Scalable Ad Hoc Network routing: from battlefield to vehicle grids and pervasive computing

Transcription

1 Scalable Ad Hoc Network routing: from battlefield to vehicle grids and pervasive computing MedHocNet June 22-24, 2005 Mario Gerla Computer Science Dept UCLA

2 Outline Large scale ad hoc nets: who needs them? Scalable ad hoc wireless projects at UCLA The ONR Minuteman project : Scalable routing The NSF WHYNET project: Scalable models and hybrid emulation testbed Scalable routing: Landmark solution Bringing ad hoc networks to commerce From large scale to small scale! The vehicular grid The future of ad hoc commercial networking

3 Infrastructure vs Ad Hoc Infrastructure Network (cellular or Hot spot) Ad Hoc, Multihop wireless Network

4 General Ad Hoc Network Characteristics Instantly deployable (no fixed infrastructure) Satisfies a temporary need Node mobility, relocation Limited battery power Multi-hopping ( to save power, overcome obstacles, enhance spatial spectrum reuse, etc.)

5 Ad Hoc Network Applications Military Automated battlefield Civilian Homeland defense Disaster Recovery (flood, fire, earthquakes etc) Law enforcement (crowd control) Search and rescue in remote areas Environment monitoring (sensors) Space/planet exploration

6 Ad Hoc Network Applications (cont) Commercial Vehicle comms (car navigation safety, etc) Patient monitoring Domotics (eg, multimedia sets indoor interworking) Ad hoc collaborative computing (Bluetooth) Sport events, festivals, conventions Networked video games at amusement parks, etc Commercial Killer Application?.stay tuned!

7 The Packet Radio project In 1971 (two years after ARPANET), DARPA starts the Packet Radio project DARPA, Army and Navy support ad hoc net research Ad hoc net technology has climbed to high sophistication and - large scale Battlefield is the killer application for ad hoc nets

8

9 Ad Hoc nets at UCLA: AINS (Autonomous Intelligent Networked Systems) Project 5 year research program (Dec 2000 Dec 2005) sponsored by ONR 7 Faculty Participants: 3 in CS Dept, 4 in EE Dept Goal: design a robust, scalable network architecture for autonomous agents

10 SURVEILLANCE MISSION AIR-TO-AIR MISSION SATELLITE COMMS UAV-UAV NETWORK SURVEILLANCE MISSION STRIKE MISSION COMM/TASKING Unmanned Control Platform COMM/TASKING RESUPPLY MISSION COMM/TASKING UAV-UGV NETWORK Manned Control Platform FRIENDLY GROUND CONTROL (MOBILE) Algorithms and Protocols for a Network of Autonomous Agents

11 The AINS theme: networking FLIR

12 UCLA Field Test May 2004

13 The UCLA Field Demo Goals - Demonstrate a real ad hoc network - Apply ODMRP (On Demand Multicast Routing Protocol) to a realistic scenario Approach - Aerial nodes: Blimps with laptops - Mobile ground nodes: men/robots carrying laptops - Routing protocol: ODMRP - Scenario: cooperative surveillance of a large area

14

15 Video from Blimp: Region of Interest QuickTime and a TechSmith EnSharpen decompressor are needed to see this picture.

16 Video from Blimp: Intruder QuickTime and a TechSmith EnSharpen decompressor are needed to see this picture.

17 Testing large scale solution: Hybrid Simulation testbed Simulated large-scale network Access Nodes & Hybrid Simulation Server Cluster Small-scale Real Testbed Internet

18 WHYNET - Network Testbed at UCLA Wireless Hybrid Networked Testbed Sponsored by NSF (2003 to 2007) A consortium of seven Universities (UCLA, USC, UCB, UCD, UCR, UCSD, U-Delaware) Main Goal: develop test environments/tools: Radios (MIMO, OFDM, UWB, sensor radios, etc) MAC protocols (directional antennae) Sensor (low energy protocols) Network protocols (QoS, Scalability, interconnection) Security Approach: share results/code/platforms Center piece: hybrid emulation environment

