[PPT] - Wireless Mesh Networks MiNEMA Winter School 2009 Dr Johnathan PowerPoint Presentation

SLIDE 1

MiNEMA Winter School 2009 Dr Johnathan Ishmael ishmael@comp.lancs.ac.uk

Wireless Mesh Networks

SLIDE 2

Talk Overview

 Middleware for Network Eccentric and Mobile

Applications Garbinato, Benoît; Miranda, Hugo; Rodrigues, Luís (Eds.) 2009, Approx. 465 p., Hardcover ISBN: 978-3-540-89706-4

 Chapter 7: Wireless Mesh Networks

SLIDE 3

Talk Overview

 Chapter 7: Wireless Mesh Networks

 Defining

 What & Why

 Building  Testing  Additional Research Challenges

SLIDE 4

What is a Wireless Mesh Network?

Defining Wireless Mesh Networks

SLIDE 5

What is a WMN?

 A Wireless Mesh Network is a form of multi-hop

wireless network providing a low cost, distributed network infrastructure

SLIDE 6

Why Wireless Mesh Networks? (1)

 Provide connectivity to places without existing

infrastructure

 Ubiquitous Access  Digital Divide

SLIDE 7

Why Wireless Mesh Networks? (2)

 Goals of WMNs:

 Provide a network which can deliver high bandwidth-

low latency services

 Provide Ubiquitous connectivity  Remain cost effective in low density / sparsely

populated areas

SLIDE 8

Why Wireless Mesh Networks? (3)

 Other uses:

 Specialised Scenarios (Security, Emergency/Disaster)

SLIDE 9

What is a WMN?

 A Wireless Mesh Network is a form of multi-hop

wireless network providing a low cost, distributed network infrastructure

SLIDE 10

Multi-hop Wireless Networks

Hybrid Wireless Network Wireless Ad hoc Networks Wireless Mesh Networks Wireless Sensor Networks

SLIDE 11

Internet connection (Gateway) Wireless Network Client

Wireless coverage area Access Point Connected to the Internet

Traditional Wireless Access Point

SLIDE 12

 As Ad-hoc nodes are

added coverage is extended

Wireless Network Client

Ad-hoc Wireless Network

SLIDE 13

 As Ad-hoc nodes are

added coverage is extended

Internet connection (Gateway) Dedicated Mesh Node Wireless Network Client

Wireless Mesh Network

SLIDE 14

Ad-hoc Vs Mesh Vs Wired

Ad-hoc Wireless Mesh Wired Wireless Communication Wireless Communication Wired Communication Dynamic Topology Static Topology Static Topology Infrastructure-less Infrastructure based Infrastructure based Battery Powered Usually Mains Powered Mains Powered Temporary in Nature Usually Permanent Permanent No services Decentralised Services Centralised Services

SLIDE 15

Low Cost?

 Ease & Simplicity:

 Rapid, scalable deployment without the need for existing

infrastructure

 Zero administration:

 Self-healing: Continuous re-configuration as routes are

broken

 Price:

 Low install cost, low administration cost, low cost to end

user

SLIDE 16

Distributed Network Infrastructure?

 In order to function, Wireless Mesh Networks

require infrastructure type services such as:

 DNS/DHCP  Access Control  Traffic Shaping  Gateway Selection

 In a wired network this would be centralised  Yet, without a single point of authority or

infrastructure

SLIDE 17

What is a WMN?

 A Wireless Mesh Network is a form of multi-hop

wireless network providing a low cost, distributed network infrastructure

SLIDE 18

What components are required to make a Wireless Mesh Network?

Building Wireless Mesh Networks

SLIDE 19

What Components make a WMN?

 The components of a

Wireless Mesh Network span a large range of the OSI model

 No single component

really “makes” a Wireless Mesh Network

 A concise overview of

each layer...

