n Quick topology changes n Scalability 9

Two Approaches n Table-driven v Like existing Internet routing protocols n On-demand 10 Two Approaches n Table-driven v Like existing Internet routing protocols n On-demand 10

Table-Driven Routing Protocols l Also called proactive routing protocols l Continuously evaluate the routes Table-Driven Routing Protocols l Also called proactive routing protocols l Continuously evaluate the routes l Attempt to maintain consistent, up-to-date routing information Øwhen a route is needed, it is ready immediately l When the network topology changes Øthe protocol responds by propagating updates throughout the network to maintain a consistent view 11

On-Demand Routing Protocols v Also called reactive routing protocols v Discover routes when needed On-Demand Routing Protocols v Also called reactive routing protocols v Discover routes when needed by the source node. v Longer delay 12

Early Ad Hoc Routing Protocols 13 Early Ad Hoc Routing Protocols 13

DSDV: Destination Sequence Distance Vector n “Highly Dynamic Destination-Sequence Distance-Vector Routing (DSDV) for Mobile DSDV: Destination Sequence Distance Vector n “Highly Dynamic Destination-Sequence Distance-Vector Routing (DSDV) for Mobile Computers” n Charles E. Perkins & Pravin Bhagwat n Computer Communications Review, 1994 n pp. 234 -244 14

DSDV Overview n DSDV = destination-sequenced distance- vector n Distance-vector routing n Each entry DSDV Overview n DSDV = destination-sequenced distance- vector n Distance-vector routing n Each entry is tagged with a sequence number originated by the destination node. Destination A B C D Distance 0 10 … Sequence # E F G 15

DSDV Route Advertisement n Each node periodically broadcasts its distance vector. v “broadcast” is DSDV Route Advertisement n Each node periodically broadcasts its distance vector. v “broadcast” is limited to one hop. v sequence numbers ØFor the sender’s entry: Sender’s new sequence number (typically, +1) ØFor other entries: originally “stamped” by the destination nodes Destination A B C D Distance 0 10 … Sequence # E F G 16

DSDV Route Updating Rules n Paths with more recent seq. nos. are always preferred. DSDV Route Updating Rules n Paths with more recent seq. nos. are always preferred. n least-cost(A, B) = min {cost(A, x) + least-cost(x, B): for all neighbors, x, of A} x A B C 17

(Source-Initiated) On-Demand Routing Protocols n. DSR n. AODV n. ABR n. SSR n. ZRP (Source-Initiated) On-Demand Routing Protocols n. DSR n. AODV n. ABR n. SSR n. ZRP 18

DSR: Dynamic Source Routing n “Dynamic Source Routing in Ad-Hoc Wireless Networks” n D. DSR: Dynamic Source Routing n “Dynamic Source Routing in Ad-Hoc Wireless Networks” n D. B. Johnson and D. A. Maltz n Mobile Computing, 1996 n pp. 153 -181 19

DSR : Outline n Source Routing n On-demand n Each host maintains a route DSR : Outline n Source Routing n On-demand n Each host maintains a route cache containing all routes it has learned. n Two major parts: v route discovery v route maintenance 20

Route Discovery of DSR n To send a packet, a source node first consults Route Discovery of DSR n To send a packet, a source node first consults its route cache. v If there is an unexpired route, use it. v Otherwise, initiate a route discovery. n Route Discovery: v Source node launches a ROUTE_REQUEST by flooding. v A ROUTE_REPLY is generated when Ø the route request reaches the destination Ø an intermediate node has an unexpired route to the destination 21

Stale Route Cache Problem n Definition: v A cached route may become stale before Stale Route Cache Problem n Definition: v A cached route may become stale before it expires. x x 22

Route Maintenance of DSR n When a node detects a link breakage, it generates Route Maintenance of DSR n When a node detects a link breakage, it generates a ROUTE_ERROR packet. v The packet traverses to the source in the backward direction. v The source removes all contaminated routes, and if necessary, initiates another ROUTE_REQUEST. x x B 23

AODV: Ad-Hoc On-Demand Distance Vector Routing n “Ad-hoc On-Demand Distance Vector Routing” n Charles AODV: Ad-Hoc On-Demand Distance Vector Routing n “Ad-hoc On-Demand Distance Vector Routing” n Charles E Perkins, Elizabeth M Royer n Proc. 2 nd IEEE Wksp. Mobile Comp. Sys. and Apps. , Feb. 1999. 24

AODV : Outline n Next-hop Routing (cf. DSR: source routing) n On-demand n Each AODV : Outline n Next-hop Routing (cf. DSR: source routing) n On-demand n Each host maintains a routing table n Two major parts: v route discovery (by flooding) v route maintenance 25

AODV vs. DSR n DSR: Routes are discovered and cached n AODV: Next-hop info AODV vs. DSR n DSR: Routes are discovered and cached n AODV: Next-hop info is stored n “Performance Comparison of Two On-Demand Routing Protocols for Ad Hoc Networks, ” Personal Communications, February 2001 26

