Self Organizing and Self-Healing Ad-hoc Networks

Transcription

1 Self Organizing and Self-Healing d-hoc Networks WLN 57- by hip (rig) Elliott, N Technologies ambridge, M, Presented at WLN 2001 by Jerry urchfiel 1

2 Example Problem N radios on a terrain Each moving No Fixed Infrastructure High-Speed ata ommunication Mobile Voice and Video heap and Reliable 2

3 The Hidden Terminal Problem and can t hear each other transmits to, while transmits to Results in one or both transmissions failing Very common event, alas! Their transmissions collide at and, and both packets are lost How could or know when to transmit? 3

4 hannel ccess Mechanisms RTS / TS pproach Timeslot pproach RTS TS ata Frame K Transmitter directed Good features of TM Needs sync d clocks Very hard distributed algorithm Receiver directed cues off s TS omplex state machines ehaves like SM (oesn t work for broadcast) x RTS TS RTS TS Note that and must be coordinated! 4

5 elay vs. Throughput Tradeoff High Power means low network throughput Low Power means long end-to-end delay Single, High-Power Transmission Multi-Hop, Low Power Transmissions 38 other radios blocked 14 other radios blocked Interference rea 5

6 Proactive Routing Protocols E F G E G F Link State (SPF) istance Vector Global Knowledge llows QOS Routing Multicast is easy ad Points Scales poorly Global bursts of control traffic E F G ctual Network Topology Less ontrol Traffic Local Repair Simple to ode ad Points ontrol traffic is hard QOS is hard to estimate Multicast is hard 6

7 On-emand Routing Protocols? 1. Flood Search. (Optimized by caching.) 2. Path iscovery. (Optimized by pruning.) 3. Traffic. (Source routed.) No global knowledge llows local repair Scales with Traffic Flows ad Points Repair routes are suboptimal ad interactions with TP Multicast is complicated enial of Service attacks 7

8 Hierarchical d Hoc Network luster Member luster luster Head RTS / TS Waveform Per-frame Power ontrol ackbone Link Hierarchical Link-State Routing Spatial reuse with good end-to-end delay 8

9 d Hoc Network with Unmanned ircraft GS, LEOs rone Nodes TM Radios High Mobility ackbone Handheld Voice/ata Nodes 9

10 Example Problem Hidden Terminal Problem and can t hear each other transmits to, while transmits to Results in one or both transmissions failing Very common event, alas! N radios on a terrain Each moving No Fixed Infrastructure High-Speed ata ommunication Mobile Voice and Video heap and Reliable Their transmissions collide at and, and both packets are lost How could or know when to transmit? hannel ccess Mechanisms elay versus Throughput RTS / TS pproach RTS ata Frame TS K Timeslot pproach Transmitter directed Good features of TM Needs sync d clocks Very hard distributed algorithm High Power means low network throughput Low Power means long end-to-end delay Single, High-Power Transmission Multi-Hop, Low Power Transmissions Receiver directed cues off s TS omplex state machines ehaves like SM (oesn t work for broadcast) RTS xts RTS TS Note that and must be coordinated! 38 other radios blocked 14 other radios blocked Interference rea 10

11 Proactive Routing Protocols On-emand Routing Protocols E F G E G F? 1. Flood Search. (Optimized by caching.) Link State (SPF) istance Vector 2. Path iscovery. (Optimized by pruning.) Global Knowledge llows QOS Routing Multicast is easy ad Points Scales poorly Global bursts of control traffic E F G ctual Network Topology Less ontrol Traffic Local Repair Simple to ode ad Points ontrol traffic is hard QOS is hard to estimate Multicast is hard 3. Traffic. (Source routed.) No global knowledge llows local repair Scales with Traffic Flows ad Points Repair routes are suboptimal ad interactions with TP Multicast is complicated enial of Service attacks Hierarchical d Hoc Network luster Member d Hoc Network with rone ircraft GS, LEOs rone Nodes luster Head luster TM Radios High Mobility ackbone RTS / TS Waveform Per-frame Power ontrol ackbone Link Hierarchical Link-State Routing Spatial reuse with good end-to-end delay Handheld Voice/ata Nodes 11