Routing and Topology Control in Ad Hoc & Wireless Sensor Networks

Transcription

1 Routing and Topology Control in Ad Hoc & Wireless Sensor Networks Marcin Brzozowski IHP Im Technologiepark Frankfurt (Oder) Germany All rights reserved

2 Overview 1. Basics of routing Network in computer science Necessity of routing in wireless networks Routing in OSI reference model Routing challenges in Wireless Sensor Networks Energy efficiency 2

3 Overview 1. Routing ideas and protocols Routing in Ad Hoc & Wireless Sensor Networks Historical classification Network structure Distance vector Link state Flat Hierarchical Backbone Clustering When updates exchanged Addressing Proactive Reactive Hybrid Identity-centric Data-centric Geographic 3

4 Network in computer science Network is a graph G = (V,E) V set of all nodes E set of all edges: (v 1,v 2 ) E V 2 e.g. V = { A, B, C,... } E = { (A,B), (B,C), (C,F),... } Routing Find a (the best) path from source to destination D A I H C G F M L K E Q P U R O J W T N V S WSN graph idealized model connectivity is function of time nodes mobility... B 4

5 Routing in wireless networks Motivation Broadcast wireless medium, routing necessary? Radio range limited, direct communication source-sink impossible......multi hop communication required A J B I C H D E F G Routing necessary Find (optimal) route to destination on source.....and on intermediates nodes (forwarding) 5

6 Routing in wireless networks Motivation J H I E A B C D F G Direct communication sometimes possible Multi hop communication (often) more efficient than a direct one TX energy ~ distance α (α >= 2) (too) many receivers (overhearing) -> MAC solution? Collisions, reduced throughput 6

7 Routing in OSI Reference Model data dest G next E Application Presentation Session Transport E dest G G data next F F G data A J B I C H D E F G E G data Network Network PHY+MAC E G data Data link (MAC) Physical Data link (MAC) Physical PHY+MAC F G data 7

8 Network in computer science Routing: main task of network layer Network layer Abstraction for higher layer: SEND PACKET RECEIVE PACKET Hide multi-hop communication Wireless Sensor Networks: OSI Model relaxed Data-centric routing Cross-layer (e.g. RSSI read by Link Layer, Routing and Application) 8

9 Routing in Wireless Sensor Networks Challenges nodes fail hardware/software failures (repairs inconvenient) hostile environment nodes move around wireless communication unreliable scalability: number of nodes from several to many thousands limited energy need of energy-aware routing -> Communication interruptions (temporary, permanent) 9

10 Routing in Wireless Sensor Networks Communication interruptions Temporary Permanent Redundancy Retries Several routes Divesion (Re-routing) 10

11 Energy-efficient Routing Goals Minimize energy/bit Example: A-B-E-H Maximize network lifetime (Time until first node failure, loss of coverage, partitioning) D 3 3 A 1 B C E H G 2 2 F 4 Source:Holger Karl, Andreas Willig, University of Paderborn 11

12 Energy-efficient Routing Maximum total available battery capacity Path metric: Sum of battery levels (A-C-F-H) Minimum battery cost routing Path metric: Sum of reciprocal battery levels (A-D-H) Min-Max battery capacity routing Path metric: largest reciprocal battery level use route with smallest metric Minimize variance in power levels Minimum total transmission power D A 1 B 1 E 1 H G C F 2 4 Source:Holger Karl, Andreas Willig, University of Paderborn 12

13 Taxonomy Routing in Ad Hoc & Wireless Sensor Networks Historical classification Network structure Distance vector Link state Flat Hierarchical Backbone Clustering When updates exchanged Addressing Proactive Reactive Hybrid Identity-centric Data-centric Geographic 13

14 Routing maintenance Routing: additional information required (e.g. neighbours, network topology,...) Routing information exchanged with other nodes What information? With which nodes exchanged? D A B I H C G F M L K E Q P U R O J W T N V S How to find route? 14

15 Distance Vector Routing Called Bellman-Ford (1957), Routing Information Protocol (RIP) in Internet each node maintains routing table (no topology graph) dest next distance C B 2 G B 2 B B 1 L J 3 nodes known distances (hops, delay...) to their neighbours periodically distance vectors sent to all neighbours Routing table determined from received distance vectors Source: A.S. Tanenbaum, Computer Networks, Pearson Education International,

16 Distance Vector Routing Fast good news propagation off on Distance to A = DV sent DV sent B gets DV from A C gets DV from B (B->A = 1 hop) Source: A.S. Tanenbaum, Computer Networks, Pearson Education International,

