1 / 39

Overview

SHORT: Self-Healing and Optimizing Routing Techniques for Ad Hoc Networks Chao Gui and Prasant Mohapatra University of California. Overview. Background SHORT Introduction Shortest Path Routing Shortcutting Established Routes PA-SHORT Algorithm, Limitations, Implementation

laith-bates
Télécharger la présentation

Overview

An Image/Link below is provided (as is) to download presentation Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author. Content is provided to you AS IS for your information and personal use only. Download presentation by click this link. While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server. During download, if you can't get a presentation, the file might be deleted by the publisher.

E N D

Presentation Transcript


  1. SHORT: Self-Healing and Optimizing Routing Techniques for Ad Hoc NetworksChao Gui and Prasant MohapatraUniversity of California

  2. Overview • Background • SHORT Introduction • Shortest Path Routing • Shortcutting Established Routes • PA-SHORT • Algorithm, Limitations, Implementation • Energy Aware Routing • EA-SHORT • Performance • Conclusion

  3. Background • Mobile Ad-hoc Network Routing Challenges: • Dynamic network topology • Low transmission power • Low available bandwidth • Reactive routing protocols lack knowledge about network topology and node movement • Routes may result in less than optimal paths • Path Length • Energy Consumption • Unbalanced Path Loads

  4. Background • Suboptimal routing • Increases end-to-end delay • Accelerates energy consumption of nodes • Decreases bandwidth • Reduces lifetime of routes • Routing protocols must adjust to a changing mobile network without: • Incurring too much control message overhead • Consuming too much energy from nodes in the network

  5. Introducing SHORT • SHORT can optimize any routing algorithm that provides a reachable path between source and destination. • SHORT includes two classes of algorithms • Path-Aware (PA-SHORT) • Energy-Aware (EA-SHORT) • Nodes neighboring a routing path monitor the route and try to optimize sub-paths: • Reducing hops • Reducing overall energy consumption • Preserving power of low-energy nodes • Goal: Optimize performance and energy conservation without incurring significant overhead.

  6. Shortest Path Based Routing • Mobile nodes cause network topology to change in unpredictable ways. • Most routing techniques perform initial path discovery but do not initiate new path discovery process until a link fails. • Some routing optimization techniques actually cause non-optimality in large networks • ER (Expanding Ring) Search • Local Repair • Simulations on large (10,000 node) networks show average path lengths increasing to more than double the optimal length with these techniques

  7. Shortest Path Degradation Most optimal path:8 hop path from A to I A G C E F H B I D J

  8. Shortest Path Degradation C No longer optimal I B H D F A E J G Most optimal path:5 hop path from A to I

  9. Shortcutting Routing Paths • In a given routing path, an (n,k) reduction implies that n routing hops can be reduced to k routing hops, where k < n. • Hops are edges of the network graph. • Intuitively, the higher the difference between n and k, the greater the performance gain will be. Original Path A-B-C A-B-C-D Reduced Path A-C A-E-D Reduction (2,1) (3,2)

  10. Shortcut Examples Gui and Mohaptra

  11. Frequency of Shortcut Scenarios • To provide significant performance gain, the shortcut scenarios must occur frequently. • Simulations were performed to determine the probability of adjacency of nodes. • Given a path v0, v1, … vn+1 of n+2 nodes in the simulated network, what is the probability that nodes v0 and vn+1 are adjacent? • We denote this probability by P[n], and term it the probability of adjacency. • A (2,1) shortcut is equivalent to P[1]. • A (3,2) shortcut is equivalent to P[3].

  12. Frequency of Shortcut Scenarios • P[1]is shown to derive as 58.6% • For greater values of n, computer simulations were performed repeatedly. • Simulation assumption: For the given set of n + 1 nodes along the n step path, the position of the next-hop node is uniformly distributed within the unit circle. • In a mobile ad-hoc network where routing paths are setup by a routing protocol, it is more probable that the next-hop node is toward the direction of the destination node.

  13. Frequency of Shortcut Scenarios • Further experiments were performed using the ns-2 simulator. • 100 node network simulated within a field of 2000m x 600m with random waypoint mobility. • Traffic simulated between source node S and destination node D using AODV for routing. • To ensure significant distance between S and D, position and movement of these nodes is restricted.

