routed mobile ad hoc networks

7/27/2019 Routed Mobile Ad Hoc Networks

http://slidepdf.com/reader/full/routed-mobile-ad-hoc-networks 1/167

1© 2005 Cisco Systems, Inc. All rights reserved.

RST-2210

11089_05_2005_c3

ROUTED MOBILE AD HOC

NETWORKSRST-2210




RST-2210

11089_05_2005_c3

Session Objectives

• Introduction to concept of Mobile, Autonomic, andPervasive Networks

• Introduction to research and development inIP Mobility

• Introduction of technology development in RoutingProtocols, focusing on mobil ity

Mobile Ad-hoc Networks (MANet)

Network Mobility (NeMo)




RST-2210

11089_05_2005_c3

Mobility Overview

• Mobility Components

• Protocol Requirements

• A Seven Layer Problem

• Network Evolution

Mobil ity is in a state of development

It is part of a larger picture

• Networking Spectrum• Mobile IP

• NeMo—Network Mobility

• MANet—Mobile Ad Hoc Networking


http://slidepdf.com/reader/full/routed-mobile-ad-hoc-networks 4/167444© 2005 Cisco Systems, Inc. All rights reserved.

RST-2210

11089_05_2005_c3

Mobility Components

• Portability

The ability to be transported easily;moveable

• Roaming

The ability to be reached anywhere(while in motion)

• Handoff

The abil ity to change your access linkwithout interruption



RST-2210

11089_05_2005_c3

Two Types of Roaming

• Nomadic

Does not maintain an active IP connection while roamingEstablishes an IP connection while at a specific location

Will typically need to re-initiate application connections

• Seamless (Cellular)

Maintains an active IP connection while roaming

Roams between multiple connection types

Applications stay connected while roaming

Always reachable at the same IP address* while roaming



RST-2210

11089_05_2005_c3

Mobility Protocol Requirements

• Movement Detection

Discover when accesslink changes

• Edge Discovery

Identify access edge

• Signaling

Inform the network of thenew location

• Path (Re)establishmentUpdate the delivery method sothat traffic is delivered to thenew location

All are logical operations, but not necessarily physical



RST-2210

11089_05_2005_c3

Network Evolution

Mobile IP

Network Mobility

Network Mobility +

Wired/StaticNetworks

Mobile Access

(Networksin Motion)

Distributedor Regional

Services

Metropolitanor Regional

Mobility

AutonomicServices

PervasiveNetworks

MESH Ad Hoc Routing

AONS

Mobile IP +MESH + Ad HocRouting + AONS

Presence

HostMobility

Service

Mobility

You are here



RST-2210

11089_05_2005_c3

Mobile Access (Networks in Motion)

• Provide connectivity to fixedservices “ back home”

Mobile access routers andmobile hosts provideconnectivity extension

Services are in the core of the network

Do not support transit traffic

• Single administrative domain

When nodes roam they musttunnel back to that domain

Public Safety/Homeland Security



RST-2210

11089_05_2005_c3

Metropolitan or Regional Mobility

• Provide connectivity to a network,typically with a fixed core, to cover

a region or metropolitan area• Has [one-to-many] wireless links

and [one-to-many] access points,points of presence, etc.

• Services remain fixed

• Different administrative domains

• May require a tunnel back to thedomain or may be peer-to-peer likea mesh network

Mesh assumes Layer 2 with backhaul

Very low mobili ty ratesVery fast STP

Metro MobileNetworks



RST-2210

11089_05_2005_c3

Distributed or Regional Services

• Provide an independent, autonomous network thatcan have connectivity back to other network(s)

while covering a large region or metropolitan area

• Mobile nodes provide forwarding and other services; allow transit traffic for interconnectivity

• May have multiple peers

Metro IPOptical Ring

Cellular Cisco Switch

802.11Bridging

City Hall

Retail




RST-2210

11089_05_2005_c3

Autonomic Services

• Dynamic network formation and service delivery inan impromptu or improvised* manner

• Network formation includes

Routing

Edge detection

Dynamic peering

Graceful insertion and deletion

• Transit traffic

• What most customers are asking for today

Currently this does not exist, but this is the short term goal




RST-2210

11089_05_2005_c3

Pervasive Networking

• Virtualizedservices in

the network

• Highlyintegratedapplications

• One network,many virtualplanes

• A network of

networks

Applications

Public Space

Closed Area

Hospitality

Mobility

10/100BaseT

DSL/Cable

802.11/RF/FSO

VPN

Technology Access

Pick your favorite

Voice, Video

Collaboration

Commerce




RST-2210

11089_05_2005_c3

Pervasive Networks

Where We Want to Go

• Self-configuring, -managing, and -healing networks• Mimic social networks rather than technical bounds

• Technology molds to fit policy and interaction,

rather than people molding to fit technical bounds

• Zillions of nodes

The vision where any person or device can be connected

• The fulfil lment of Metcalf's Law

The value of the network is proportional to connectivity




RST-2210

11089_05_2005_c3

End Game: Autonomic and Pervasive

Any Device, Any Place, Any Time, Always Connectable*




RST-2210

11089_05_2005_c3

How Do We Get There?

• “ Maintain continuous IP connectivity whileindependent of the connection type”

• …but a continuous IP connection to what?

People talk to people, not machines

People want information available from machines (video,mapping, etc.)

Connecting devices is fine, but people want services

Local and centralized services

Services available on any host anyplace and anytime




RST-2210

11089_05_2005_c3

Current Challenges

• Last-Mile converged services

• On-the-Move communications• Real-time situational awareness

• Ubiquitous IP connectivity with always on,constant communications

• Interoperability

• Adaptable, seamless, robust architectures




RST-2210

11089_05_2005_c3

Issues: Mobile Networking Spectrum

Large HQ

FOB/Airbase Task Group

Thousands of Nodes Tens of NodesScalability

Always Intermittent/Peer-to-Peer Client/Server Access

Relatively Fixed Ad Hoc Mobility

Static Addressing

Mobile IP Ad HocRouting

Flooding

Increasing Mobil ityWired and Wireless Wireless only

Continual IntermittentReach Back to HQ or Central

Fixed Assets,

Predefined Connections

Mobile Assets

On-the-Fly ConnectionsSecurity

And do not forget about Power

Mobile IP+



DEFINING THE PROBLEM SPACE

RST-2210

11089_05_2005_c3 202020© 2005 Cisco Systems, Inc. All rights reserved.




RST-2210

11089_05_2005_c3

Defining the Problem Space

No LateralConnections

Network Moves as a Unit

AlwaysHasReachbackto “ Home”

Device/Network Mobil ity

• A host or network(including attachedhosts) moves as a unit

• The host or network canalways reach back to a“ home” without worryingabout bandwidth

uti lization, etc.• Each host or network

connects only to its

upstream—no lateralconnections




RST-2210

11089_05_2005_c3


• Hosts move independently

• If a network is going tomove as a unit, hostsattached to it would be

numbered in the same IPaddress space

• No “ reach back” to any

specific network isguaranteed

• Lateral connectionsbetween devices arecommon

LateralConnections

Independent DeviceMovement

Ad Hoc Mobility




RST-2210

11089_05_2005_c3


• Device/Network Mobility

Mobile IP/Mobile Router/NeMo fit these requirements

Reach back consumes limited bandwidth across networks

Placing the “ home agent” is a difficult puzzle to solve

Lack of lateral connections between “ trees”

• Ad Hoc

Mobile Ad Hoc Networks (MANet) fits these requirements

Ad hoc networks may have a major impact onupstream “ clouds”

Will take time to develop and deploy

Scale and administration are possible concerns




RST-2210

11089_05_2005_c3


• NeMoLarge, relatively stable

networksGuaranteed reachback

Only one path out, or mult iple“ equally good” paths (optimumrouting is not a requirement)

• MANets can attach behindthe NeMo

Routing information has to floatup the NeMo and be injectedinto routing at the edge

The MANet only receives adefault route (no optimalrouting through the NeMo)

Back doors cannot be used toroute to NeMo devices

MANet

B a c

k D

o o r

Home NetworkReachback for Mobile IP

Always

Available

NeMo

RoutingInformationPassed Through

to Edge




RST-2210

11089_05_2005_c3


• MANet

Higher mobility (not the same

thing as faster mobility)

No “ home,” or no guaranteedreachback

Multihoming with optimalrouting important

• NeMos can attach behinda MANet

Reachback is through theMANet

The NeMo devices don’t see

the MANet as anything out of the ordinary

HomeNetwork

NeMo

Reachback

Through MANet




RST-2210

11089_05_2005_c3


• Classify each device as MANet or NeMo

Examine the environment and assumptions for each deviceor network

Compare these to the limitations and capabili ties of eachtechnology

• Resolve the technical issues

Make certain NeMos have the ability to leak routinginformation upstream through the cloud