17 Distance Vector Routing Slow bad news propagation Slow bad news propagation off on Distance to A DV sent (no A s DV) DV sent (no A s DV) A B C D E C gets DV from B (B->A B = gets 3 hops) DV from C (C->A = 2 Count-to-infinity hops) problem Source: A.S. Tanenbaum, Computer Networks, Pearson Education International,

18 Count-to-infinity solution Destination Sequenced Distance-Vector Routing (DSDV) Count-to-infinity problem: use of sequence numbers (in routing tables and update packets) Source: Ch.Perkins, P.Bhagwat, Highly Dynamic Destination-Sequenced Distance-Vector Routing (DSDV) for Mobile Computers, ACM SIGCOMM'94 Conference on Communications Architectures, Protocols and Applications,

19 Link State Routing Maintain (store) network topology graph Calculate the best path locally (e.g. Dijkstra shortest path algorithm) Network graph maintenance: Discover neighbours (ID, link cost) Broadcast neighbour list (link state packets) - send to all nodes! On RX update network graph Source: A.S. Tanenbaum, Computer Networks, Pearson Education International,

20 Distance Vector vs Link State Routing Distance Vector Routing Link State Routing Send routing table to neighbours Send link state (neighbour list) to all nodes large overhead Suffers from count-to-infinity (long convergence) Better (faster) convergence than DV In ARPANET: Distance Vector Routing replaced by Link State Routing Routing updates necessary, but when? Periodically On event, what event? 20

21 Taxonomy Routing in Ad Hoc & Wireless Sensor Networks Historical classification Network structure Distance vector Link state Flat Hierarchical Backbone Clustering When updates exchanged Addressing Proactive Reactive Hybrid Identity-centric Data-centric Geographic 21

22 Proactive Protocols Routing structures (routing tables, network graph) maintained continually -> route updates sent regardless of data traffic Route updates sent even when no data traffic overhead Maintained continually: periodically on topology change (e.g. Node fails) Referred as table-driven 22

23 Destination Sequenced Distance-Vector (DSDV) Modified Distance-Vector protocol Dest Next Dist Seq I am next to I am 1 to 5 I am 1 to 5 I am 2 to Routing table at node 15 I am next to 5 10 I am 1 to 5 15 I am 2 to 5 2 I am 3 to 5 Source: Dr.-Ing. Falko Dressler, University of Erlangen-Nürnberg 23

24 Destination Sequenced Distance-Vector (DSDV) Improvements on Distance-Vector Aging mechanism against count-to-infinity problem / loops each route entry with sequence number originated by destination Optimization Full route updates sent periodically Incremental route updates (smaller) sent on topological changes Don t forward updates immediately, wait for settling time 24

25 Fisheye State Routing (FSR) Goal: reduce Link State protocol overhead Detailed knowledge about distant nodes necessary? Imprecise knowledge = routing overhead reduction Data moves closer to dest -> enters regions with detailed knowledge Source: G.Pei, M.Gerla, and Tsu-Wei Chen. Fisheye state routing: A routing scheme for ad hoc wireless networks. In Proceedings of the IEEE International Conference on Communications, pages 70 74, New Orleans, LA, June

26 Fisheye State Routing (FSR) Link State with limited flooding LS packets exchanged with neighbors (instead of flooding like in LS routing) Frequent TX update for near stations Rare TX update for stations afar Source: G.Pei, M.Gerla, and Tsu-Wei Chen. Fisheye state routing: A routing scheme for ad hoc wireless networks. In Proceedings of the IEEE International Conference on Communications, pages 70 74, New Orleans, LA, June

27 Optimized Link State Routing (OLSR) Link State with efficient flooding Multiple Relay Node (MPR) must be one-hop neighbour MPR set covers all 2-hop nodes Each node selects its own minimal MPR set (for flooding) Suited for dense networks Source: T. Clausen, P. Jacquet, A. Laouiti, P. Muhlethaler, A. Qayyum, and L. Viennot, Optimized link state routing protocol for ad hoc networks, in Proceedings of the IEEE International Multi Topic Conference, December

28 Reactive Protocols Idea: No data traffic = no route maintenance Find route just before data transmission No route information (next hop, distance,...) initially; Steps to be taken: Discover path to dest (routing packets flooded) Destinations unicast response Data transmission Discovered path used for future data transmissions Referred as on-demand 28

