1 / 45

Off By One Power-Save Protocols

Off By One Power-Save Protocols. Corey Andalora Keith Needels. Agenda. Paper 1: BECA / AFECA Paper 2: GAF Paper 3: Span Framework Design. Power Save Protocols. Our topic for this project is power save protocols.

kapono
Download Presentation

Off By One Power-Save Protocols

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. Off By OnePower-Save Protocols Corey Andalora Keith Needels

  2. Agenda • Paper 1: BECA / AFECA • Paper 2: GAF • Paper 3: Span • Framework Design

  3. Power Save Protocols • Our topic for this project is power save protocols. • Power save protocols save energy by keeping node radios off as much as possible. • Why completely turn radios off? • Radio power usage is dominated by the idle state!

  4. Power Save Protocols • A node sending data at 2 Mbps and receiving data at 2 Mbps from each of four neighbors uses only 1% more power than an idle interface. [4] • So by reducing the amount of transmissions, reducing the transmission radius, reducing packet size, etc. you can only hope to save a fraction of 1% of energy usage.

  5. Protocol Energy Consumption [2]

  6. Adaptive Energy-Conserving Routing for Multihop Ad Hoc Networks Ya Xu, John Heidemann, and Deborah Estrin. “Adaptive energy-conserving routing for multi-hop ad hoc networks.” Technical Report 527, USC/Information Sciences Institute, October 2000.

  7. BECA/AFECA • This paper was an early power save protocol. • Two algorithms are presented: • BECA: Basic Energy-Conserving Algorithm • AFECA: Adaptive Fidelity Energy-Conserving Algorithm

  8. BECA • Each node wakes up every Ts seconds and listens for traffic for Tl seconds. • If no traffic for this node is received within Tl seconds, go back to sleep for another Ts seconds. • If traffic is received, go into the active state. • Transition back into the sleep state when no traffic is received for Ta seconds.

  9. BECA Node States

  10. Route Setup in BECA • All nodes are initially asleep when A decides to send data to B. B C D E A

  11. Route Setup in BECA • A continues to try to send RREQs to C. Within Ts seconds, C enters listening state and receives the RREQ, and C then enters the active state. B C D E A

  12. Route Setup in BECA • C continues to try to send the RREQ to B, D, and E, until within Ts seconds they receive it and transition into the active state. • B replies with the RREP and starts communicating with A via C. B C D E A

  13. Route Setup in BECA • Since no traffic is being passed through D and E, within Ta seconds they will enter the sleeping state. Now only the nodes that need to be awake are awake. B C D E A

  14. BECA • This is a very simple power save protocol. • Should not be used with “a priori” routing algorithms like distance vector. • Can you see why? • Results show nodes save an average of 40-50% of their energy over plain AODV! From http://www432.pair.com/linton/hugg/gorecan.jpg

  15. AFECA • Everything is the same as BECA, except node sleep time varies with network density. • When a node overhears a neighbor, it adds the neighbor to a neighbor set. After some time (Te) without overhearing this neighbor, the neighbor is removed from the neighbor set. • Let N be the number of nodes in the neighbor set.

  16. AFECA Sleep Time • AFECA sleep time is denoted by TSA. • Tsa = Random(1,N) x Ts • What do we gain by sleeping for Tsa instead of Ts? • In dense areas, we should be able to sleep longer (on average) since there are more nodes capable of routing traffic.

  17. AFECA Issues • AFECA is only intended for networks of uniform density. • The middle node below has a high number of neighbors, so he sleeps longer. This is bad!

  18. AFECA Results • AFECA saves 2-5% more energy over BECA • The big advantage of AFECA is network lifetime (time until all nodes die): • BECA extends network lifetime by 20% over AODV • AFECA extends network lifetime by 55% over AODV • A fourfold increase in AFECA node density doubles network lifetime.

  19. Geography-informed Energy Conservation for Ad Hoc Routing Ya Xu, John Heidemann, and Deborah Estrin. “Geography-informed energy conservation for ad hoc routing,” in Proceedings of 7th Annual International Conference on Mobile Computing and Networking, pp. 70-84, July 2001.

  20. GAF: Geographical Adaptive Fidelity • Main ideas: • Each node knows its location • Nodes are partitioned into grid squares, where any two nodes in adjacent grid squares are within range of each other. • At any given time, only one node in each grid square needs to be awake to route data. • Active grid square nodes are cycled.

  21. Adjacent Grids • Any node in the yellow grid squares are within range of all nodes in the red grid square.

  22. Grid Square Size • All nodes (red circles) are within range of each other. If R is our radio range, how big can our grid squares be?

  23. Grid Square Size • Our grid squares have to be small enough for these two nodes (worst case) to reach each other. Let R be our node’s radio range and x be the grid square’s width. Solve for x! x x x x x R x x

  24. Grid Square Size – Easy Math • From high school: a2+b2=c2 • (x)2+(2x)2=R2 x x x x x R x x

  25. Grid Square Size – Easy Math • From high school: a2+b2=c2 • (x)2+(2x)2=R2

  26. Active Nodes • Here is an example ad-hoc network broken up into grids. Only the red nodes are on, the yellow nodes can sleep.

  27. Active Nodes • After a while, some of the inactive nodes will take over the role of the active node when the current active node’s power level drops below theirs.