Discover the NeMo/MANet/infrastructure edge and respondcorrectly

• Resolve the policy issues

Nesting




RST-2210

11089_05_2005_c3

Network Virtualization and Distribution

Mobile access network

M o b

i l e B a c k b

o n e

Authorization

ServicesMove to the

MobileBackbone

Distributed Protocols

Bind Them

Content

Mi c r om

o b i l i t y

Gr o u pM

o b i l i t y

A c c e s

s t o S er v i c e s :

I nf r a s t r u c t ur e

an d M o b i l i t y

Servers Management




RST-2210

11089_05_2005_c3

Vertical Market Applications

• Public Services911 Services (Police/Fire/Medical)

• Armed Services Army, Navy, Air Force, Marines, NATO

• Commercial Markets

Package delivery fleetsTrucking

Rental fleets

• Consumer Telematics

Railroads

Airplanes

Entertainment




RST-2210

11089_05_2005_c3

Sensors and Surveillance

IPv4/IPv6

Mobile

Router

Movementdetection

application

Central Site

Home Network

IPv4/IPv6

Mobile

Router

Internet

IPv4/IPv6

Mobile

Router

IPv4/IPv6Mobile

Router




RST-2210

11089_05_2005_c3

Public Safety/Personal Area Networks

Attach to Homes, Buses, Cars, Planes,Trains, Hotspots…

Body Sensors

Position

EnvironmentalSensors

Way-Out Navigator

Mesh NetworkBattery saving

Determine position and save historyCalculates route to the exit and navigate

Body temperature, heartbeat, pressure

Temperature, Smoke, positioning grid

Nodes and Sensors acting as routers

Tune radio power dynamically

• Drop sensors to provide wireless backbone andpositioning grid

• Body sensor indicates a sudden drop of blood pressure

• Application navigates the fire fighter to the troubledfire fighter

• A fire fighter in confined area can only reach the base stationthrough other nodes or sensors

• Environmental sensor indicates temperature is too high• The fire fighter uses the way-out navigator to reach the exit




RST-2210

11089_05_2005_c3

Smart Vehicle/Smart Road

Portal

Internet

GPRS

WGB

Modem

GPRS

AP

DSL Modem

MR

IPv4IPv4

IPv4

IPv4

IPv4

IPv4

IPv6IPv4IPv6IPv4

HA

PC

IPv6

Adding IPv6 toCisco Mobile

Networking Solution

Cisco MAR 3200

MIPv6 in GPRS v4 Tunnel




RST-2210

11089_05_2005_c3

Routing Protocol Review

What Does a Routing Protocol Do?

• Routing considers throughput and delay

• Finds neighbors

• Selects best paths

• Updates based on changes and failures

Standard IP routing follows the longest prefix to reach thedestination address

Routing information is aggregated




RST-2210

11089_05_2005_c3

IP Routing and Full Mobility

Full Mobility Requires a Constant AddressEven When Changing Access Links

10.82.79/24

?

Mobile Node192.168.100.10 D

B CInternet

192.168.100/24

Route Next

192.x.x.x B

Route Next

192.168.x.x A

A




RST-2210

11089_05_2005_c3

Components of a Routing Protocol

• High-level characteristics

Path selection method

Convergence (Path [re]establishment)

Scalability

Complexity

• Implementation details

Update messages (Update signaling)

Neighbor discovery (Location discovery)

Metrics

Special use features (Movement detection)




RST-2210

11089_05_2005_c3

Mobility with Traditional Routing Protocols

• Local Area Mobility (LAM)—ARP discovery-basedhost routing

• Mobile host routes redistributed into IGP

• More hosts means more host routes and

route updates• IGPs are not designed to reconverge many times

a second

• The solution… Mobile IP



MOBILE IP REVIEW

RST-2210





RST-2210

11089_05_2005_c3

The Solution: Mobile IP

Route Next

192.168.100.10 Tunnel

Route Next

192.x.x.x B

Route Next

192.168.x.x A

CB

A

D

Mobile IP Is Not a Routing Protocol; Consider It

a Protocol for Mobile Devices to Signal and

Establish Dynamic Tunnels for Location

Updates, Eliminating Host Route Propagation

Internet

Route Next

192.168.100.10 E0

Mobile Node

192.168.100.10




RST-2210

11089_05_2005_c3

Mobile IPv4 and Mobile IPv6 Terminology

MN, MOBILE NODE

An IP host that maintains network connectivityusing its “ home” IP address, regardless of which subnet (or network) it is connected to

IP Networks

D

C

10.1.1.1

Subnet

30.1.1.0/24

CN

AFA

MN

10.1.1.1

MN

HA MN

Home Address

CN,CORRESPONDENTNODE

Destination IP hostin session with aMobile Node

HA, HOME AGENT

Maintains an associationbetween the MN’s “ home”IP address and its Care Of

Address (loaned address)

on the foreign networkRedirects and tunnelspackets to the care of address on the foreignnetwork

FA, FOREIGN AGENT (V4 ONLY)Provides an addressable point of attachmentto the MN called Care Of Address (COA)

Maintains an awareness for all v isi ting MNs

Acts as a ‘relay’ between the MN and itsHome Agent

Receives all packets for the MN from the MN’sHome Agent

Subnet

20.1.1.0/24




RST-2210

11089_05_2005_c3

Mobile IPv4: Triangle Routing

• Inbound traffic is sent to the home subnet

• The Home Agent intercepts the traffic while themobile node is registered as away

• Traffic is tunneled to the COA and forwarded to MN

• Traffic from the mobile node is forwarded directlyto the correspondent host via the FA

MobileNode

Home Agent

Foreign Agent

CorrespondentHost

COA




RST-2210

11089_05_2005_c3

Mobile IPv4

HA

FA

C3200 Mobile

Access Router

Roamed Connection = WLAN, CDMA,…

IP Traffic Path to the mobile IP Host/Device

All IP Traffic to Devices on a Mobile Network Will AlwaysRoute Through the Mobile Networks Home Agent (HA) onTheir Home Network (RFC 3344)

M bil IP 4




RST-2210

11089_05_2005_c3

Mobile IPv4

Enhanced IPv4 (RFC 3344) Mobile Networks Optimizes theRouting Path Between an IP Host Connected to the Foreign Agent (FA) and Mobile Networks That Roam and Connect to

the Same FA

Roamed Connection = WLAN, CDMA,…

IP Traffic Path to the mobile IP Host/Device

HA

FA

C3200 Mobile

Access Router

M bil IP 6 B i O ti




RST-2210

11089_05_2005_c3

Mobile IPv6 Basic Operation

1.

CN

HA

MN

4. 3.

2.

1) CareOf Address

The Mobile Nodeobtains an address inthe visited networkby autoconfigurationand neighbor discovery

2) Bind Update

The Mobile Nodefinds a Home Agentand negotiates abinding betweenCoA and Home Address

3) Tunnel and RO

Packets are exchangedwith the CorrespondentNode via a reversetunnel or directly usingRoute Optimization

M bil IP D li i C




RST-2210

11089_05_2005_c3

Mobile IP: Delivering Convergence

• Standards–based

• Unifying access• Transport and

network

independence• Improving the

user experience

Applications roamwith the user

Seamless roamingmaintains sameaddress



NETWORK MOBILITY (NeMo)

RST-2210


Introduction to NeMo Protocol




RST-2210

11089_05_2005_c3

Introduction to NeMo Protocol

• Part of IPv6

• Leverages Mobile IPv6 defined in RFC 3775Home agent, bi-directional tunnel, correspondent node

• Layer 2 agnostic

• Traffic is better optimized if it is between “ home”network and mobile nodes

• Routing is required to forward tunneled traffic tothe home

• Concept proved in IPv4—Cisco Mobile Networks v4

What Is NeMo?




RST-2210

11089_05_2005_c3

What Is NeMo?

“ The NeMo Working Group is concerned with managingthe mobility of an entire network, which changes, as a

unit, its point of attachment to the Internet and thus itsreachability in the topology. The mobile networkincludes one or more mobile routers (MRs) whichconnect i t to the global Internet.

A mobile network is assumed to be a leaf network,i.e., it will not carry transit traffic. However, it could bemultihomed, either with a single MR that has multiple

attachments to the internet, or by using multiple MRsthat attach the mobile network to the Internet.”