29 Ad Hoc On-Demand Distance Vector Routing (AODV) Source sends route request (RREQ) Intermediate nodes record (memory) where RREQ came from required for reverse path Destination sends back directed route reply (RREP) reverse path used Intermediate nodes activate route routing table Source: E.Royer and C-K. Toh. A review of current routing protocols for ad hoc mobile wireless networks. In IEEE Personal Communication, volume 6, pages 46-55, April

30 Ad Hoc On-Demand Distance Vector Routing (AODV) Link failure maintenance How to detect a failure? Source moves route discovery reinitiated Failure along route Upstream neighbor detects failure Link failure notification to upstream nodes Source gets link failure notification Route discovery reinitiated Local repair Source: E.Royer and C-K. Toh. A review of current routing protocols for ad hoc mobile wireless networks. In IEEE Personal Communication, volume 6, pages 46-55, April

31 Dynamic Source Routing (DSR) Source sends route request (RREQ) Intermediate nodes records add ID in RREQ and forward......or send back RREP when route to destination is known Destination sends back RREP Asymmetric links supported: Destination initiates route discovery to source Source: E.Royer and C-K. Toh. A review of current routing protocols for ad hoc mobile wireless networks. In IEEE Personal Communication, volume 6, pages 46-55, April

32 Dynamic Source Routing (DSR) Optimizations: Route caching (in both directions: to source and to destination) from forwarded packets overhearing RREPLY created from local caches RREQ with hop limit (e.g. expanding ring starting from 0 hops) 32

33 Dynamic Source Routing (DSR) Error recovery basic DSR recovery DATA DATA DATA A B C D E ERR ERR package salvaging use local cache DATA DATA DATA A B C D E DATA DATA DATA F G 33

34 Zone Routing Protocol (ZRP) Hybrid Protocol (Proactive and Reactive) Nodes maintains two zones Intra-zone: proactive Inter-zone: reactive Zone radius in hops Bordercast instead of broadcast (e.g. in RREQ) Behaviour adjustable Larger radius: more proactive Smaller radiusor more reactive Source: N.Beijar, Zone Routing Protocol, Networking Laboratory, Helsinky University of Technology 34

35 Proactive and Reactive Protocols Proactive Reactive Maintain routing structures (routing tables, network graph) continually periodically on event (topology change) Route updates sent even when no data traffic overhead Do nothing when no data traffic Find path shortly before data transmission 1. Discover path (routing packets) 2. Transmit data Referred as table-driven Referred as on-demand 35

36 Summary Proactive Reactive Advantages: Advantages: No delays before data transmission Alternative paths without delay QoS support Overhead proportional to data transmission mostly much smaller than proactive (what if nodes move frequently?) Drawbacks: Drawbacks: Large protocol overhead Delay before data transmission 36

37 Taxonomy Routing in Ad Hoc & Wireless Sensor Networks Historical classification Network structure Distance vector Link state Flat Hierarchical Backbone Clustering When updates exchanged Addressing Proactive Reactive Hybrid Identity-centric Data-centric Geographic 37

38 Geographic Routing Physical locations (e.g. GPS, time of arrival, time difference of arrival,...) No route maintance required Routing overhead decreased Forward decision locally (based on positions) 38

39 Greedy Routing Forward to node closer to destination Known positions own destination (in packet) last sender (in packet) neighbours (e.g. one-hop broadcasts) Strategies most forward within radius (MFR) nearest with forward progress (NFP) neighbor closest to straight line between sender and destination (compass routing) Source: M.Ilyas, The Handbook of Wireless Ad Hoc Wireless Networks, CRC Press,

40 Greedy Routing...not always possible, what then? Traversal of planar graph when greedy fails F local optimum, no neighbour closer to dest -> switch to graph traversal G closer to dest than F, resume greedy More sophisticated approaches: Greedy Perimeter Stateless Routing Face-2 Source: M.Ilyas, The Handbook of Wireless Ad Hoc Wireless Networks, CRC Press,

41 Geographic Addressing and Routing (GeoCast) messages sent to all nodes in given area (point, circle, polygon) instead of IDs (geocasting) multi-level GeoRouters each router calculates its service area source passes data to its GeoRouter GeoRouter checks whether dest in its service area if not, forwards to the higher-level GeoRouter Source: X.Hong, K.Xu, M.Gerla, Scalable routing protocols for mobile ad hoc networks, IEEE Network, Volume 16 Issue 4, pages 11-21, Jul/Aug

42 Identity-centric Directed towards a well-specified particular destination (sink) Support for unicast, multicast, and broadcast messages 7 TO:2 1 Data TO:2 Data 11 Source TO:2 Data 2 Sink Source: Dr.-Ing. Falko Dressler, University of Erlangen-Nürnberg 42