ABR: Associativity-Based Routing n “Associativity-Based Routing for Ad-Hoc Mobile Networks, ” C. K. Toh. ABR: Associativity-Based Routing n “Associativity-Based Routing for Ad-Hoc Mobile Networks, ” C. K. Toh. n ABR considers the stability of a link. vcalled the degree of association stability. vmeasured by the number of beacons received from the other end of the link. v. The higher degree of a link’s stability, the lower mobility of the node at the link’s other end. 27

ABR Outline n Route Discovery: v Same as DSR except the following. v Each ABR Outline n Route Discovery: v Same as DSR except the following. v Each ROUTE_REQUEST packet collects the association stability information along its path to the destination. v The destination node selects the best route in terms of association stability. 28

n Route Reconstruction: v On route error, a node performs a local search in n Route Reconstruction: v On route error, a node performs a local search in hope of repairing the path. v If the local search fails, a ROUTE_ERROR is reported to the source local searched zone destination 29

SSA: Signal Stability-Based Adaptive Routing n “Signal Stability-Based Adaptive Routing (SSA) for Ad Hoc SSA: Signal Stability-Based Adaptive Routing n “Signal Stability-Based Adaptive Routing (SSA) for Ad Hoc Wireless Networks” n University of Maryland n R. Dube, C. D. Rais, K. -Y. Wang & S. K. Tripathi n IEEE Personal Communications, ‘ 97 30

Basic Idea of SSA n Observation: v The ABR only considers the connectivity stability. Basic Idea of SSA n Observation: v The ABR only considers the connectivity stability. n Two more metrics: v signal stability: Øthe strength of signal over a link v location stability Øhow fast a host moves 31

ZRP: Zone Routing Protocol n The Zone Routing Protocol (ZRP) for Ad Hoc Networks ZRP: Zone Routing Protocol n The Zone Routing Protocol (ZRP) for Ad Hoc Networks n Cornell University n Z. J. Haas and M. R. Pearlman n draft-ietf-manet-zone-zrp-01. txt, 1998 32

ZRP Outline n Hybrid of table-driven and on-demand!! n Each node is associated with ZRP Outline n Hybrid of table-driven and on-demand!! n Each node is associated with a zone. n Within a zone: table-driven (proactive) routing. n Inter-zone: on-demand routing (similar to DSR). 33

Route Discovery n By an operation called “boardercast”: v sending the route-request to boarder Route Discovery n By an operation called “boardercast”: v sending the route-request to boarder nodes 34

ZRP Example 35 ZRP Example 35

Scalability Problem in Large-Scale Network Routing n Internet solution 36 Scalability Problem in Large-Scale Network Routing n Internet solution 36

Geographic Routing n Make use of location information in routing Geographic Routing n Make use of location information in routing

Assumptions n Each node knows of its own location. v outdoor positioning device: ØGPS: Assumptions n Each node knows of its own location. v outdoor positioning device: ØGPS: global positioning system Øaccuracy: in about 5 to 50 meters v indoor positioning device: ØInfrared Øshort-distance radio n The destination’s location is also known. v How? (via a location service) 38

LAR: Location-Aided Routing n Location-Aided Routing (LAR) in mobile ad hoc networks n Young-Bae LAR: Location-Aided Routing n Location-Aided Routing (LAR) in mobile ad hoc networks n Young-Bae Ko and Nitin H. Vaidya n Texas A&M University n Wireless Networks 6 (2000) 307– 321 39

Basic Idea of LAR n All packets carry sender’s current location. n This info Basic Idea of LAR n All packets carry sender’s current location. n This info enables nodes to learn of each other’s location. 40

Basic Idea of LAR (cont. ) n Same as DSR, except that if the Basic Idea of LAR (cont. ) n Same as DSR, except that if the destination’s location is known, the ROUTE_REQ is only flooded over the “route search zone. ” D Expected zone of D S Route search zone 41

DREAM n. A Distance Routing Effect Algorithm for Mobility (DREAM) n S. Basagni, I. DREAM n. A Distance Routing Effect Algorithm for Mobility (DREAM) n S. Basagni, I. Chlamtac, V. R. Syrotiuk, B. A. Woodward n The University of Texas at Dallas n Mobicom’ 98 42

Basic Idea of DREAM n Dissemination of location information: v Each node periodically advertises Basic Idea of DREAM n Dissemination of location information: v Each node periodically advertises its location (and movement information) by flooding. v This way, nodes have knowledge of one another’s location. 43

Basic Idea of DREAM n Data Packet carries D’s and S’s locations. n Forwarded Basic Idea of DREAM n Data Packet carries D’s and S’s locations. n Forwarded toward only a certain direction. D Expected zone of D S 44

GRID Routing n “GRID: A Fully Location-Aware Routing Protocol for Mobile Ad Hoc Networks” GRID Routing n “GRID: A Fully Location-Aware Routing Protocol for Mobile Ad Hoc Networks” n Wen-Hwa Liao, Yu-Chee Tseng, Jang-Ping Sheu n NCTU n Telecommunication Systems, 2001. 45

Basic Idea of GRID Routing n Partition the physical area into d x d Basic Idea of GRID Routing n Partition the physical area into d x d squares called grids. 46