Network Mobility (NeMo) IETF Working Group Charter

http://www.ietf.org/html.charters/NeMo-charter.html

NeMo Basic

http://www.ietf.org/html.charters/nemo-charter.html





RST-2210

11089_05_2005_c3

NeMo Basic

• Adopts the methods for host mobility support in

Mobile IP• Extends the host mobility

support to “ group”mobility support

Group can be a singlesubnet or multiple subnets

NeMo router differentiatesitself from mobile nodeusing a new R flag in theBinding Update

When acting as a mobilerouter, “ Mobile NetworkPrefix option” may be

included in theBinding Update

CN

Correspondent Node

Home Agent

Access Router

IPv6 Backbone

Mobile Router

B i - d i r e c t i o n a l T

u n n e l

Mobile Network

Node(Mobile Network Node—MNN)

Mobile Networks v6: NeMo Basic




RST-2210

11089_05_2005_c3

Mobile Networks v6: NeMo Basic

Mobile Networks v4 Like Capabilities

• Ability to roam across multiple links withinthe router

• Roaming over a single link at a time

• Connectivity to one Home Agent at a time

• Ability to prefer one link over another for roamingdecisions ‘Bandwidth and/or Priority’

• Mobile Router registers one or more directlyconnected subnets (stub network)

• Static and Dynamic Network Registration (implicitand explicit)

• Statically configure Home Agent

Advertise Mobile Network(s) Information




RST-2210

11089_05_2005_c3

Advertise Mobile Network(s) Information

• Explicit approach:

Includes the “ Mobile Network Prefix” option in the

Binding UpdateSimilar to “ dynamic mobile networks” in Cisco mobilenetwork v4

• Implicit approach:

Home Agent already knows which prefixes belong to aMobile Router, such as via static configuration

Similar to “ static mobile networks” in Cisco mobilenetwork v4

• Running routing protocols across thebi-directional tunnel

Routing domain within the group network is typically stable

Dynamic Home Agent Address Discovery




RST-2210

11089_05_2005_c3

Dynamic Home Agent Address Discovery

R o u t e r

A d

v e r t i s e m e n t

R o u t e r Ad v e r t i s e m e n t

ICMP HAD Request

ICMP HAD Reply

• STEP 1: Each HA receivesRouter Advertisement from allother HAs and maintains an

ordered list of the Current HAs• STEP 2: MR send an ICMP HA

Discovery Request to IPv6 HA Anycast address

•

STEP 3: First HA receivingreplies with the Global IPaddresses of the HAs in order of preference

• STEP 4: MR sends a Binding

Update to the first Home Agentin the list, HA answers backwith a Binding Acknowledgment

• STEP 5: A bi-directional tunnelis established between the

Mobile Router and theHome Agent

Home Network

HA 1

HA 2

HA 3Home Agent list:

HA 1 : li fetime

HA 2 : li fetime

HA 3 : li fetime

Home Agent list:

HA 1 : li fetime

HA 2 : li fetimeHA 3 : li fetime

Home Agent list :

HA 1 : li fetime

HA 2 : li fetime

HA 3 : li fetime

Mobile

Router

B i n d i n g

U p d

a t e

B i n d

i n g A c k

Mobile Network Sample Configuration




RST-2210

11089_05_2005_c3

Mobile Network Sample Configuration

• Home Agent configuration:interface Ethernet1

ipv6 address CA5A:4::BB4/64

ipv6 enable

ipv6 mobile home-agent

ipv6 route D093::/64 CA5A:4::9

• Mobile Router configuration:

ipv6 unicast-routingipv6 mobile router

home-network CA5A:4::BB4/64

home-address home-network ::9

interface FastEthernet0/1

ipv6 address autoconfig

ipv6 enable

ipv6 nd suppress-ra

ipv6 mobile router-service roam

interface FastEthernet1/0

ipv6 address D093::1/64

ipv6 enable

Î Home Agent Address

Î Static Route to Mobile Network

Î Home Network

Î Mobile Network

Î Home Address of Mobile Router

Î Roaming interface

Mobile Tunnel Support (IPv4/IPv6 in IPv6)




RST-2210

11089_05_2005_c3

Mobile Tunnel Support (IPv4/IPv6 in IPv6)

IPV6/V4Mobile Network

IPV6

Infrastructure

IPV4

Network

IPV6

Home Agent

IPV6Mobile

Router IPv4 Host

IPV6

Network

IPv6 Tunnel

• Configuration of an IPv6 tunnel between the MRand its Home Agent

• Both IPv4 and IPv6 traffic can go throughthis tunnel

• Mobility is handled at the IPv6 level

Mobile Networks v6: Doors




RST-2210

11089_05_2005_c3

Mobile Networks v6: Doors

• Enables MR to traverse an IPv4 network

• Dual stack Doors Gateways on the edge of the IPv6

network• MR obtains a CoA and derives an IPv6 address

• MR is configured with the Doors Gateway address

• Mobile IPv6 messages are encapsulated in an IPv4UDP packet

• Doors Gateway removes the IPv4 wrapper andforwards the Mobile IPv6 packets to the destination

Home Agent

• Doors Gateway is the default router for the MR, for all traffic to the MR

IPv4 Transport: Doors




RST-2210

11089_05_2005_c3

IPv4 Transport: Doors

IPV4

IPV6

Home Network

Door Gateway

IPV6Mobile Router 6to4 Gateway

IPV6Mobile Network

IPV6

Roaming

IPV6Home Agent

• Allows MN/MR to roam to IPv4 networks seamlessly

• A IPv4/UDP bi-directional tunnel is built

automatically, similar to 6to4 tunnel in IPv6• The UDP wrapper makes NAPT (PAT)

traverse possible

• MR and the Doors gateway need to be dualstack routers

Doors Protocol Overview




RST-2210

11089_05_2005_c3

Doors Protocol Overview

IPV4

NAT/PAT

IPV6

Door GW

Home Agent

MR

CoAv4Private

IPV4

Mapping of IPv4 CoA into

2002:IPv4 Door:1:Door UDP port:IPv4 CoA:MN UDP ports

• The Door gateway’s IPv4 address is known by MN/MR

• As the home agent is in IPv6, it needs to have a IPv6 CoA toreach the MN/MR

MR derives the IPv6 CoA and uses it as the source IP address

• MR encapsulates with the IPv4/UDP

• When the Door receives the packets, it decapsulates theIPv4/UDP and rewrites the IPv4 CoA and MN UDP port in thesource (to support NAT/PAT) with the source address of IPv4

•

When HA receives it, it builds the proper binding cache entryfor the MN/MR

Reverse Routing Header




RST-2210

11089_05_2005_c3

e e se out g eade

RRH Routing Header Type 4Outer IPV6 Header Encapsulated IPV6 Packet

MR3’s HA

MR2’s HA

MR1’s HA

MR3

MR2

MR1

LFN1

Packets route

Access Router

Internet

CNNew Routing Header:Type 4

oNAF S= MR3 Coa D= MR1 HA MR2 Coa MR1 Coa MR1 HoA iNAF S= LFN1 D= CN1 iPAYLOAD

Tunnel

• Works with plain V6 hosts on both ends

• Home Equivalent Privacy option via HA

• First MR or MN builds a tunnel with RRH

• Next MRs add the source to the RRH and overwri te source with their COA

•

A combination of IP Routing in Infrastructure and of On Demand SourceRoute in the mobile cloud to adapt faster to topology changes

Tree Organization with Route Optimization




RST-2210

11089_05_2005_c3

g p

• Each Mobile router hasonly one COA (MIPV6)

• Each Mobile Router attaches to another onefollowing rules that force a

Tree topologyBased on autonomousdecision of eachMobile router

Based on Loop avoidancemechanism

Internet

TLMR

TLMR

Tree Discovery




RST-2210

11089_05_2005_c3

y

References




RST-2210

11089_05_2005_c3

• IETF NeMo Working Grouphttp://www.ietf.org/html.charters/NeMo-charter.html

• IETF Mobility for IPv6 Working Grouphttp://www.ietf.org/html.charters/mip6-charter.html

• IETF Network Mobility (NeMo) Basic Support

Protocolhttp://www.ietf.org/rfc/rfc3963.txt

• Selected NeMo drafts:http://www.ietf.org/internet-drafts/draft-thubert-NeMo-ro-

taxonomy-04.txthttp://www.ietf.org/internet-drafts/draft-thubert-tree-discovery-01.txt

http://www.ietf.org/internet-drafts/draft-thubert-NeMo-

global-haha-00.txt


http://www.ietf.org/html.charters/mip6-charter.html

http://www.ietf.org/rfc/rfc3963.txt

http://www.ietf.org/internet-drafts/draft-thubert-nemo-ro-taxonomy-04.txt


http://www.ietf.org/internet-drafts/draft-thubert-tree-discovery-01.txt



https://datatracker.ietf.org/public/idindex.cgi?command=id_detail&id=12344

https://datatracker.ietf.org/public/idindex.cgi?command=id_detail&id=12344







http://www.ietf.org/html.charters/mip6-charter.html




MOBILE AD HOC NETWORKS(MANet)

MANet: Agenda




RST-2210

11089_05_2005_c3

• Defining MANet

• Mobile OSPF

• MANet Complements and Futures

• Other MANet Protocols

Defining MANet




RST-2210

11089_05_2005_c3

• MobileNodes travel according to some

movement model• Ad Hoc

No fixed infrastructure, but can befixed relative

• NetworkLayer 2—Wireless (CSMA, FDMA,…)

Layer 3—Routed (addressing)

• Most Important Question:What does MANet mean to you?

• Or Specifically:What problem do you think MANet isgoing to solve?

