1. Wave Relay: Multi-hop Wireless Ad hoc Network Baruch Awerbuch, David Holmer, Herbert Rubens {baruch dholmer herb}@cs.jhu.edu Johns Hopkins University Department of Computer Science www.cnds.jhu.edu/archipelago/

2. Goals: Design a system that… Supports a large number of nodes Supports a large number of nodes thousands thousands Moving at high speeds Moving at high speeds greater then 40 mph greater then 40 mph In an urban environment In an urban environment High multi-path, rapidly fluctuating channels High multi-path, rapidly fluctuating channels Running real-time applications Running real-time applications Voice, video, interactive distributed applications Voice, video, interactive distributed applications With or without help from fixed infrastructure With or without help from fixed infrastructure If its available use it to be more efficient If its available use it to be more efficient

3. Wave Relay Test-bed Over 50 Wave Relay Routers deployed across JHU Campus Over 50 Wave Relay Routers deployed across JHU Campus Urban City Environment Urban City Environment Internet Access, Ad hoc Access Points, Voice over IP Internet Access, Ad hoc Access Points, Voice over IP Mobility testing from automobiles Mobility testing from automobiles System tested at Holcim Industrial Plant (Chicago, IL) System tested at Holcim Industrial Plant (Chicago, IL) Complex propagation environment Complex propagation environment Massive multi-path Massive multi-path Enabled real-time industrial process control Enabled real-time industrial process control Currently Deployed Custom Applications Currently Deployed Custom Applications Military Distributed Battlefield Mapping Military Distributed Battlefield Mapping GPS based interactive map GPS based interactive map Eventual reliability Eventual reliability Locality Specific Messaging System Locality Specific Messaging System GPS based messaging system GPS based messaging system Messages targeted to any user at a specific location Messages targeted to any user at a specific location

4. Wave Relay Device Pulse Protocol [Infocom’04, Milcom’04, WONS’05] Pulse Protocol [Infocom’04, Milcom’04, WONS’05] Scalable ad hoc routing protocol Scalable ad hoc routing protocol Active path tracking Active path tracking Based on Tree Routing strategy Based on Tree Routing strategy Medium Time Metric [MONET,WONS’04] Medium Time Metric [MONET,WONS’04] High Throughput Path Selection High Throughput Path Selection Increased Path Elasticity Increased Path Elasticity Efficient Multi-rate Operation Efficient Multi-rate Operation Leader Election Algorithm Leader Election Algorithm Handles merge, partition, failure Handles merge, partition, failure Embedded Linux Distribution Embedded Linux Distribution Less then 8 MB storage requirement Less then 8 MB storage requirement Linux Kernel Module 2.4 and 2.6 compatibility Linux Kernel Module 2.4 and 2.6 compatibility Operates at layer 2 Operates at layer 2 Distributed virtual switch architecture provides seamless bridging Distributed virtual switch architecture provides seamless bridging Embedded Single Board Computer Embedded Single Board Computer NS Geode SC1100 266 MHz Processor 64 Mb Ram onboard 2 mini-PCI interfaces 1 Compact flash interface Serial port 10/100 Ethernet Hardware Watchdog Power over Ethernet +7V to +18V DC Input Atheros 802.11g/b Wireless Card Atheros 802.11g/b Wireless Card 400 mW (26 dBm) output power 16 MB Industrial Compact Flash 16 MB Industrial Compact Flash Stores OS & Wave Relay software Garmin GPS 16 receiver Garmin GPS 16 receiver Li-Ion Battery Pack Li-Ion Battery Pack ~20 hours continuous runtime Industrial NEMA 67 Enclosure Industrial NEMA 67 Enclosure 4 N-type antenna mounts 2 Ethernet Ports (6) protection against dust (7) water submersible Software Hardware

5. Existing Approaches Source Receivers Destination Multi-path fading & shadowing Rapidly changing channel conditions On-demand protocols have no knowledge of channels conditions A RREQ packet provides only a single sample of a complex distribution On-Demand Protocols (AODV, DSR) Channel is continuously changing Continuous flooding from every node in the network Link State Protocols (OLSR, TBRPF) Urban Channel Environment You can not accurately track channel with control packets!