Application Presentation Session Transport Network Data Link Physical

SLIDE 20

Physical Layer (1)

 Choosing the right radio technology

 Power usage  Range  Data rate

 Antennas

 What Antennas? – Directional?  Smart Antennas  Multiple Antennas (MIMO)

SLIDE 21

Physical Layer (2)

 Greater throughput, greater resilience, lower latency &

less power

 Choosing the correct Physical Layer components

depends upon the Application of the network

 IPTV solution compared to Disaster Scenario

 Research Topics  MIMO-based physical layers for WMNs  Higher physical layer capacity  Ability to operate in high interference levels  Ability to operate in flexible radio bands (Cognitive Radio)

SLIDE 22

Medium Access Control (MAC) Layer (1)

 Wireless MAC layer manages communications

between devices in a shared communication channel

 Correct operation vital for WMNs due to large

number of nodes and the multi-hop nature

SLIDE 23

Medium Access Control (MAC) Layer (2)

 What does the MAC Layer provide?

 Collision Avoidance  Energy Conservation  Interference Resistance  Rate Adaption

SLIDE 24

Medium Access Control (MAC) Layer (3)

Time New Contention Window (8 slots)

D I F S

RTS

S I F S

Ack

D I F S

New Contention Window (5 slots) New Contention Window (9 slots)

D I F S D I F S

Old Contention Window (1 Slot) Defer Access (Based on RTS) Defer Access Station 1 Station 2 Station 3

CTS

S I F S

Data

S I F S

Station 4 (Hidden from Station 1) Defer Access (Based on CTS)

RTS

Defer Access Defer Access

SLIDE 25

Medium Access Control (MAC) Layer (4)

 Research Topics

 MAC incompatibilities with higher level protocols  Dual-Multi Channel MAC

SLIDE 26

Network Layer

 Provide functional and procedural means to transfer

variable length data from a source to a destination

 Routing  Addressing  Quality of Service

SLIDE 27

 As Ad-hoc nodes are

added coverage is extended

Internet connection (Gateway) Dedicated Mesh Node Wireless Network Client

Routing Metrics

SLIDE 28

Routing & Routing Metrics (1)

 Selects the “best” available path between two hosts

n a network

 Routing protocols determine a cost for each

potential path available and use the lowest cost path

 The cost for each path is based on a routing metric

SLIDE 29

Routing & Routing Metrics (2)

 What metrics?

 Stable (Uptime)  Most throughput (Bandwidth)  Lowest latency (Round Trip Time)  Cost effective (Access Network Cost)  High power availability (Battery Power)

 Routing Protocols based on application and network

environment

SLIDE 30

Routing & Routing Metrics (3)

 What makes a good routing protocol?

 Ensuring route stability (no route flapping)  Lowest cost paths actually have good performance

 AODV by default uses minimum hop count

 Efficient (low cost) algorithms for calculating minimum

cost paths

 Loop free

SLIDE 31

Routing & Routing Metrics (4)

 Protocols divided into (at least) three camps:  Proactive (Table-driven) e.g.:

 DSDV (Destination-Sequenced Distance Vector)  OLSR (Optimized Link State Routing Protocol)

 Reactive (On-demand) e.g.:

 AODV (Ad-hoc on-demand Distance Vector)  DSR (Dynamic Source Routing)

 Hybrid e.g.:

 ZRP (Zone Routing Protocol)  Some more examples of routing protocols.....

SLIDE 32

Routing & Routing Metrics (5)



And.....



AWDS



Babel



CGSR



DFR



DBF



DSDV



Guesswork



HSR



IARP



LCA



MMRP



OLSR



And.....



TBRPF



WAR



WRP



CHAMP



GB



IERP



LBR



LMR



LQSR



LUNAR



MOR



MPRDV



And.....



B.A.T.M.A.N.



QuaSAR



RDMAR



SrcRR



SSR



PLBR



VRR



TORA



ARPAM



HRPLS



HSLS



OORP



ZRP



And.....



CBRP



CEDAR



DART



DDR



FSR



GSR



HARP



HSR



HSR



LANMAR



ATR



ALARM



And.....



BGR



DREAM



GLS



LAR



GPSAL



ZHL



GPSR



GFG



SFT



FACE



ISAIAH



PARO



EADSR



And.....