Defining MANet




RST-2210

11089_05_2005_c3

• Is this your problem?

• Random Walk Mobility

• How about this?

• Gauss Markov Mobility

• Or is this your problem?

• Random Waypoint Mobil ity

• And finally

• Group Mobility

Defining MANet




RST-2210

11089_05_2005_c3

• Micro-Mobility—small geographic area with fast,frequent handoffs

• Macro-Mobility—larger geographic area withfewer handoffs

• Seven Synthetic entity mobility models and fivegroup mobility models

• Use Layer 2 or Layer 3 to handle mobility?

• Performance of an ad hoc routing protocol will varysignificantly depending on the mobility model

Actual mobility will probably have multiple models

• No silver bullet—MANet is all about compromiseIncreased complexity

Decreased security/QoSIncreased latency

Highly Mobile Sites




RST-2210

11089_05_2005_c3

Pure Ad Hoc Networking Is Difficult

MANet vs. Structured MANet



“ Ad Hoc Network — A collection of [wireless] (mobile) nodes forming adynamic autonomous network. Nodes may communicate without an

existing infrastructure or centralized administration in a standalone

fashion or may be connected to the larger Internet. They are free to

move randomly and organize arbitrarily with the potential for rapid or unpredictable change. Thus, each node acts both as a router and a

host.”

Public Domain


RST-2210

11089_05_2005_c3

What Is MANet?




RST-2210

11089_05_2005_c3

“ The purpose of the MANet working group is to standardize IP routingprotocol functionality suitable for wireless routing application withinboth static and dynamic topologies with increased dynamics due tonode motion or other factors.

Approaches are intended to be relatively lightweight in nature,suitable or mult iple hardware and wireless environments, andaddress scenarios where MANets are deployed at the edges of an IPinfrastructure. Hybrid mesh infrastructures (e.g., a mixture of f ixed

and mobile routers) should also be supported by MANetspecifications and management features.

Using mature components from previous work on experimentalreactive and proactive protocols, the WG wil l develop two Standardstrack routing protocol specifications:

• Reactive MANet Protocol (RMP)

• Proactive MANet Protocol (PMP).”

Mobile Ad-hoc Networks (MANet) IETF Working Group Charter http://www.ietf.org/html.charters/manet-charter.html

Characteristics of MANets (RFC2501)

http://www.ietf.org/html.charters/manet-charter.html

http://www.ietf.org/html.charters/manet-charter.html




RST-2210

11089_05_2005_c3

• Dynamic topologiesNodes are free to move arbitrarily; network topology may changerandomly and rapidly at unpredictable times

• Bandwidth-constrained, variable capacity linksWireless links have significantly lower capacity than their hardwired counterparts; after accounting for the effects of multiple access, fading, noise, and interference conditions, etc.;

the actual throughput is often much less than a radio's maximumtransmission rate

• Energy-constrained operationSome or all of the nodes in a MANet may rely on batteries or other

exhaustible means for their energy; network and routingoptimization must be cognizant of energy conservation

• Limited physical securityMobile wireless networks are more prone to physical security

threats (i.e., eavesdropping, spoofing, and DOS attacks) thanhardwired networks

What Is a MANet?




RST-2210

11089_05_2005_c3

• Characteristics

Host movement “ frequent”

Topology change “ frequent”

No infrastructure; multi-hop wireless links

Data must be routed via intermediate nodes

• Applications

Emergency services

Military tacticalBusiness/Commercial fleets

Consumer environments

Personal area networking

Dynamic Topology




RST-2210

11089_05_2005_c3

• Random interconnection

Minimal or no engineering

Low (varying) bandwidth links

• Constant or frequent change (motion)

Neighbor changes:New neighbor may be less reliable connection

More reliable connection may be at lower bandwidth

Resulting information propagation (flooding)

• Ad Hoc Routing should consider space and power

Energy-Constrained Operation




RST-2210

11089_05_2005_c3

• Some nodes (e.g., hand-held or laptop devices) arepowered by batteries

• Others (e.g., vehicle-based) may be able to rely on a“ constant” power source

• Battery drain will influence a node’s abili ty to

participate as a routing next-hopYou could suspend a node if it hasn’t participated for someperiod of time, but then how do you wake it up?

Route cost should take energy constraints intoconsideration

• Inefficient data link, MAC, or network layer design

can result in more power being consumed

Radio Characteristics




RST-2210

11089_05_2005_c3

• Directional antennae

Some radios send in a stated direction; have to send when

peer is listening in that direction

• Varying signal strength, link quality

Route cost should take link quality into consideration

• Overlapping connectivity

No unifying concept l ike Designated Router

Haphazard connections

Radio Regions Overlap Haphazardly




RST-2210

11089_05_2005_c3

Limited Physical and Routing Security




RST-2210

11089_05_2005_c3

• RF transmission is inherently less secure thanwired transmission

Easier to snoop or eavesdrop

• Detection avoidance for military applications

• More susceptible to DoS attacks

• Highly (quickly) flapping routes

Network stability

Trapped packets

• False routing information

Black Hole

Full tables

Routing and Mobility Overall




RST-2210

11089_05_2005_c3

• Issues

Frequent route (topology) changes—arbitrary movement

Low (varying) bandwidth links

Route changes may be related to node movement(adjacency changes/hidden and exposed nodes)

Low power/CPU nodes

• Requirements

Decrease control overheadFind short routes

Find “ Stable” routes (despite mobility)

• Two classes of Routing

Proactive Routing




RST-2210

11089_05_2005_c3

• Characteristics

Traditional distributed shortest-path protocols

Maintain routes between every host pair at all t imes

Based on periodic/triggered updates; high routingoverhead

• Trade offs

Always maintain routes

Litt le or no delay for route determinationConsume bandwidth to keep routes up-to-date

Maintain routes which may never be used

Reactive Routing




RST-2210

11089_05_2005_c3

• Characteristics

Determine route if and when needed

Source initiates route discovery

• Trade offs

Lower overhead since routes are determined on demandSignificant delay in route determination

Employ flooding (global search)

Control traffic may be bursty

Problems




RST-2210

11089_05_2005_c3

• Traditional routing metrics

Consider delay and throughput

Provide reliable reachability informationCannot enforce performance

• BGP can be especially challenging

Interior and exterior

Routing decisions based on path vector

Does not reveal information about path quality

• No visibil ity into situational requirements

Managing bandwidth monetary cost

Performing optimal load-distribution

MANet Working Group




RST-2210

11089_05_2005_c3

IETF WG Chartered to Bring into ExperimentalStatus a Set of Core Routing ProtocolsDesigned Specifically for MANet

• Ad hoc On-Demand Distance Vector (AODV)

• Optimized Link State Routing Protocol (OLSR)

• Topology Dissemination Based on Reverse-PathForwarding (TBRPF)

• Dynamic Source Routing Protocol (DSR)

• Dynamic MANet On-demand Routing Protocol(DYMO)

IETF: MANet




RST-2210

11089_05_2005_c3

• Charter Change

Primary focus is to address MANet use as a routing

technology extension

IPv4 and IPv6 compatible designs

Support for deployment of hybrid mesh infrastructures

• Autoconfiguration

• Multicast

• DYMO

• OSPF and CDS (Connected Dominating Set)



MANet FOR OSPFv3


RST-2210

11089_05_2005_c3

Why OSPF?




RST-2210

11089_05_2005_c3

• Why not a MANet WG protocol?No ‘optimal’ protocol

Performance depends on scenario and application(traffic pattern)

Testing studies—not validated in real-world deployments

Experimental RFCs

• AdvantagesProactive protocol

Integration with existing networks—devices can moreeasily migrate between a MANet and infrastructure network

Stabil ity of code that comes from years of deployment andtesting

Industry knowledge—Standard Protocol

Time to market

Why OSPF?




RST-2210

11089_05_2005_c3

• Baseline requirements

• Protocol

Needs to be well known and widely deployed

Will integrate well with wireline networks

Will be able to provide input into a wireline routingprotocols, and take new optimizations and capabilities asthey are proposed and deployed

Time to market is a consideration, a known working code

base is a plus

• OSPFv3 is the initial protocol

Why OSPF?