  14. Frequency of Shortcut Scenarios • More frequent mobility and faster mobility within denser networks results in highest shortcut availability. • About 40% of packets encountered (n,2) shortcuts. • About 10% of packets encountered (n,1) shortcuts. • Thus, a routing scheme that utilizes both shortcuts will converge to a shortest path much faster than one that uses only (n,1) shortcuts. Gui and Mohaptra

  15. Path Aware (PA)-SHORT • PA-SHORT monitors an established routing path, identifies shortcuts, and shortens the path according to those shortcuts. • Two subclasses of PA-SHORT • PA-SHORT-DV works with distance vector algorithms such as AODV. • PA-SHORT-SR works with source routing protocols such as DSR.

  16. PA-SHORT: Basic Idea • Packet forwarding is broadcast wirelessly, so all nodes within range can hear the packet. • In a typical wireless broadcast: • Each node checks the packet header to see if they are the next hop for the packet. • The node that is the next hop captures the packet and processes it for the next hop. • Other nodes generally disregard the packet. • So neighboring nodes get to check all packets that they can overhear.

  17. PA-SHORT: Header Information • To support PA-SHORT each packet header must carry the following field: • Hop Count field: HC • HC is initially zero at the source node and is incremented by one at every hop. • Other important information in the header: • Source address: SA • Destination address: DA

  18. PA-SHORT: Hop Comparison Array • Hop Comparison Array • Stored at each node, consists of: • Source Address: SA • Destination Address: DA • Hop Count: HC • Neighbor Address: NA • NA is the neighbor address from which the overheard packet was broadcast • Prior to the first transmission, the HCA of node SAk contains <SAk, DAk, 0, SAk >

  19. PA-SHORT: Algorithm Process the packet’s final destination. Write to HCA if the packet is a new flow. Process the packet’snext hop. Search for shortcuts. “Have I seen this same packet at least 2 hops ago?” Modify routing tables and clean up the hop comparison array. Gui and Mohaptra

  20. PA-SHORT: Example C <A to I, HC=2> B I <A to I, HC=5> <A to I, HC=1> H D <A to I, HC=7> <A to I, HC=3> F A <A to I, HC=6> <A to I, HC=0> E J G <A to I, HC=4> Suppose node A wants to send a message to node I. Best viewed in slideshow mode.

  21. PA-SHORT: Overhead • Overhead is minimal • Each node requires only a small amount of space to store the hop comparison array. • Only a small amount of processing required to detect shortcuts. • Only one extra message is necessary to inform a neighboring node to update its routing table.

  22. Limitation: Source (2,1) Shortcut • A (2,1) shortcut involving the source node will not be detected. Gui and Mohaptra

  23. Limitation: Destination Shortcut • Shortcuts involving the destination node will not be detected. Gui and Mohaptra

  24. PA-SHORT: Implementation • Ephemeral Shortcuts • When a fast-moving node forms a shortcut, and causes routing table changes, it may only remain in the effective shortcut position range for a short period of time. • Avoidance requires position and speed data. • Multiple Shortcuts • More than one node lies within the position range of a certain shortcut, and all report the shortcut to the relevant nodes, causing multiple route changes. • Can be avoided by using a stable period in which new or updated routing entries cannot change.

  25. Energy Aware Routing • Mobile devices typically have limited power. • Wireless network interfaces are a major source of energy consumption. • Energy aware routing aims at continuously monitoring changing conditions in link quality, contention rate, and tries to heal and optimize the path by diverting it to a better path, when one is available. • Route packets through nodes with sufficient power supplies, avoid nodes that are low on power, and balance traffic.

  26. Energy Aware Load Balancing • A heavily-used routing path will leave other nodes relatively free from traffic load. • Energy of nodes on this path will drain early. • An energy-aware routing scheme will divert traffic to an alternate path. • EA-SHORT: Basic Idea • Listen to packets broadcast from neighbor nodes. • Monitor energy levels of neighboring nodes. • Nodes with higher energy levels volunteer to forward packets for nodes with lower energy levels.