6. The Pulse Protocol Proactive Component Proactive Component Tracks minimum amount of information to avoid flooding for route establishment and maintenance Tracks minimum amount of information to avoid flooding for route establishment and maintenance Periodic flood operation (similar to Hello Protocol) Periodic flood operation (similar to Hello Protocol) Rebuilds spanning tree Rebuilds spanning tree Estimates neighbors, density, SNR, loss rates, capabilities, number of radios, MTM metric Estimates neighbors, density, SNR, loss rates, capabilities, number of radios, MTM metric On-Demand Component On-Demand Component Route establishment Route establishment Using only UNICASTS! Using only UNICASTS! Gratuitous mechanism Gratuitous mechanism Neighbors promiscuously monitor packets Neighbors promiscuously monitor packets Metric tracked at the speed of data packets NOT control packets! Metric tracked at the speed of data packets NOT control packets! Path switches as metrics change Path switches as metrics change Local changes in connectivity only generate local traffic Local changes in connectivity only generate local traffic Unlike BOTH on-demand and link state protocols Unlike BOTH on-demand and link state protocols

7. Ad hoc Nodes

8. Network Connectivity

9. Pulse Flood

10. Spanning Tree

11. Source and Destination Need to Establish a Path

12. Pulse Response Sent to Root

13. Destination Paged on Next Pulse

14. Destination Sends Pulse Response

15. Initial Path: Tree Shortcut Tree Shortcut Path 3 Hops Shortest Path 2 Hops This is the initially selected path of the Pulse protocol.

16. Path Optimization: Gratuitous Reply Optimized Path 2 Hops Shortest Path 2 Hops Node sends gratuitous reply

17. Proactive Route Maintenance

20. Pulse Protocol Concepts Aggregation – for scalability Aggregation – for scalability Spanning tree represents a compressed view of the network topology Spanning tree represents a compressed view of the network topology Pro-active component maintains the minimum amount of information to allow efficient route establishment Pro-active component maintains the minimum amount of information to allow efficient route establishment De-Aggregation – for efficiency De-Aggregation – for efficiency The routing metric is tracked at the speed of the data flow The routing metric is tracked at the speed of the data flow Changes to the metric are only reported locally Changes to the metric are only reported locally Routes are continuously adjusted as the metrics change Routes are continuously adjusted as the metrics change High speed accurate route tracking is essentially an on-demand decompression of the topology High speed accurate route tracking is essentially an on-demand decompression of the topology However, it occurs ONLY in areas of the network with active data flows However, it occurs ONLY in areas of the network with active data flows Result: a scalable routing structure which tracks paths at the speed of the data flow Result: a scalable routing structure which tracks paths at the speed of the data flow

21. Future Work Security (NDSS 2005) Security (NDSS 2005) Wormholes, black-holes, flood rush, replay Wormholes, black-holes, flood rush, replay Provide Provide Node authentication Node authentication End-to-end encryption End-to-end encryption Broadcast/Routing Encryption Broadcast/Routing Encryption Efficient node addition/removal Efficient node addition/removal Distributed commit (CNDS-02) Distributed commit (CNDS-02) Consistent, persistent, group communication Consistent, persistent, group communication e.g. coordinated battlefield view and control e.g. coordinated battlefield view and control Opportunistic Gradient Forwarding Opportunistic Gradient Forwarding

22. Thank You! Questions?? http://www.cnds.jhu.edu/archipelago/ (baruch,dholmer,herb)@cs.jhu.edu Wave Relay Ad hoc Networking Test-bed http://www.cnds.jhu.edu/research/networks/archipelago/testbed/testbed.html Secure Ad hoc Networking for Industrial Process Control http://www.cnds.jhu.edu/research/networks/archipelago/industrial/industrial.html Baruch Awerbuch, David Holmer, Herbert Rubens