19 Testbed Picture

20 Hybrid Simulation Experiment 36 UGVs, 36 UAVs, 3 battleships, 1 submarine, 3 high elevation UAVs UGV scouts the area; it transmits video using ODMRP broadcast QuickTime and a Microsoft Video 1 decompressor are needed to see this picture. Video packets travel in the simulated network and are played on an actual camera at the command post IMPACT: perception based video adaptation protocol design

21 Challenge in AINS : Scalable routing Tens of thousands of nodes Nodes move in various patterns QoS communications requirements Hostile environment jamming

22 Existing ad hoc routing solutions On demand routing (DSR, AODV) Proactive routing (eg, DSDV, Optimal Links State Routing - OLSR) Explicit hierarchical routing

23 On-demand Routing reply(0) query(0) reply(0) query(0) 1 0 query(0) 3 Advantages: - no periodic routing O/H - no large routing tables Limitations: - flood-search does not scale to large networks 2 reply(0) query(0) query(0) 5 4 query(0) query(0)

24 On-demand Routing query(0) 0 reply(0) reply(0) query(0) 1 query(0) 3 2 query(0) reply(0) query(0) 5 4 query(0) query(0)

25 On-demand Routing query(0) reply(0) reply(0) query(0) 1 0 query(0) 3 Advantages: - no periodic routing O/H - no large routing tables Limitation: flood-search does not scale to large networks 2 query(0) reply(0) query(0) 5 4 query(0) query(0)

26 Distance Vector 0 Routing table at node 5 : Destination Next Hop Distance É É É 2 3 Tables grow linearly with # nodes 4 Control O/H grows with mobility and size 5

27 Link State Routing At node 5, based on the link state pkts, topology table is constructed: 0 {1} {0,2,3} 1 3 {1,4} Dijkstra s Algorithm can then be used for the shortest path O/H grows with N {1,4,5} 2 {2,4} 5 4 {2,3,5}

28 Making Link State more scalable Link State explodes because of Link State update overhead Question: how can we reduce the O/H? Answer: Topology reduction (1) if the network is dense, use fewer forwarding nodes (2) if the network is dense, advertise only a subset of the links Result: IETF MANET OLSR (Optimal Link State Routing)

29 OLSR Overview In LSR protocol a lot of control messages unnecessarily duplicated In OLSR only a subset of neighbors (Multipoint Relay Selectors) retransmit control messages Reduce flooding overhead OLSR retains all the advantages of LSR: Does not depend upon any central entity; Tolerates loss of control messages;

30 Optimized Link state routing (OLSR) 24 retransmissions to diffuse a message up to 3 hops 11 retransmission to diffuse a message up to 3 hops Retransmission node Retransmission node

31 Multipoint Relays (MPR) cont. Every node keeps a table of routes to all known destination through its MPR nodes Every node periodically broadcasts list of its MPR Selectors (Reduced Link State list) Upon receipt of MPR information each node recalculates and updates routes to each known destination Route is a sequence of hops through MPR s from source to destination All the routes are bidirectional

32 Neighbor sensing Each node periodically broadcasts Hello message: List of neighbors with bidirectional link List of other known neighbors. (If node sees itself in this list it adds the sender to neighbors with bidirectional link) Hello messages permit each node to learn topology up to 2 hops Based on Hello messages each node selects its set of MPR s

33 Example of neighbor table One-hop neighbors Two-hop neighbors Neighbor s id State of Link Neighbor s id Access through 2 Bidirectional Unidirectional MPR 15 3 Also every entry in the table has a timestamp, after which the entry in not valid

34 MPR Selection MPR set need not to be optimal hard problem to find an optimal set Greedy heuristic: select node with best 2-hop cover increment MPR is recalculated after a change in onehop or two-hops neighborhood topology

35 Fisheye Concept Routing update frequency decreases with distance to destination Higher frequency updates within a close zone and lower frequency updates to a remote zone Highly accurate routing information about the immediate neighborhood of a node; progressively less detail for areas further away from the node

36 Scope of Fisheye Hop=1 Hop=2 Hop>2 31

37 OFLSR = OLSR + Fisheye Node density fixed, variable # of nodes, no mobility Data Packet Delivery Ratio OLSR OLSR + FSR Network Size (# of nodes) Delivery rate vs Network Size

38 Where do we stand? OLSR can dramatically reduce the state sent out on update messages It effectively reduces the working topology in dense networks. Fisheye further reduces control traffic O/H However, the table size still grows with O(N) Cannot handle large scale nets in the thousands of nodes! What to do?