RST-2210

11089_05_2005_c3

h e l l o

t w o w a

y

d a t a b a

s e e x c h a n g e

t o p o l o g y c h a n g e

A

B

• Technical challenges

• Consider normal link state

operation:

Discover neighbors

Verify two way connectivity

Exchange link state databases

Flood new information

• We need to optimize at eachof these points

Cisco’s MANet Direction




RST-2210

11089_05_2005_c3

• Data Link Management System

• Incorporating L2 feedback into routing metrics

• Hybrid MANet interface

• Incremental hellos

• Optimized flooding

• Temporary l ink state database

• Address families

• Future work

General Radio/RF Link Assumptions




RST-2210

11089_05_2005_c3

• Radio detects and authenticates neighbors, reportsneighbor joining/loss to router

• Data link rates range from 80 Kbps to 29 Mbpseffective throughput

• Data link rates are dynamic, controlled by radio

• RF bandwidth changes, assigned radio channel(s),etc. are controlled by radio (independent of router)

Layer Two and One-Half




RST-2210

11089_05_2005_c3

• Definition of a new interface type for use bydynamic link layers (wireless)

• Based on the assumption that link layers can do amuch better job of assessing their status than canthe IP layer

• Algorithms exploiting inter-layer interaction in astable fashion

Alternatively,

• Routers calculate trends (Bayesian Matrix)

• Link-state moves from binary to probabilistic

• Topology morphs from flat to location and distance

Basic Router-Radio Architecture

R t R di I t f H Wh t A t t




RST-2210

11089_05_2005_c3

Router-Radio Interface Has What Amounts toVirtual Circuits:• Each routing neighbor has a

different data rate, managed by

the radio

• Therefore each neighbor musthave its own windowed protocol,and it must be windowed tocontrol rate

• There must be separate QoSdata structures per routingneighbor

• PPPoE sessions are establishedbetween the router and the

radio—a session for eachneighbor

• Radio detects and authenticatesneighbors, reports neighbor joining or leaving (LOS) to

routers

Radio

Raven System

Router

PPPOE with

extensions

Radio-Router Interface




RST-2210

11089_05_2005_c3

Router Radio Router Radio

PPPoE PPPoERF

PPP Data Session

Credit Based Flow Control




RST-2210

11089_05_2005_c3

• Design intent:

Push delayed data back into router queues

Queues provide QoS support (priority, isolation)

• Many similarities to TCP and LAPB

Receiver offers credit to sender Sender may send up to that amount

Credit in bytes (correlates to time)

• Not a reliable protocol

Retransmission at higher layers or application

Metrics Reported by Radio (L2 Feedback)




RST-2210

11089_05_2005_c3

• Current data rate

• Maximum data rate of link technology/policy

• Relative link factor

Intended as a radio-vendor-specif ic measure of l ink quality

• Resources

A node specific value that represents a percentage of resources remaining that effect continued operation (e.g.battery power)

• LatencyOne-way propagation delay in milliseconds

• Metrics are used by OSPF to calculate route cost

based on link quality

OSPF Link Local Signaling

OSPF Link Local Signaling: Backward Compatible




RST-2210

11089_05_2005_c3

OSPF Link-Local Signaling: Backward-CompatibleTechnique to Exchange Arbitrary Data on a Link

• A special data block is added at the end of OSPF

packets (or right after the authentication data block)

• The LLS data block may be attached to OSPF hellosand the DD packets

Options Field

0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 2 3-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+--+--+--+--+--+--+--+| | | | | | | | | | | | | | | |L|AF|DC| R| N|MC| E|V6|

-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+--+--+--+--+--+--+--+

Hybrid MANet Interface




RST-2210

11089_05_2005_c3

• Media (air) is multi-access, but may also behave aspoint-to-point (unidirectional antennas) with both

exposed and hidden nodes• OSPF: Interface is modeled as point-to-multipoint

Adjacencies treated as point-to-point

May take advantage of multi-access characteristics

Link metric can be set per neighbor

B A C

Incremental Hellos




RST-2210

11089_05_2005_c3

• Objective: reduce the control overhead

OSPF hello messages

PeriodicOverhead increases with rate

Overhead increases with the size and density of thenetwork (neighbor list included in every hello)

• OSPF: incremental State Hellos (don’t include fullneighbor list)

Two-way connectivity check

Incrementally update as neighbor state changesCompatible with graceful restart

Include information about capabilities (overlapping relays,willingness)

The Hello Process




RST-2210

11089_05_2005_c3

• OSPF discovers neighborsthrough multicast hello packets

• Each hello carries a list of neighbors we’ve heard from

Ensures two connectivity existsbefore building an adjacency

State increases with eachaddit ional neighbor

Constant state that’s not

interesting to other neighbors

• We need to reduce this state,both a new neighbor discovery

and normal operation

B

I can see BI can see B and CI can see B–DI can see B–EI can see B–FI can see B–GI can see B–HI can see B–I

I can see AI can see AI can see AI can see AI can see AI can see AI can see A

A

C D E F G H I

I can see A

The Hello Process




RST-2210

11089_05_2005_c3

• Replace the state with astate sequence

A small (32-bit) number

Indicates “ current hello state”

• Each time the sender changes state

Include new informationIncrement the state sequence

• Receiver can request a state

update to synchronize• Reduces the state carried in

hellos to the minimumpossible, while ensuring two

way connectivity

D

I c a n s e e B – D

s e q 3

C

I c a n

s e e B

a n d C

s e q 2

I c a n

s e e B

s e q 1

B

s e q 3

s e q 3

A

Incremental Hellos: Details




RST-2210

11089_05_2005_c3

• The original neighbor list should be kept for adjacency formation purposes

The I-bit (in the Options Field) indicates only newlydiscovered neighbors are listed in the list of neighbors

Options Field1 2

0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 2 3-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+--+--+--+--+--+--+--+

| | | | | | | | | | | | | | |I|L|AF|DC| R| N|MC| E|V6|-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+--+--+--+--+--+--+--+

Optimized Flooding




RST-2210

11089_05_2005_c3

• All new LSAs are flooded to all neighbors, exceptthe one from which the advertisement was

received—this behavior may result in duplication of information in some segments

• OSPF: Overlapping relays

Use knowledge of two-hop neighborhood, allowing moreintelligent flooding decisions and intelligent ACKingdecisions to be made

Active Overlapping Relays (ORCA) immediately flood, whilebackups delay the flooding until successful transmission isconfirmed

Algorithm used to select ORCAs is similar to OLSRs MPR

Selection

Standard OSPF Flooding




RST-2210

11089_05_2005_c3

Overlapping Relays




RST-2210

11089_05_2005_c3

Optimized Flooding

Only One Flood E F




RST-2210

11089_05_2005_c3

Only One Flood

Is Required Here

Routing Change

Flood

Flood

B C D

A

E F

• New reachability andtopology change

information is flooded toall adjacent neighbors

Several copies of an LSAmay reach a neighbor two

hops away

This is wasteful of availablebandwidth andprocessing power

• We need to optimizeflooding

Optimized Flooding

Fi d “ t h ”E F




RST-2210

11089_05_2005_c3

• Find common “ two hop”neighbors

Each neighbor advertises its

willingness to become an activeoverlapping relay

WILL_NEVER means not tochoose this neighbor

Heuristic based on the OLSRMPR selection algorithm

• Calculate minimum set of overlapping relays (ORCAs)

Pick one neighbor from eachgroup of neighbors with thesame “ two hop” neighbors

Group neighbors based on their

neighbor sets

B and C CanBoth See E

Willingness

Flood

B C D

Only DCan See F A

Optimized Flooding

E F

C l l t i i t f




RST-2210

11089_05_2005_c3

Declare Either Bor C and D Active as

OverlappingRelays

This Flood IsEliminated

A

Flood

Flood

B C D

Routing Change

• Calculate minimum set of overlapping relays

Pick one neighbor from eachgroup of neighbors with thesame “ two hop” neighbors

• Signal overlapping relaysto flood LSAs

• When A floods a new LSA

D refloods to F

B refloods to E

C doesn’t reflood at all

Optimized Flooding

C a tomaticall backs p B

E F




RST-2210

11089_05_2005_c3

• C automatically backs up B,since it’s not an ORCA

• When C receives the newLSA from A, it starts a“ pushback timer”

When B refloods, this timer

is stoppedIf the timer expires, C assumes Bis not operational, so it refloods

• This backup processprevents dynamic networkconditions from causingdatabase synchronizationissues and thus, packet loss

A

Flood

B C D

Flood

Routing Change

Set Pushback

Pushback/Expires

Intelligent Acknowledgements

Anything A Assume A Floods a New LSA




RST-2210

11089_05_2005_c3

Anything BTransmits, Both

A and C Receive

Anything CTransmits,Only B Receives

y gTransmits,Only B Receives

A

New LSA

ACK

Reflood

• When B receives this LSA,

it acknowledges its receipt

• B then refloods the LSA to C

• C sees the acknowledgementto A for the new LSA

• A sees the reflood to C by B

B

C

Intelligent Acknowledgements

Anything AWhy should B acknowledge




RST-2210

11089_05_2005_c3

Anything BTransmits, Both

A and C receive

Anything CTransmits,Only B Receives

y gTransmits,Only B Receives

• Why should B acknowledge A’s LSA if A is going to hear B’s reflood to C?