24. Minimum Hop Metric (Traditional Technique) Not designed for multi-rate networks Not designed for multi-rate networks A small number of long slow hops provide the minimum hop path A small number of long slow hops provide the minimum hop path These slow transmissions occupy the medium for long times, blocking adjacent senders These slow transmissions occupy the medium for long times, blocking adjacent senders Selecting nodes on the fringe of the communication range results in reduced reliability Selecting nodes on the fringe of the communication range results in reduced reliability

25. New Approach: Medium Time Metric (MTM) Assigns a weight to each link proportional to the amount of medium time consumed by transmitting a packet on the link Assigns a weight to each link proportional to the amount of medium time consumed by transmitting a packet on the link Enables the Pulse protocol to discover the path that minimizes total transmission time Enables the Pulse protocol to discover the path that minimizes total transmission time

26. MTM Example Source Destination 11 Mbps 5.5 Mbps 2 Mbps 1 Mbps 1 0.85 Mbps 2.5ms 3.7ms 7.6ms 13.9ms 11 Mbps 5.5 Mbps 2 Mbps 1 Mbps 13.9ms Medium Time Usage 4.55 Mbps 3.17 Mbps 1.54 Mbps 0.85 Mbps Path Throughput Path Medium Time Metric (MTM) = 13.9 ms Link Throughput

27. MTM Example Source Destination 11 Mbps 5.5 Mbps 2 Mbps 1 Mbps 5.5 + 2 1 1.04 Mbps 0.85 Mbps 2.5ms 3.7ms 7.6ms 13.9ms 11 Mbps 5.5 Mbps 2 Mbps 1 Mbps 7.6ms3.7ms 13.9ms = 11.3 ms Medium Time Usage 4.55 Mbps 3.17 Mbps 1.54 Mbps 0.85 Mbps Path Throughput Path Medium Time Metric (MTM) = 13.9 ms Link Throughput

28. MTM Example Source Destination 11 Mbps 5.5 Mbps 2 Mbps 1 Mbps 11 + 2 5.5 + 2 1 1.15 Mbps 1.04 Mbps 0.85 Mbps 2.5ms 3.7ms 7.6ms 13.9ms 11 Mbps 5.5 Mbps 2 Mbps 1 Mbps 2.5ms7.6ms 3.7ms 13.9ms = 10.1 ms = 11.3 ms Medium Time Usage 4.55 Mbps 3.17 Mbps 1.54 Mbps 0.85 Mbps Path Throughput Path Medium Time Metric (MTM) = 13.9 ms Link Throughput

29. MTM Example Source Destination 11 Mbps 5.5 Mbps 2 Mbps 1 Mbps 11 + 11 11 + 2 5.5 + 2 1 2.38 Mbps 1.15 Mbps 1.04 Mbps 0.85 Mbps 2.5ms 3.7ms 7.6ms 13.9ms 11 Mbps 5.5 Mbps 2 Mbps 1 Mbps 2.5ms 7.6ms 3.7ms 13.9ms = 5.0 ms = 10.1 ms = 11.3 ms Medium Time Usage 4.55 Mbps 3.17 Mbps 1.54 Mbps 0.85 Mbps Path Throughput Path Medium Time Metric (MTM) = 13.9 ms Link Throughput

30. MTM Advantages Paths which minimize network utilization, maximize network capacity Paths which minimize network utilization, maximize network capacity Global optimum under complete interference Global optimum under complete interference Excellent heuristic in even larger networks Excellent heuristic in even larger networks Avoiding low speed links inherently provides increased route stability Avoiding low speed links inherently provides increased route stability High speed links operate with greater margin and are more elastic under changes High speed links operate with greater margin and are more elastic under changes