43 Data-centric Data is focus of attention Doesn t matter which node exactly (which ID) sends data (several nodes sense same data) Forwarding of messages to all / some appropriate nodes Routing decisions according to the data, i.e. encoding rules are needed ID: B Data ID: B ID: B Data Data Waiting for B B Source ID: B Data B Waiting for B Source: Dr.-Ing. Falko Dressler, University of Erlangen-Nürnberg 43

44 Identity-centric vs. Data-centric Routing approach Prerequisites Routing techniques Identity-centric Identification of a path according to the destination address of the data message Network-wide unique addresses Proactive routing (continuous state maintenance) or reactive routing (on-demand path finding) Data-centric Determination of the destination of a data message according to the content of the packet Pre-defined message types and semantics (probabilistic) flooding schemes or interest-based reverse routing Source: Dr.-Ing. Falko Dressler, University of Erlangen-Nürnberg 44

45 Data-dissemination Data dissemination forwarding of data though the network Data dissemination Flooding Gossiping Agent-based approaches Reverse path techniques Source: Dr.-Ing. Falko Dressler, University of Erlangen-Nürnberg 45

46 Flooding Basic mechanism: Each node that receives a packet re-broadcasts it to all neighbors The data packet is discarded when the maximum hop count is reached Step 1 Step 2 Step 3 Source: Dr.-Ing. Falko Dressler, University of Erlangen-Nürnberg 46

47 Flooding Advantages No route discovery No topology maintenance Delay minimal (source-sink) Disadvantages TTL 3 TTL 4 Implosion: duplicate messages are sent to the same node Overlap: same events sensed by several nodes -> duplicate reports of the same event Resource blindness: available energy not considered and redundant transmissions may occur -> limited network lifetime Source: Dr.-Ing. Falko Dressler, University of Erlangen-Nürnberg 47

48 Gossiping GOSSIP(p) Probabilistic version of flooding Packets are re-broadcasted with a gossiping probability p for each message m if random(0,1) < p then message m Step 1 Step 2 Step 3 p p p p p p p p p p p Source: Dr.-Ing. Falko Dressler, University of Erlangen-Nürnberg 48

49 Gossiping Advantages Avoids packet implosion Lower network overhead compared to flooding Disadvantages Long propagation time throughout the network Does not guarantee that all nodes of the network will receive the message (similarly do other protocols but for gossiping this is an inherent feature ) p p p p n-1 n p p 2 p (n-1) p n Source: Dr.-Ing. Falko Dressler, University of Erlangen-Nürnberg 49

50 Publish/subscribe paradigm Decoupling in space: publisher and subscribers not aware of each other (e.g. sink ID not known to source) in time: publishing and notification at different times (Event Service as intermediate storage) in flows: asynchronous (no blocking) Source: P.T. Eugster, P.A. Felber, R. Guerraoui, and A.M. Kermarrec. The many faces of publish/subscribe. ACM Computing Surveys, 35(2): , June

51 Directed Diffusion Realization of publish/subscribe paradigm Data dissemination based on reverse path Basic behavior Each sensor node names its data with one or more attributes Other nodes express their interest depending on these attributes The sink node has to periodically refresh its interest if it still requires data to be reported to it Data is propagated along the reverse path of the interest propagation Source: Dr.-Ing. Falko Dressler, University of Erlangen-Nürnberg 51

52 Directed Diffusion Named data Interest propagation type = four-legged animal interval = 1s rect = [-100, 200, 200, 400] timestamp = 01:20:40 expiresat = 01:30:40 Data transmission type = four-legged animal instance = elephant location = [125, 220] intensity = 0.6 confidence = 0.85 timestamp = 01:20:40 // type of animal seen // instance of this type // node location // signal amplitude measure // confidence in the match // event generation time Source: Dr.-Ing. Falko Dressler, University of Erlangen-Nürnberg 52

53 Directed Diffusion Steps: Interest propagation send named data to neighbours Convergast tree creation: gradients set up store who interest came from (no global unique id required) Data propagation send data according to gradients 53

54 Rumor Routing Agent-based path creation algorithm Motivation: Query: sink->sources I need data... Event: sources -> sinks I collected data... Few queries, many events query flooding (+ paths to sinks set up) Many queries, few events event flooding (+ gradients towards event set up) Rumor routing in between Source: Dr.-Ing. Falko Dressler, University of Erlangen-Nürnberg 54