If B is reflooding the LSA, A canassume B received it correctly

• A uses B’s reflood to C as anacknowledgement

Cuts down on traffic on the wire

A

New LSA

Reflood B

C

Acknowledgements and Authentication

• OSPF neighbor adjacencies are vulnerable to resets




RST-2210

11089_05_2005_c3

g j

Attacker sends a spoofed hello

• Spoofed routing information

Conflicting

Full tables

• MD5 authentication

Unauthorized devices transmitting routing information

• OSPv3 does not support MD5 authentication,it uses IPSec

Overlapping Relays

• Enhance OSPF v3 to




RST-2210

11089_05_2005_c3

Enhance OSPF v3 tominimize the controldata needed to maintain

the mobile networkwithin an area

http://www.ietf.org/internet-drafts/draft-chandra-ospf-manet-ext-01.txt

• Provide dynamic linkfeedback to be used asrouting metrics for pathdecisions—Data Link

Management Systemhttp://www.ietf.org/internet-drafts/draft-bberry-pppoe-credit-01.txt

Overlapping Relays: Initial Results





http://www.ietf.org/internet-drafts/draft-bberry-pppoe-credit-01.txt












RST-2210

11089_05_2005_c3

Figure 20: OSPF’s traffic sent of OSPFv2 vs. OSPFv3 with overlappings relays

Temporary Link State Database

• Only LSAs from adjacent (> Exchange) neighbors




RST-2210

11089_05_2005_c3

• Only LSAs from adjacent (> Exchange) neighborsare kept in local Link State Database

• OSPF: Create a Temporary Link State Database tostore LSAs received from non-adjacent neighbors

Listen in promiscuous mode

Leverage these LSAs to reduce link state requests

Database Exchange

• Routers exchange their link state A




RST-2210

11089_05_2005_c3

Routers exchange their link statedatabase (LSDB) to become fullyadjacent

Verify two way connectivity

Exchange database descriptors

Request LSAs as needed

• MANet connections may notlast long

Reduce adjacency build time

Route across connection faster • Reduce bandwidth required

Hold information speculatively

Synchronize over a longer period of time

d e s c r i b e d a t a b a s e

d a t a , p l e a s e

d e s c r i b e d a t a b a s e

A

F U L L

B

Database Exchange

• Listen to new information while




RST-2210

11089_05_2005_c3

Listen to new information whilebuilding adjacency

Hold until synchronized

Reduces data exchanged indatabase exchange

• Send a summary of the

reachability informationInstant reachability through newneighbor

Schedule a full synchronization later

Neighbor state brought to FULL

• Synchronize databases

Out of band synchronization used

for OSPF graceful restart

v e r i f y t w o

w a y c o n n e c t i v i t y

( s e n d n e w d

a t a a s a v a i l a b l e )

( h o l d d a t a u n t i l s y n c h r o n i z

e d )

d

a t a b a s e s u m m

a r y

s c h e d u l

e s y n c h r o n i z a t i o n

s y n c h r o n

i z e

Supporting IPv4 in MANet Environments

• Implement draft-ietf-ospf-af-alt-xx.txt




RST-2210

11089_05_2005_c3

Implement draft ietf ospf af alt xx.txt

• Allocate “ Instance ID” number space in OSPFv3

packet header to signify Address Family (AF)0-31 means IPv6, 64-95 means IPv4

• Need to run two routing instances, if both IPv6 and

IPv4 Address Family needs to be supported

• Each AF will establish different adjacency, havedifferent link state database and compute differentshortest path tree

OSPFv3 Address Family

IPv4 Legacy Router

Dual Stack IPv4/IPv6

IPv6/IPv4 cloud:Routing Protocol : OPSFv3

IPv4 and IPv6 dual stack enabled




RST-2210

11089_05_2005_c3

IPv4 Island

OSPFv2IPv4 Island

OSPFv2

IPv6/IPV4 MANetOSPFv3 Area 0

Dual Stack

MANet Enabled Router

IPv4 Networks

IPv4 networks

IPv4 and IPv6 dual stack enabled

IPv4 packet thru IPv4 forwarding

IPv6 packet thru IPv6 forwarding

MANet Optimized

Routes redistributed from OSPFv2

IPv4 cloud:

Routing Protocol : OSPFv2

Routes redistributed from OSPFv3

IPv4 Data

Forwarding

IPv4 RoutesRedistribution

IPv4/IPv6

networks

Future Work: Scalability

• Requirements




RST-2210

11089_05_2005_c3

q

Reduce the amount of information on the media and to

be stored• Solutions

Edge Detection: dynamically limit the size of the SPT

Smart Peering: limit the number of peers intelligently

Database Synchronization Enhancements: reduce theoverhead when synchronizing with nodes that may have

redundant information

Future Work: Flooding Boundaries

• In link state protocols, routers floodf f

A B




RST-2210

11089_05_2005_c3

information about the state of their links to all other routers

• All the routers receiving the floodedinformation are said to be in thesame flooding domain

• We summarize topology information

into reachability information at aflooding domain border

• OSPF has manually configuredflooding boundaries

• In MANet networks, floodingboundaries are not well defined

Based on the distance from the originator of the routing information?

Based on edges detected in the network?

C

G

D

E F

Border

Connected to E,F, and 10.1.1.0/24

Connectedto D and G

Connectedto D and G

10.1.1.0/24

Future Work: Contiguous Mobility

• Situation




RST-2210

11089_05_2005_c3

While moving in the network, a node may find itself in adifferent domain

• Solutions

Enhanced Neighbor Discovery: establish adjacencies with

nodes in other domains Automatic Clustering: change local characteristics andseamlessly join new domains; combination of edgedetection and auto-configuration (find friends amongst

acquaintances)

Future Work: Auto-Configuration

• Situation




RST-2210

11089_05_2005_c3

Through noncontiguous mobility, a node may need toreconfigure (or receive initial configuration parameters)itself

• Solutions

Discovery and Selection of DHCP Server : selection of “ closest” server (to provide locally significantconfiguration)

Dynamic DNS

Switchover to OSPF-Lite/Other protocols



MANet COMPLEMENTS AND FUTURES


RST-2210

11089_05_2005_c3

MANet: Complements and Futures

• Quality of Service




RST-2210

11089_05_2005_c3

• Multi-Topology Routing

• Application-Oriented Networks

• Cognitive Radio

• Optimized Edge Routing

• Mobile BGP

• Other

Quality of Service (QoS)

• What is “ Quality”




RST-2210

11089_05_2005_c3

• Statistical services like the Internet provide

statistical guarantees

• Sometimes, people need non-statistical guarantees:

A certain amount of bandwidth between here and there,

no excuses

Move a file across the network in a fixed amount of t ime

Deliver a message to a set of people within a fixed period

of time

• Currently, there is no standard policy in place

QoS for MANet

• Multilayer Problem Application




RST-2210

11089_05_2005_c3

Layers must work together

• As mobility increases,

Latency increases

Security and QoS decrease

• Three types of nodes

Ingress, Egress, Interior

Multiple roles

Physical

Link

Network

pp

Multi-Topology Routing Concepts

• Goalf f ff




RST-2210

11089_05_2005_c3

To influence the path that certain types of traffic would takebased upon attributes such as DSCP and Application Type

Traffic Separation across network infrastructure

• Creation of multiple topologiesLogical path that traffic will take across the given network

Each topology wil l route a subset of the traffic as definedby the classification criteria

• Topology mappingDetermine which traffic is subject to topology-specif ic

forwarding• MTR provides path-based service differentiation

within a single domainQoS provides per-hop service differentiation

Multi-Topology Routing Traffic Paths




RST-2210

11089_05_2005_c3

Traffic is marked

at the network edge;DSCP value is usedto assign traffic to

a topology

As traffic traverses the networkit is constrained to its owncolored topology

Critical Data TopologyVoice TopologyBase Topology

Application-Oriented Networks

Lower Cost• Lower capital and

operational costs• Application security

(encryptionM




RST-2210

11089_05_2005_c3

andComplexity

operational costs

• Simpler, integratedfunct ionality andmanagement

(encryption,authentication, digitalsignatures, certificate

management, PKI)

MoreSecurity

Greater Reliability

MoreVisibility

• Guaranteed delivery

• HTTP reliablemessaging

• Workload balancingand failover

• Real-time monitoringof events

• Logging for auditing,non-repudiation, andanalysis

Greater

IntegrationFlexibility

• Protocol translation

(HTTP, SMTP, FTP,SOAP, MIME, FIX)