39 APPROACH: use hierarchical routing to reduce table size and table update overhead

40 Hierarchical State Routing (HSR) Loose hierarchical routing in Internet Main challenge in ad hoc nets: maintain/update the hierarchical partitions in the face of mobility Solution: distinguish between physical partitions and logical grouping physical partitions are based on geographical proximity logical grouping is based on functional affinity between nodes (e.g., tanks of same battalion, students of same class) Physical partitions enable reduction of routing overhead Logical groupings enable efficient location management strategies using Home Agent concepts

41 First try: Hierarchical Routing - multilevel partitions Level = HSR table at node 5: 2 3 DestID 1 Path 5-1 Level = <1-2-> <1-4-> 5-7 Level = 0 (MAC addresses) <3--> Hierarchical addresses 5-7 HID(5): <1-1-5> HID(6): <3-2-6> Problem: Mobility!!

42 HSR - logical partitions and location management Logical (IP like) type address <subnet,host> Each subnet corresponds to a particular user group (e.g., tank battalion in the battlefield, search team in a search and rescue operation, etc) logical subnet spans several physical clusters Nodes in same subnet tend to have common mobility characteristic (i.e., locality) logical address is totally distinct from MAC address

43 HSR - logical partitions and location management (cont d) Each subnetwork has at least one Home Agent to manage membership Each member of the subnet registers its own hierarchical address with Home Agent periodical/event driven registration; stale addresses are timed out by Home Agent Home Agent hierarchical addresses propagated via routing tables; or queried at a Name Server After the source learns the destination s hierarchical address, it uses it in future packets

44 Better Approach: Landmark Routing Main observation: nodes move in groups Groups are dynamically recognized Landmark elected in each group Landmarks advertise their position to entire net Landmark Logical Group

45 LANMAR Routing (cont) Builds upon existing routing protocols (1) local routing algorithm that keeps accurate routes within local scope < k hops (e.g., OLSR, FSR) (2) Landmark routes advertised to all mobiles using a Distance Vector approach Landmark Logical Group

46 Landmark Routing In action (cont) A packet to local destination is routed directly using local tables A packet to remote destination is routed to Landmark corresponding to logical address Once the landmark is in sight, the direct route to destination is found in local tables. Benefits: low storage, low update traffic O/H; and, it tolerates motion Landmark Logical Subnet

47 Dynamic Group Formation QuickTime and a Microsoft Video 1 decompressor are needed to see this picture.

48 Delivery Ratio LANMAR-DSDV LANMAR-FSR OLSR LANMAR-OLSR DSDV FSR DSDV and FSR degrade rapidly when number of nodes increases OLSR generates excessive control packets, cannot exceed 400 nodes

49 What about georouting? Very scalable No state in routers nor in packet header No flood search Cons: It requires GPS (or virtual coordinate computation) It requires GLS (Geo Location Service) Suffers of the hole problem (GPSR not very efficient) Pros: Geo routing very effective in high mobility forwarding We are exploring joint geo-routing and LANMAR

50 Mobile Backbone Landmark Overlay provides routing scalability However the network is still flat - paths have many hops poor TCP and QoS performance!! Solution: Mobile Backbone Overlay

51 Backbone Node Automatic Deployment Objectives Robust and autonomous backbone network maintenance Uniform distribution to cover the field Approach Dynamic backbone node election: Deploy redundant backbone capable nodes and select a few Backbone node automatic placement: Relocate backbone nodes from dense to sparse regions

52 Mobile Backbone QuickTime and a Microsoft Video 1 decompressor are needed to see this picture.

53 Important breakthroughs in ad hoc networking Large, mobile network routing - exploit mobility to achieve scalability Disruption tolerant networks for extreme conditions Swarming: unmanned (in the air, under water..) Multipath routing: scatter/fuse MAC + routing protocols (cross layer interaction) Multi-antenna radios (MIMO) - jam protection Spectrum agile radios (use whatever portion of spectrum is available) Most of above work sponsored by DoD

54 Ad Hoc net commercial applications? Most of the on-going mobile ad hoc network research is aimed at: Military, large scale applications Civilian applications (disaster recovery, homeland defense, planetary exploration, etc) Large mobile sensor platform deployments Is this technology ready for transfer to commodity ad hoc applications? Where are the commercial applications?