Protocol Overview n In each grid, a leader is elected, called gateway. n Responsibility Protocol Overview n In each grid, a leader is elected, called gateway. n Responsibility of gateways: v forward route discovery packets v propagate data packets to neighbor grids v maintain routes which passes the grid n Routing is performed in a grid-by-grid manner. 47

Route Search Range Options 48 Route Search Range Options 48

Strength of Grid Routing x x 49 Strength of Grid Routing x x 49

Gateway Election in a Grid n Any “leader election” protocol in distributed computing can Gateway Election in a Grid n Any “leader election” protocol in distributed computing can be used. n Multiple leaders in a grid are acceptable. n Preference in electing a gateway: v near the physical center of the grid Ølikely to remain in the grid for longer time v once elected, a gateway remains so until leaving the grid 50

Taxonomy of Geographic Routing Algorithms n Also called position-based routing n Three major components Taxonomy of Geographic Routing Algorithms n Also called position-based routing n Three major components of geographic routing: v Location services (dissemination of location information) ØNext topic v Forwarding strategies v Recovery schemes 51

Forwarding Strategies n Basic greedy methods n Directional flooding n Geographical source routing n Forwarding Strategies n Basic greedy methods n Directional flooding n Geographical source routing n Power-aware routing 52

Basic greedy methods n Most Forward within Radius (C), 1984 n Nearest Forward Progress Basic greedy methods n Most Forward within Radius (C), 1984 n Nearest Forward Progress (A), 1986 n Compass Routing (B) , 1999 n Random Progress (X), 1984 n The above schemes’ 2 -hop variants 53

Directional Flooding n DREAM (in data packet routing) n LAR (in route discovery) n Directional Flooding n DREAM (in data packet routing) n LAR (in route discovery) n GRID (in route discovery) 54

Geographical Source Routing n Source specifies a geographical path v Needs an anchor path Geographical Source Routing n Source specifies a geographical path v Needs an anchor path discovery protocol Terminode routing n GRID n 55

Terminode Routing n “Self Organized Terminode Routing, ” Blazevic, Giordano, Le Boudec Cluster Computing Terminode Routing n “Self Organized Terminode Routing, ” Blazevic, Giordano, Le Boudec Cluster Computing Journal, Vol. 5, No. 2, April 2002 n Remote destinations: v Use geographical routing n Local destinations: v Use non-geographical, proactive routing n Similar to Zone Routing in this sense 56

Terminode Routing n Remote Routing v Anchored Geodesic Packet Forwarding v Geodesic Packet Forwarding Terminode Routing n Remote Routing v Anchored Geodesic Packet Forwarding v Geodesic Packet Forwarding (if no anchored path known) v Friend Assisted Path Discovery ØBased on Small World Graphs 57

Small World Graphs n Two nodes are connected if they are acquainted n Sparse, Small World Graphs n Two nodes are connected if they are acquainted n Sparse, small diameter 58

Terminode routing 59 Terminode routing 59

Power-Aware Routing n “Geographical and Energy Aware Routing: a recursive data dissemination protocol for Power-Aware Routing n “Geographical and Energy Aware Routing: a recursive data dissemination protocol for wireless sensor networks” n Y. Yu, R. Govindan, D. Estrin n UCLA 60

Recovery Schemes With any of the above forwarding strategies, packets may get stuck (hitting Recovery Schemes With any of the above forwarding strategies, packets may get stuck (hitting a hole). n A recovery scheme is invoked to get around the hole. n v Initiate a route discovery v GPSR (enter the perimeter mode) D S Stuck, initiating a recovery procedure 61

GPSR n “GPSR: Greedy Perimeter Stateless Routing for Wireless Networks” n Brad Karp, H. GPSR n “GPSR: Greedy Perimeter Stateless Routing for Wireless Networks” n Brad Karp, H. T. Kung n Harvard University n Mobi. Com 2000 n Two modes: v Greedy (for regular forwarding) v Perimeter (for recovery) 62

Perimeter Mode of GPSR n Suppose nodes x and D are connected by a Perimeter Mode of GPSR n Suppose nodes x and D are connected by a planar graph. n The graph divides the plane into faces. n Line x. D crosses one or more faces. D x 63

Planar Graphs n Graphs without crossing edges. Not Planar 64 Planar Graphs n Graphs without crossing edges. Not Planar 64

Planar Subgraph n G: communication graph n Relative neighborhood graph (RNG): v Subgraph of Planar Subgraph n G: communication graph n Relative neighborhood graph (RNG): v Subgraph of G v Keep edge (u, v) iff there are no nodes in the overlapped area. n RNG is planar u v 65

Evolution n Distance Vector, Link State n Proactive n On demand n Hybrid (zone Evolution n Distance Vector, Link State n Proactive n On demand n Hybrid (zone routing) n Geographical routing v Location Service v Location-based Forwarding v Recovery 66

Next? n Location service n Geographical routing without location services n Geocasting: v sending Next? n Location service n Geographical routing without location services n Geocasting: v sending a message to every node within a region. Geocast region Geocast group 67