  27. Alternative Sub-Paths • In a dense network, alternate sub-paths of optimal path length may occur frequently. • Utilization of these sub-paths will help balance out the energy levels of nodes in the network. Gui and Mohaptra

  28. EA-SHORT: Packet Information • To support EA-SHORT, we add two fields to each packet, P. • Hop Counter: hc(P) • Residual energy level: lvl(P) • Other packet information of interest: • Source: s(P) • Destination: d(P) • Sender: nid(P) • Sequence Number: seq(P) • Source/Dest. Pair: s-d(P)

  29. EA-SHORT: Eavesdropping • Each node maintains an “overhear table” • Each entry represents a traffic flow in the network, identified by the Source/Destination pair field. • An overhear table entry is represented by e. • Source/Destination Pair: s-d(e) • Sequence Number: seq(e) • Overhear List: ovlist(e) • Overhear List Entry: hc, lvl, nid

  30. EA-SHORT: Algorithm Check Overheard Packetsand Add If From a New Flow For Familiar Flows:Ignore Earlier Sequences,Reset on Later Sequences, Record Matching Sequences Identify a 2-Step Alternate Sub-path and Activate New Sub-path Gui and Mohaptra

  31. EA-SHORT: Algorithm Identify Upstream Node of a 3-Step Alternate Sub-path Identify Downstream Node of a 3-Step Alternate Sub-path Listen for Cooperating 3-Step Path Nodes and Activate New Sub-path Gui and Mohaptra

  32. EA-SHORT: Example Consider a high-traffic path A-B-C-D. 5 3 4 8 6 7 3 5 4 7 8 6 3 7 6 8 4 5 6 7 3 4 5 8 A B C D E F G 3 4 5 6 7 8 7 3 4 8 6 5 3 4 5 6 7 8 Let’s enable the EA-SHORT algorithm. Best viewed in slideshow mode.

  33. EA-SHORT: Example <S to D, seq=2, hc=1, lvl=6> <S to D, seq=2, hc=2, lvl=6> <S to D, seq=2, hc=3, lvl=6> 3 4 5 6 5 6 3 4 3 4 5 6 6 7 8 9 A B C D E F G 3 4 5 6 7 8 8 7 6 4 3 5 3 4 5 6 7 8 Waiting: B Waiting: C SHORT: B Best viewed in slideshow mode.

  34. EA-SHORT: Example <S to T, seq=4, hc=1, lvl=4> <S to T, seq=4, hc=4, lvl=4> 1 2 3 4 2 3 4 5 1 4 2 3 5 6 7 8 A B C D <S to T, seq=4, hc=2, lvl=7> <S to T, seq=4, hc=3, lvl=7> <S to T, seq=4, hc=5, lvl=8> E F G 3 3 4 5 6 7 3 7 6 4 5 2 6 8 3 4 7 5 Alternate Path Found Waiting: C Best viewed in slideshow mode.

  35. PA-SHORT: Performance • AODV and DSR path optimality suffer as mobility increases. • SHORT algorithm enhances path optimality significantly. • AODV and DSR delivery rate suffer as mobility increases. • SHORT algorithm enhances delivery rate significantly. Gui and Mohaptra

  36. PA-SHORT: Route Overhead • SHORT algorithms have less routing overhead and fewer route requests. • Performance gain of AODV-SHORT is greater than that DSR-SHORT in both measures. • Again, SHORT performance gains are most significant in high-mobility scenarios. Gui and Mohaptra

  37. EA-SHORT: Performance 5 sourcenodes 5 sourcenodes 5 sinknodes 5 sinknodes • Simulated network field designed to create high-traffic nodes. • The network capacity timeline depicts a scenario with 200 seconds pause time, 10m/s movement and 6 packets per sec. • AODV interruptions begin at 1210 seconds and are data flow is not sustainable when resumed. High Traffic Nodes AODV Failure Zone Gui and Mohaptra

  38. Conclusion • SHORT improves routing optimality through: • Continuous monitoring of routing paths • Gradually redirecting toward a more optimal path • SHORT exploits available information about: • Relative local topology • Link quality • Simulations show that SHORT achieves: • Higher delivery rates • Lower routing overhead • Longer network lifetimes

  39. Slides prepared by James Bernsen

More Related