• Service virtualization

Higher Performance

• Hardwareacceleration

• Message loadbalancing, caching,compression

AONS: Information in Action

HIGH AVAILABILITY,

GUARANTEED DELIVERY

P t




RST-2210

11089_05_2005_c3

AONS

AONS

AONS

AONS

AONS

AONS

AONS

AONS

AONS

AONS

AONS

IMPORTANT

INFORMATION

Analyzed

Stored

Alerted

Message Sent:Product Created

and Shipped

Document RoutedBased upon

Contents

Partners

FieldOrganizations

Manufacturing


APPLICATION LEVEL SECURITY,

SERVICE VIRTUALIZATION

HIGH AVAILABILITY,

GUARANTEED DELIVERY

P t




RST-2210

11089_05_2005_c3

Partners

FieldOrganizations

Manufacturing

AONS

AONS

AONS

AONS

AONS

AONS

AONS

AONS

AONS

AONS

AONS

IMPORTANT

INFORMATIONMessage Sent:Partner Receives

Shipment

Translated for the Format of theERP System

Original Stored,

Queries Processed,Sales Alert Created

NewInformation

Available


APPLICATION LEVEL SECURITY,

SERVICE VIRTUALIZATION

HIGH AVAILABILITY,

GUARANTEED DELIVERY

QUALITY OF SERVICE, LOGGING,

REAL TIME ANALYSIS

Partners

Message




RST-2210

11089_05_2005_c3

Partners

FieldOrganizations

Manufacturing

AONS AONS

AONS AONS

AONS AONS

AONS AONS

AONS AONS

AONS AONS

AONS AONS

AONS AONS

AONS AONS

AONS AONS

AONS AONS

IMPORTANT

INFORMATION

MessagesProcessed,

UpdatedClient

Information

Sales Report

ResultsIntegrateinto a NewDocument

UpdatedInformation

MessageDistributed

gDistributed

MessageDistributed

Building Blocks of AON Solutions

Native Understanding of Application Messageswith the Ability to Apply a Range of Services




RST-2210

11089_05_2005_c3

B2BB2B OthersOthersRFIDRFID Application Application

SecuritySecurityServicesServicesOrientedOriented

Activity Activity

MonitoringMonitoring

AONs:Intell igent Network Embedded Servicesfor Secure Application Communications

(A Network for Intelligent Applications)

Cognitive Radio: SDR Fundamentals

• DSPs provide significant programmability




RST-2210

11089_05_2005_c3

• All modulation, cryptography, protocols, and

source coding (data, voice, video) are establishedvia software

• Many types of modulation can be achieved over a

broad range of frequencies; thus an SDR is capableof servicing more than one class of service*

• Field serviceable—as requirements change,

upgrades/modifications are relatively easy

Cognitive Radio Means “ Smart” and “ Alert”

WAN(Wide Area Network)

• The Radio Knows:Where it isA il bl i




RST-2210

11089_05_2005_c3

LAN(Local Area Network)

MAN

(Metropolitan Area Network)

PAN(Personal Area

Network)

PAN LAN MAN WAN

Speed < 1 Mbps2 to 54+

Mbps22+ Mbps

10 to 384Kbps

Range 10 feet 200 feet 0.5 – 2 miles Long

Available servicesUser’s interests

Needs and futurelikelihood of needs

• Learns andrecognizes usagepatterns

• Applies model-based reasoningabout needs,content, andenvironment

Cisco Optimized Edge Routing

• Automatic policy-based distribution of traffic

P fi h bil it d l k t l




RST-2210

11089_05_2005_c3

Prefixes: reachabil ity, delay, packet loss

Links: load, distribution, monetary cost

• Provides functionality unavailable with traditionalrouting metrics

• Optimized exit path based upon

Latency

Packet Loss

Reachability

Cost

Load Sharing

What Is Cisco OER?

Automatic Outbound Route Optimization for Multi-Homed Enterprises by Selecting “ Optimal”Exit According to Performance Cost and Load




RST-2210

11089_05_2005_c3

Customer

Access

SP A SP B

SP D SP ESP C

BR1

BR2

BR3

CR1

CR2eBGP

SP F

iBGPand/or OSPF,EIGRP,

etc.

OER Master

Server(s)

SLA A

SLA B

SLA C

Cisco IOS®

or Appliance

Exit According to Performance, Cost, and LoadDistribution Policy*

CONTENT PROVIDER TRANSIT NETWORK CONSUMER

* OER 1.0

Mobile BGP: Agenda

• Why Mobile BGP?

T h i l Ch ll




RST-2210

11089_05_2005_c3

• Technical Challenges

• Promiscuous eBGP

• iBGP

• Routing Through the AS

• Reducing BGPs Weight

Why Mobile BGP?

• Administrative Separation

In wired networks BGP is used to separate administrative




RST-2210

11089_05_2005_c3

In wired networks, BGP is used to separate administrativedomains

Any topology under one administration is considered anautonomous system (AS)

AS are interconnected using BGP

In MANet networks, we anticipate the same need

• Scaling Properties

In wired networks, BGP is used to scale the routing domain

Routes are carried through autonomous systems usingBGP, rather than into the AS

In MANet networks, we anticipate the same need

Technical Challenges

• BGP splits internetworking autonomous systemsinto two logical categories




RST-2210

11089_05_2005_c3

g g

• eBGPHandles connections to other autonomous systems

Heavy on policy

Uses the AS Path to prevent routing loops (path vector routing protocol)

• iBGP

Carries routes through the autonomous system

Not designed to prevent routing loops

• We need to address both cases

Technical Challenges

• Neighbor discovery—dynamic peers

BGP requires manually configured neighbors




RST-2210

11089_05_2005_c3

BGP requires manually configured neighbors

• Routing through the ASiBGP sessions are multihop

If a router along the path doesn’t know how to reach the

destination, routing fails

• Heavy packet formats

BGP packets carry a lot of policy and other information

Promiscuous eBGP

router bgp 65000

neighbor 10.1.1.1 remote-as 65001

router bgp 65000

neighbor 10.1.1.1 remote-as 65001• BGP requires that neighborsbe manually configured




RST-2210

11089_05_2005_c3

router bgp 65000


router bgp 65000


I P v 6 n e i g h b o r d i s c o

v e r y

p

e r i o d i c p o l l i n g

y g

This isn’t conducive to MANetenvironments

The average time we canexpect a neighbor to be within

reach is only about one-half

• Neighbors can be found

IPv6 neighbor discovery

Periodic polling on a wellknown multicast address

Promiscuous eBGP

router bgp 65000

neighbor 10.1.1.1 remote-as 65001• Once a new peer is discovered,Promiscuous eBGP* can setupthe peering relationship

router bgp 65000


p




RST-2210

11089_05_2005_c3

router bgp 65000


the peering relationship

Accept peering relationshipswithout configuration

Effectively the same asautomatically building a neighbor configuration on the fly from

discovered information

Normal BGP state machine

• Security is critical in this

processWe have to rely on outside securitymechanisms, since BGP doesn’thave configured peers

Central authority, pre-shared

router bgp 65000


I P v 6 n e i g h b o r d i s c o

v e r y

p

e r i o d i c p o l l i n g

b u i l d p e e r i n g r e l a t i o

n s h i p

rely on outsidesecurity

mechanisms

iBGP

iBGP Is More Difficult than eBGP

• Peer relationships are normally multihop




RST-2210

11089_05_2005_c3

Peer relationships are normally multihop

• We need to detect possible iBGP peers

• We need to detect autonomous system edges

• We need to transmit BGP routing informationoptimally through the AS without route reflectors or other aids

Routing Through the AS

• Routing Flow

A advertises 10.1.1.0/24 to BeBGP

A




RST-2210

11089_05_2005_c3

B advertises 10.1.1.0/24 to DD advertises 10.1.1.0/24 to E

• Traffic Flow

E sends a packet to 10.1.1.1 via D

D sends it to C, its next hoptowards B

C has no information about10.1.1.0/24, so it drops the packet

iBGP

eBGP

B

C

D

E

10.1.1.0/24

10.1.1.1

Routing Through the AS

• The wired resolution

Make C a route reflector, so it hasf ll ti i f ti

eBGP A




RST-2210

11089_05_2005_c3

full routing information

Synchronize the iBGP and IGProuting tables throughredistribution

Make the AS a single hop deep

• Possible MANet solutions

Tunnel the packet from D to B

Conditionally inject information

about 10.1.1.0/24 into the IGPBoth of these solutions may beuseful in different situations, soboth may be considered

iBGP

eBGP

C

E

10.1.1.0/24

10.1.1.1

route reflector

synchronizeiBGP and IGP

B

D

BGP’s Heavy Packet

• BGP carries a large number of attributes

Local preference




RST-2210

11089_05_2005_c3

MEDOrigination code

AS Path

Communities

• BGP requires a lot of processing to run thebestpath algorithm

Somewhere between a 12–18 step process

• We would like to reduce BGP’s weight in both of these areas

Reducing BGP’s Weight

• Cost Community

• At the inbound edge, convert to a single communityt i i ll t i




RST-2210

11089_05_2005_c3

containing all metricsMED, Local Preference, AS Path length compiled into asingle number carried in a community

• Drop attributes through the iBGP cloud

Drop MED, Local Preference, and OriginKeep AS Path and communities

• Run bestpath as a single number compare amongcost communities

• At the outbound edge, convert back to BGPattributes

• Contrarian View: Add “ delay”



SUMMARY

RST-2210


Mobility Review

• There is more to mobility than just moving

Portability roaming and handoff Fire




RST-2210

11089_05_2005_c3

Portability, roaming, and handoff

• Layer 3 mobility is key in largenetworks

• IPv6 mobility is a stepping stone

to ad hoc, autonomic, andpervasive networking1 bill ion Mobile Phone Users(lack of IPv4 addresses)