  28. Active Nodes • In this image, three inactive nodes have taken over the role of active node for their grid square.

  29. GAF Node States • Sleeping: Radio is in sleep mode. • After Ts seconds, node enters discovery state. • Discovery: Nodes are in the process of selecting this grid square’s active node. • After Td seconds with no other higher ranking node becoming active, node enters active state. • If a higher ranked node sends a discovery message, this node re-enters the sleep state. • Active: This node is acting as a grid square’s active node. • If a higher ranked node sends a discovery message, this node re-enters sleep state. • After Ta seconds, this node re-enters discovery state.

  30. GAF Node States

  31. Paper Results • GAF saved each node an average of 40-60% of its energy over bare AODV. • Varies between 40-60% based on simulated node movement patterns. • GAF was able to triple to quadruple network lifetime over AODV. • Higher node density increases network lifetime. • Practically no increase in latency or decrease in packet loss if GAF nodes are transit only.

  32. GAF Wrap-Up • Parameters are customizable (Ts, Td, Ta, etc.) • Ranking is customizable (usually, your amount of remaining power is your rank.) • Like Span, GAF runs independently of the routing protocol. • Since the active node might move out of its grid square before the other nodes wake up, active nodes can advertise the time they expect to leave the grid square. • Any questions on GAF?

  33. Span: An Energy-Efficient Coordination Algorithm for TopologyMaintenance in Ad Hoc Wireless Networks Benjie Chen, Kyle Jamieson, Hari Balakrishnan, and Robert Morris. “Span: An energy-efficient coordination algorithm for topology maintenance in ad hoc wireless networks.” ACM Wireless Networks Journal, 8(5), 481-494, September 2002.

  34. Span Requirements • Allow as many nodes as possible to turn their radio receivers off most of the time. • Forward packets between any source and destination with minimally more delay than if all nodes were awake. • Provide about as much total capacity as the original network, since otherwise congestion may increase. • Don’t make assumptions about link layer’s facilities for sleeping. • Inter-operate with whatever routing system the ad hoc network uses.

  35. Span Approach • Distributed backbone selection algorithm. • Nodes periodically decide whether to sleep or stay awake. • Nodes announce coordinator willingness by a random time interval two factors: • Battery level • Neighbor count • Nodes switch state periodically between being a coordinator and being a non-coordinator.

  36. Span Achieved Goals • Elect enough coordinators so that every node is in radio range of at least one coordinator. • Rotate the coordinators in order to ensure that all nodes share the task of providing global connectivity roughly equally. • Minimize the number of nodes elected as coordinators. • Increases network lifetime • No significant loss of capacity or increase in latency • Elect coordinators using only local information in a decentralized manner.

  37. Coordinator Eligibility Rule • A non-coordinator node should become a coordinator if it discovers, using only information gathered from local broadcast messages, that two of its neighbors cannot reach each other either directly or via one or two coordinators. C N N

  38. Backoff Delay Ni = the number of neighbors for node i Ci = the number of added connections among neighbors if i were coordinator. Er = remaining energy of node Em = maximum energy of node R = random value between 0 and 1 T = round-trip delay of a packet • The likelihood of becoming a coordinator falls as a coordinator uses up battery. • A node that connects partitions together will always be elected a coordinator.

  39. Coordinator Withdrawal WT = 3 x Ni x T sleeping every pair of neighbors can reach each other through another coordinator tentative after delay after CT coordinator after WT

  40. Span HELLO Packets • Source ID • Node position • Is Coordinator • Is Tentative • Coordinator list • Neighbor list

  41. Span Scenario Any questions on Span?

  42. NodeRangeListener NodePowerListener +Point getPosition() +void nodeEnteredRange(AdHocNode node) +void nodeLeftRange(AdHocNode node) +nodeTurnedOn(AdHocNode node) +nodeTurnedOff(AdHocNode node) Packet AdHocNode -AdHocNode source -AdHocNode destination -Date startTime -int size -int nodeId -AdHocWorld world -Point position -Point movement -double batteryLevel -boolean isOn -Vector<AdHocNode> neighbors -Vector<NodePowerListener> listeners +void sendPacket(Packet packet) +void start() UML: AdHocNode

  43. PacketListener +void packetDelivered(Packet packet) +void packetLost(Packet packet) AdHocWorld -Dimension size -double transmitRange -double transferRate -double consumptionRateOn -double consumptionRateOff -double maxSpeed -Vector<AdHocNode> nodes -Graph<AdHocNode> graph +AdHocNode getNextNode(AdHocNode src, AdHocNode dest) +void start() UML: AdHocWorld

  44. UML: Algorithms AdHocNode SpanNode AFECANode GAFNode

  45. Questions? • Ya Xu, John Heidemann, and Deborah Estrin. “Adaptive energy-conserving routing for multi-hop ad hoc networks.” Technical Report 527, USC/Information Sciences Institute, October 2000. • Ya Xu, John Heidemann, and Deborah Estrin. “Geography-informed energy conservation for ad hoc routing,” in Proceedings of 7th Annual International Conference on Mobile Computing and Networking, pp. 70-84, July 2001. • Benjie Chen, Kyle Jamieson, Hari Balakrishnan, and Robert Morris. “Span: An energy-efficient coordination algorithm for topology maintenance in ad hoc wireless networks.” ACM Wireless Networks Journal, 8(5), 481-494, September 2002. • Stefano Basagni, Marco Conti, Silvia Giordano, and Ivan Stojmenovic. Mobile Ad Hoc Networking. John Wiley & Sons, 2004. ISBN 0-471-37313-3.

More Related