55 Rumor Routing Agents, or ants are long-lived entities created at random by nodes These are basically packets which are circulated in the network to establish shortest paths to events that they encounter Event A Agent A Known path to A Agent B Known path to B Event B Source: Dr.-Ing. Falko Dressler, University of Erlangen-Nürnberg 55

56 Rumor Routing Optimizations: Propagation of several events Path shortening Y Event A 2 Distance Event A Event B Z Event Distance Direction A 43 XY B 1 X X Event Distance A 4 B 2 Source: Dr.-Ing. Falko Dressler, University of Erlangen-Nürnberg 56

57 In-Network Processing Lifetime...several years required limited battery capacity, energy harvesting in initial phase Communicaton consumes much more energy that CPU TMOTE Sky: Energy for sending of 1 byte = hundreds of CPU instructions (100-byte packet = tens thousand CPU instructions) Goal: Network traffic reduction 57

58 In-Network Processing Local data processing compression aggregation (but large packets -> higher PER Distributed data processing cooperation in-network processing 43 C 22 C 43 C 22 C 22 C 22 C 21.7 C 21 C 21 C General communication high network traffic tardy data error discovery Data aggregation reduced network traffic early error discovery 58

59 Data aggregation chain 2 chain 3 Chain-based aggregation Tree-based aggregation chain 1 sink cluster 2 chain 4 A A A A cluster 1 C C C cluster 3 sink Grid-based aggregation sink Source: Dr.-Ing. Falko Dressler, University of Erlangen-Nürnberg 59

60 Data-aggregation Tradeoff: latency vs efficiency Δt Δt Δt Δt n-1 n Δt 2 Δt (n-1) Δt n Δt Source: Dr.-Ing. Falko Dressler, University of Erlangen-Nürnberg 60

61 In-Network Processing data dest G next E Application Presentation Session Transport E dest next G F G data F G Content Layer data A J B I C H D E F G E G data Network Network PHY+MAC E G data Data link (MAC) Physical Data link (MAC) Physical PHY+MAC F G data 61

62 Summary Forwarding based on: Locations, Addresses or/and Data Geographic routing (locations): Routing overhead decreased (e.g. No route maintenance) Data-centric routing Focus of attention: data Often more suitable for WSN than pure id-centric: Flexibility (source-sink decoupling) In-network processing 62

63 Taxonomy Routing in Ad Hoc & Wireless Sensor Networks Historical classification Network structure Distance vector Link state Flat Hierarchical Backbone Clustering When updates exchanged Addressing Proactive Reactive Hybrid Identity-centric Data-centric Geographic 63

64 Topology control Motivation Tasks carried out by dozens, hundreds, thousands...of nodes Topology control Massive routing overhead Collisions Reduced throughput Choose Backbone, links clusterheads explicitly Control Number transmit of routing power nodes reduced 64

65 Taxonomy Topology control Flat network all nodes have essentially same role Hierarchical network assign different roles to nodes; exploit that to control node/link activity Power control Backbones Clustering 65

66 Connectivity Network characteristic preserved (e.g. connectivity)......can be overpriced Maximum component size 5000 Probability of connectivity 1 Average size of the largest component ,8 0,6 0,4 0,2 Probability of connectivity Maximum transmission range Source:Holger Karl, Andreas Willig, University of Paderborn 66

67 Metrics Connectivity If two nodes connected in G, they have to be connected in G0 resulting from topology control Stretch factor should be small Hop stretch factor: how much longer are paths in G0 than in G? Energy stretch factor: how much more energy does the most energyefficient path need? Throughput removing nodes/links can reduce throughput, by how much? Robustness to mobility Algorithm overhead Source:Holger Karl, Andreas Willig, University of Paderborn 67

68 Flat Network Power control Idea: Controlling transmission power corresponds to controlling the number of neighbors for a given node 68

69 Flat Network Power control (?) Basic idea: Take a graph G = (V,E), produce a graph G0 = (V,E 0 ) that maintains connectivity with fewer edges Assume, e.g., knowledge about node positions Construction should be local (for distributed implementation) Source:Holger Karl, Andreas Willig, University of Paderborn 69

70 Relative Neighborhood Graph (RNG) Edge between nodes u and v if and only if there is no other node w that is closer to either u or v (remove longest edge from triangle) Formally: RNG maintains connectivity of the original graph Easy to compute locally Power control: find closest neighbours and use minimal power to reach them (?) This region has to be empty for the two nodes to be connected Source:Holger Karl, Andreas Willig, University of Paderborn 70