ABAM



ADMR



AMRIS



AMROUTE



AQM



BEMRP



CAMP



CBM



DCMP



DDM



DSR-MB



FGMP And many more! Source: http://en.wikipedia.org/wiki/Ad_hoc_routing_protocol_list

SLIDE 33

Routing & Routing Metrics (6)

 Research Topics

 Multiple Radio Cards  (Yet more) New metrics  Cross-layer design

 Dynamic Channel Assignment

SLIDE 34

Transport Layer (1)

 Responsible for encapsulating application data in

datagrams (or vice-versa)

 Services include:  Same Order Delivery  Error correction / Segment retransmission  Flow Control  Congestion Avoidance  Ports  Can be connection-orientated (TCP) or connectionless (UDP)  We‟ll focus on TCP...

SLIDE 35

Transport Layer (2)

Link Rate (Mbps) Optimal UDP Throughput (Mbps) TCP Throughput (Mbps) ACK Overhead % 6 5.3 4.6 13.2 12 10.3 8.4 18.4 18 14.9 12.0 19.4 24 19.3 16.4 14.5 36 26.8 22.6 15.7 48 32.9 26.9 18.2

 Impact of TCP ACKs traversing multiple hops on the

network

 TCP performs poorly on WMNs

SLIDE 36

Transport Layer (3)

Channel Condition TCP Throughput (Mbps) Optimal UDP Throughput (Mbps) TCP Underutilisation(%) Very bad 0.08 0.87 90.8 Bad 3.37 6.07 44.5 Average 14.5 18.6 22.0 Very Good 26.9 32.9 18.2

 Poor congestion control, TCP can‟t differentiate the

different between congestion and poor link quality

Link rate of 48Mbps

SLIDE 37

Transport Layer (4)

 Within a WMN transport layer goals are to

provide:

 End-to-end reliability  High throughput  Capability to handle network asymmetry  Handle dynamic nature of network

SLIDE 38

Transport Layer (5)

 Research Topics:

 Improve performance (cross-layer)  ...without deviating from TCP... or  Do we need a new transport layer protocol for WMNs?

 Awareness of link quality, link failure and dynamic nature of

the network

SLIDE 39

Application Layer

 Primary function of WMNs is to provide access to

the Internet. Others include:

 Distributed backup  VoIP

 Service discovery mechanisms are required  Should applications be aware they‟re running on a

WMN and behave differently?

 P2P?

SLIDE 40

IEEE 802.11s Standard (1)

 An IEEE 802.11 amendment for mesh networking  Aims to create a Layer 2 ad-hoc network by

 Defining an architecture that supports

broadcast/multicast and unicast delivery

 Provides radio-aware metrics over self-configuring

multi-hop technologies

SLIDE 41

IEEE 802.11s Standard (2)

 802.11s defines a device as a Mesh Station (Mesh

STA)

 Multiple Mesh STAs form links with one another,

which any routing protocol can be utilised

 Default is Hybrid Wireless Mesh Protocol (AODV

derivative)

 Mesh STAs can connect to a 802.11portal which

provides gateway access to non-802.11 networks

 But... doesn‟t fit my original definition of a WMN

SLIDE 42

Results from Deployed Testbeds

Testing Wireless Mesh Networks

SLIDE 43

Wray Wireless Mesh Network

 Johnathan Ishmael, Sara Bury, Dimitrios Pezaros, Nicholas Race,

"Deploying Rural Community Wireless Mesh Networks," IEEE Internet Computing, vol. 12, no. 4, pp. 22-29, Jul/Aug, 2008

 Wray Community Wireless Mesh Network  Technical Challenges  Research Directions  http://wray.lancs.ac.uk

SLIDE 44

SLIDE 45

What is a WMN?