55 Potential divide : Military want big systems Military and civilian requirements/ characteristics: Typically, large scale Instant deployment Infrastructure absent (so, must recreate it) Very specialized mission/function (eg, UAV scouting behind enemy lines) Critical: QoS, jam protection, survivability Not critical:cost, Standards, Privacy

56 Commercial customers want small scale and cheap! Commercial, commodity requirements Mostly, small scale Cost is a major issue (eg, ad hoc vs W-LAN vs 2.5 G) Connection to Internet often available Need not recreate infrastructure, rather bypass it whenever it is convenient Proximity applications Standards are critical to cut costs and to assure interoperability Privacy, security is critical

57 Vision: Opportunistic Ad Hoc Networking Commodity ad hoc networks will not happen as isolated, self configured nets (as in the military model) Rather, they will emerge as a no cost extension of the infrastructure Examples of Ad Hoc extensions (of Wireless Internet) Indoor W-LAN extended coverage Hot spot (Mesh Networks) extensions Group of friends sharing 3G access via Bluetooth Peer 2 peer networking in the urban vehicle grid

58 Urban opportunistic ad hoc networking From Wireless to Wired network Via Multihop

59 Urban Ad Hoc net in action: Safe Driving Vehicle type: Cadillac XLR Curb weight: 3,547 lbs Speed: 75 mph Acceleration: + 20m/sec^2 Coefficient of friction:.65 Driver Attention: Yes Etc. Alert Status: None Vehicle type: Cadillac XLR Curb weight: 3,547 lbs Speed: 65 mph Acceleration: - 5m/sec^2 Coefficient of friction:.65 Driver Attention: Yes Etc. Alert Status: None Alert Status: Inattentive Driver on Right Alert Status: Slowing vehicle ahead Alert Status: Passing vehicle on left Vehicle type: Cadillac XLR Curb weight: 3,547 lbs Speed: 75 mph Acceleration: + 10m/sec^2 Coefficient of friction:.65 Driver Attention: Yes Etc. Alert Status: Passing Vehicle on left Vehicle type: Cadillac XLR Curb weight: 3,547 lbs Speed: 45 mph Acceleration: - 20m/sec^2 Coefficient of friction:.65 Driver Attention: No Etc.

60 Opportunistic piggy rides in the urban mesh Pedestrian transmits a large file in blocks to the passing cars, busses The carriers deliver the blocks to the hot spot

61 Hot Spot Hot Spot

62 STOP Power Blackout Hot Spot Hot Spot

63 STOP Power Blackout

64 CarTorrent : Car to car downloading of multimedia files

65 You are driving to Vegas You hear of this new show on the radio Video preview on the web (10MB)

66 Highway Infostation download Internet file

67 Problem: Stopping at gas station to download is a nuisance Observation: many other drivers are downloading the same files (like in the Internet) Solution: Co-operative P2P Downloading using the Car to Car ad hoc nets (Car-Torrent)

68 Partial download from Infostation Internet Download

69 Co-operative P2P Download Internet P2P Exchange of Pieces Vehicle-Vehicle Communication Routing via osmosis: an overlay routing approach

70 Car-Torrent: opportunistic incentives Bandwidth at the infostation is limited and not convenient It can become congested if all vehicles stop It is a nuisance as I must stop and waste time GPRS and 3G bandwidth is limited The car to car bandwidth is unlimited and FREE! Car to car ad hoc net already exists (safe navigation requirements)

71 Commodity Ad Hoc Nets: the Future Commercial ad hoc networks will happen as opportunistic extensions Commercial applications will have small scope - proximity concept New research (beyond military) critical: P2P overlay protocols (for proximity networking) Security/privacy Third Party forwarding incentives Soft handoff Integration with infrastructure etc

72 The End Thank You!