71 Backbone Construct a backbone network Some nodes control their neighbors; they form (minimal) dominating set Each node should have controlling neighbor Controlling nodes have to be connected (backbone) Only links within backbone and from backbone to controlled neighbors are used Formally: Given graph G = (V,E), construct D V such that Source:Holger Karl, Andreas Willig, University of Paderborn 71

72 Backbone Idea: Select some nodes from the network/graph to form a backbone A connected, minimal, dominating set (MDS or MCDS) Dominating nodes control their neighbors Protocols like routing are confronted with a simple topology from a simple node, route to the backbone, routing in backbone is simple (few nodes) Problem: MDS is an NP-hard problem Hard to approximate, and even approximations need quite a few messages Source:Holger Karl, Andreas Willig, University of Paderborn 72

73 Backbone by growing a tree 1: 2: 3: 4: Source:Holger Karl, Andreas Willig, University of Paderborn 73

74 Backbone by growing a tree Problem: Which gray node to pick? (turn whitest node gray) u u u d d d Lookahead using nodes g and w g v d u v d u v Lookahead: what would happen if not only gray node turned black, but also its white neighbours turned black? v=w v Source:Holger Karl, Andreas Willig, University of Paderborn 74

75 Clustering Partition nodes into groups ( clusters ) Each node in exactly one group (Except for nodes bridging between two or more groups) Groups can have clusterheads Typically: all nodes in a cluster are direct neighbors of their clusterhead Clusterheads are also a dominating set, but should be separated from each other they form an independent set Formally: Given graph G = (V,E), construct C V such that Source:Holger Karl, Andreas Willig, University of Paderborn 75

76 Clustering Partition nodes into groups of nodes clusters Many options for details Are there clusterheads? One controller/representative node per cluster May clusterheads be neighbors? If no: clusterheads form an independent set C: Typically: clusterheads form a maximum independent set May clusters overlap? Do they have nodes in common? Source:Holger Karl, Andreas Willig, University of Paderborn 76

77 Clustering Further options How do clusters communicate? Gateways between clusters No cluster overlap -> distributed gateway How many gateways exist between clusters? Are all active, or some standby? What is the maximal diameter of a cluster? If more than 2, then clusterheads are not necessarily a maximum independent set Is there a hierarchy of clusters? Source:Holger Karl, Andreas Willig, University of Paderborn 77

78 Clustering Computing a maximum independent set is NP-complete Can be approximate within ( +3)/5 for small, within O( log log / log ) else; bounded degree Maximum independent set not necessarily intuitively desired solution Source:Holger Karl, Andreas Willig, University of Paderborn 78

79 Clustering Basic construction of independent sets Init: Use some attribute of nodes to break local symmetries Node identifiers, energy reserve, mobility, weighted combinations - matters not for the idea as such (all types of variations have been looked at) Step 1: Step 2: Make each node a clusterhead that locally has the largest attribute value Black: cluster Gray: cluster member White: undecided Step 3: Step 4: Source:Holger Karl, Andreas Willig, University of Paderborn 79

80 Clustering Rotating clusterheads Serving as a clusterhead can put additional burdens on a node For MAC coordination, data forwarding... Let this duty rotate among various members Periodically reelect useful when energy reserves are used as discriminating attribute LEACH determine an optimal percentage P of nodes to become clusterheads in a network Use 1/P rounds to form a period In each round, np nodes are elected as clusterheads At beginning of round r, node that has not served as clusterhead in this period becomes clusterhead with probability P/(1-p(r mod 1/P)) Source:Holger Karl, Andreas Willig, University of Paderborn 80

81 Adaptive node activity Remaining option: Turn some nodes off deliberately Only possible if other nodes remain on that can take over their duties Example duty: Packet forwarding Approach: Geographic Adaptive Fidelity (GAF) Observation: Any two nodes within a square of length r < R/5 1/2 can replace each other with respect to forwarding R radio range Keep only one such node active, let the other sleep r R r Source:Holger Karl, Andreas Willig, University of Paderborn 81

82 Summary Large networks: Massive routing overhead Collisions...Topology control required Flat networks (power control) Hierarchy (backbone, clustering) Adaptive node activity 82

83 Routing in Ad Hoc & WSN Summary Internet routing (Distance-Vector, Link State) not directly usable in Ad Hoc & WSN No routing one size fits all, very application specific Data-centric: efficient means to energy efficiency Topology control essential in large networks 83

84 Summary Thank you for your attention! 84