 A Wireless Mesh Network is a form of multi-hop

wireless network providing a low cost, distributed network infrastructure

SLIDE 46

Agreement with Wray Community

 University offered to:

 Provide free backhaul to community  Install LocustWorld WMN equipment & antennas at the village

school

 Supply additional Mesh equipment for use throughout the village  Ensure connectivity to the village on a “best effort” basis

 University is not focused on:

 Solving end user connection issues  “My PC doesn’t work”, etc!  People Management/Access control of Mesh network

SLIDE 47

 Participated in Deployment of the Wray Wireless Mesh

Network:

 Initial configuring of hardware  Establishing a community-based team for day-to-day mesh

management

 Social Challenges  Large learning curve for local community, both in using and

maintaining the network

 Users‟ expectations were considerably high  Expected wireless adaptors to function from anywhere  Lack of understanding of underlying technology  Technical Challenges  Aerials incorrectly deployed  Software not operating to the „ethos‟ of Wireless Mesh Networks

(not self-managing)

Wray WMN – Phase 1:

SLIDE 48

Wray WMN – Phase 2:

Raised existing aerials,

shortened cables

Expanded number of

Mesh boxes

Fitted aerials to clients

with poor connectivity

Community-based team

and users gaining increased confidence in I.T.

School Community Hall Public House The Barn

100m

House or Collection of houses Prominent Building Coverage Area Final Mesh Traffic Flow Initial Design Expected Traffic Flow Initial and Final Design Mesh Box Final Design Mesh Box

SLIDE 49

WMN Problems

Users can be a problem...
“The BBC is broken, can you fix it?”
After several months of continued operation the mesh

began to behave unexpectedly:

Routes would fail and change periodically
Clients with poor access suddenly had no access
Unpredictable service to well connected clients
No significant capabilities of visualising and

understanding Wireless Mesh Network behaviour

SLIDE 50

Monitoring Platform

Developed a bespoke (centralised)

monitoring platform to analyse the behaviour of the Wray WMN

Captured data at 60

second intervals including:

Packet Flows
Routing tables
Signal Strength
Connected Clients

SLIDE 51

WMN Issues

Investigation of the data showed a growing change in

user behaviour :

short lived, low bandwidth  long lived, high bandwidth
Route Flapping: Network easily disrupted by long

lived flows and high number of connections from applications such as P2P

Flash Crowds: Community Element effecting the

network -- The YouTube Effect

Poor Design: This problem was compounded by the

inability of the Mesh to restore its state after disruption

SLIDE 52

Fixing the Problems

To alleviate the problems on the Wray WMN the

following actions were taken:

The gateway choice for each Mesh Node was hard coded
Available routing paths were heavily restricted, resulting in
nly „good‟ paths being used
Automatic pathway blocking disabled
Traffic shaping enabled
Boot order of Mesh Boxes closely observed
In summary, mesh elements of the network removed or

heavily restricted

SLIDE 53

Wray WMN Summary

Individual components can act poorly, but...
Failure of the network was not down to poor

behaviour of a single component

Decisions made by one component adversely effected

another

Data on the conditions of WMNs are important

SLIDE 54

Tegola Tiered Mesh Network Testbed

SLIDE 55

 Work by Edinburgh University  G. Bernardi, P. Buneman and M. K.

Marina, "Tegola Tiered Mesh Network Testbed in Rural Scotland," In Proc. ACM MobiCom 2008 Workshop on Wireless Networks and Systems for Developing Regions (WiNS-DR'08), Sep 2008.

 http://www.tegola.org.uk

Tegola Tiered Mesh Network Testbed in Rural Scotland

SLIDE 56

 Focused on  Embedded Applications  Mast Location  Self-Powered Masts (See Picture)  Adverse Environmental Conditions

 Transmission Over Water (Water reflecting

signal)

Tegola Tiered Mesh Network Testbed in Rural Scotland

SLIDE 57

Additional Research Questions

SLIDE 58

WMN Security -- Attack Space

SLIDE 59

WMN Intrusion Detection Systems

Technical Socio- technical

SLIDE 60

Administration & Fault Detection

 Who manages a WMN?  Automated fault detection and management?

 Who/What gathers statistics,  Who/what processes them?  Who/What acts on them?  Trust? / Security?

SLIDE 61

Ownership

 Who owns a Wireless Mesh Network?  Who is legally responsible?

SLIDE 62

Wrap up...

 Wireless Mesh Networking is a large research

space covering a vast array of topics

 Research is typically focused around one key area,

little considerations are taken to think about the big picture

SLIDE 63