Better adaptation than IPv4Mobility-related standardizationEven 4G is considering IPv6

• No global architecture

Fire

Fighters

Convoys

Sensors L i n k

I n s t a b i l i t y

Connectedness

Layer Three Mobility Review

• Seamless access whileroaming across variousnetworks around the

• Transparent to apps

• No IP address change*

• Allows push




RST-2210

11089_05_2005_c3

Internet• Allows push

InternetDial/DSL

Hotel

HomeConferences

Meeting Rooms

ISP Access

Through the

Internet

Cellular or Mobile

Commuting

LANs andVLANs

Mobile Ad Hoc Networks Review

• Mobile Ad hoc Networks(MANet) is a broad topic

n c y




RST-2210

11089_05_2005_c3

• Every host is a router • Closely tied to Layer 2

• There may or may not be a

structured network

• Scale and administrativeissues can be major factors

• Large number of nodesmoving quickly is notpossible today

H a n d

o v e r F r e q u

e n

Number of Nodes

Service Delivery Review

• Today’s Network

Three classes of devices

Network devices (routers

• Pervasive Network

Two classes of physical devices

Network devices (routers,




RST-2210

11089_05_2005_c3

and switches and other

infrastructure) Application servers (Web,mail, storage, computing,print, etc.)

Clients (personalcomputers, phones,PDA’s, sensors, etc.)

switches, servers, etc.) that

provide routing capabilit iesin addition to services suchas Web, mail, storage,computing, print, etc.

Clients (personal computers,

phones, PDA’s, dedicateddevices, sensors, etc.) thatprovide a user interface andparticipate in rout ing of traffic for other clients and

embedded devicesPlus “ virtual devices” enabledby the advent of virtualmachines layered on top of thetwo classes of physical devices

Putting it All Together

Fixed BackboneMobile IP




RST-2210

11089_05_2005_c3

Mobile access network

Mobile Backbone M o b i l e

a c c

e s s

n e t w o r k M

o b i l e

u s e

r

IP Ad Hoc

NeMo

Radio Ad Hoc

Mesh

M o b i l e

B a c k b o n

e

Military Tactical Scenario

Fixed

Infrastructure




RST-2210

11089_05_2005_c3

Mobile BackboneGroup Mobility

Micro-Mobility

Main

Command

Posts

Group

Mobility

Deployable

Tactical Internet

Mobile

Tactical Internet

Highly Mobile

Tactical Internet

Other Issues to Be Addressed

• Who sets Policy, how is it equalized,and how is it enforced?

Religion

PoliticalIssues at Many Layers




RST-2210

11089_05_2005_c3

and how is it enforced?

• Is there a difference in applicationqueuing, addressing, or other resources?

• Can a transport provide predictablethroughput? How does distanceaffect this?

• Can the network layer provide guarantees

or security to the packets?

• Will the physical network fight theupper layers?

Physical

Link

Network

Transport

Session

Presentation

Application

Summary

• MANet is immersed in and emerging from research

• Continuous development will be necessary to meetincreasing and changing mobil it demands




RST-2210

11089_05_2005_c3

increasing and changing mobil ity demands

• Objective is to provide a “ wired-compatible” and“ wired-like” solution

• Ultimately, it is all about virtualizationService and resources

Create an infrastructure that scales dynamically

Adapts to changing usage patterns without modification

Network resources automatically optimize themselves

• Virtualization will help yield pervasive networks

References

• MANet Characteristics: http://www.ietf.org/rfc/rfc2501.txt

• AODV: http://www.ietf.org/rfc/rfc3561.txt

• OLSR: http://www.ietf.org/rfc/rfc3626.txt










RST-2210

11089_05_2005_c3

p g

• TBRPF: http://www.ietf.org/rfc/rfc3684.txt• DSR: http://www.ietf.org/internet-drafts/draft-ietf-manet-

dsr-09.txt

• DYMO: http://www.ietf.org/internet-drafts/draft-ietf-manet-dymo-00.txt

• OSPFv3 Enhancements: http://www.ietf.org/internet-drafts/draft-chandra-ospf-manet-ext-00.txt

• PPPoE: http://www.ietf.org/internet-drafts/draft-bberry-pppoe-credit-01.txt

• IPv6:http://www.cisco.com/application/pdf/en/us/guest/products/iosswrel/c1127/cdccont_0900aecd8018e369.pdf

Associated Sessions

• AGG-1010: Introduction to Wireless LAN Technology

• AGG-2012: Design and Deploying of Outdoor



http://www.ietf.org/internet-drafts/draft-ietf-manet-dsr-09.txt



http://www.ietf.org/internet-drafts/draft-ietf-manet-dymo-00.txt










http://www.cisco.com/application/pdf/en/us/guest/products/iosswrel/c1127/cdccont_0900aecd8018e369.pdf


















RST-2210

11089_05_2005_c3

Wireless LAN/Bridging Networks• RST-1210: Introduction to IPv6

• RST-2213: IP Mobility

• RST-3261: Introduction to Multi-Topology Routing

Recommended Reading

• Mobile IP Technology and Applications [158705132X]

http://www.amazon.com/gp/product/images/158705132X/ref=dp_primary-product-display_0/103-8834844-3720649?%5Fencoding=UTF8&n=507846&s=books




RST-2210

11089_05_2005_c3

• Wireless Networks First-Step[1587201119]

• Routing TCP/IP [1578700418]

• Ad Hoc Networking [0201309769]

• BGP Design and Implementation[1587051095]

http://www.amazon.com/exec/obidos/tg/detail/-/1587201119/qid=1117220608/sr=8-1/ref=sr_8_xs_ap_i1_xgl14/103-8834844-3720649?v=glance&s=books&n=507846

http://www.amazon.com/gp/product/images/158705132X/ref=dp_primary-product-display_0/103-8834844-3720649?%5Fencoding=UTF8&n=507846&s=books



Q and A


RST-2210

11089_05_2005_c3

Complete Your Online Session Evaluation!

• Win fabulous prizes! Give us your feedback!

• Receive 10 Passport Points for eachsession evaluation you fill out




RST-2210

11089_05_2005_c3 158158158

• Go to the Internet stations locatedthroughout the Convention Center

• Winners wil l be posted on the Internetstations and digital plasma screens

• Drawings will be held in theWorld of Solutions

Monday, June 20 at 8:45 p.m.

Tuesday, June 21 at 8:15 p.m.

Wednesday, June 22 at 8:15 p.m.

Thursday, June 23 at 1:30 p.m.



RST-2210




MANet ROUTING PROTOCOLS


RST-2210

11089_05_2005_c3

AODV

• Based on Destination-Sequenced Distance-Vector (DSDV) routing algorithm




RST-2210

11089_05_2005_c3

• Routes are discovered as-needed by broadcastinga route-request (RREQ) through the network, andwaiting on a unicast route-reply (RREP)

• Routes are maintained “ as long as needed”• Route errors are signaled by a Route Error (RERR)

message to all effected destinations

Ad Hoc Distance Vector (AODV)

• A route between two nodesis found by sending an routerequest to a locality

C




RST-2210

11089_05_2005_c3

Initial locality small,grows with failure

After that, a little larger than the locality targetlast found in

• Route response sent

By target if necessary

By neighboringrouting node if possible to “ join”existing route

• Network stores the route

d

j k l

ihg

f

A

e

B

AODV (Cont.)

• Each route is to a router

• Each route advertisement has a sequence number




RST-2210

11089_05_2005_c3

Originator bumps sequence number on new informationOthers bump only when withdrawing failed route

• Effect: we always know relative order of information

No count to infinity

No looping routes

Optimized Link State Routing (OLSR)

• Systems trade

Some form routingbackbone




RST-2210

11089_05_2005_c3

Some act as “ hosts”

• As devices move

Topologicalrelationships change

Routes change

Backbone shape andcomposition changes

Dynamic Source Routing (DSR)

• Reactive protocol

Establishes the route as needed K




RST-2210

11089_05_2005_c3

• Client software needed tostore routes

• Routers are just

forwarding devices• Peer to peer routing

B

H I J

GFE

DC

Detected Source Route:

(A, B, F, J, K)

Topology Broadcast Based onReverse-Path Forwarding (TBRPF)

• OSPF variant

• Proactive routing




RST-2210

11089_05_2005_c3

Establish routes not on demand

• Topology updates for a set of routers

• Peer to peer routing

• Does not build a large tree but a subset

Otherwise known as “ Fisheye Lens” routing

Dynamic MANet On-Demand RoutingProtocol (DYMO)

• Unicast

• IPv4 and IPv6




RST-2210

11089_05_2005_c3

• Generic element and packet handling

• Avoid expiring good routes

Update reverse route lifetime on data reception

Update forward route lifetime on data transmission

routed mobile ad hoc networks

Documents