Talk Outline

Ad Hoc Networks and the Ad Hoc Networks and the Associativity-Based Associativity-Based

Routing ProtocolRouting Protocol

Shafinaz BuruhanudeenShafinaz BuruhanudeenInformatics WorkshopInformatics Workshop

Supervisors:John Mellor, Prof Demetres KouvatsosSupervisors:John Mellor, Prof Demetres Kouvatsos

Talk OutlineTalk Outline•What are Ad Hoc networks

• Issues with routing in Ad Hoc Networks

• Existing Ad Hoc Routing Protocols

•Associativity Based Routing -ABR

•Simulation and results

A collection of two or A collection of two or more devices equipped more devices equipped with wireless with wireless communications and communications and networking capability.networking capability.

Devices communicate Devices communicate with one another with one another within/outside the radio within/outside the radio range to forward range to forward information packets from information packets from source to destination.source to destination.

Formed and deformed on the fly, thus making it Formed and deformed on the fly, thus making it self organised, adaptive and does not require self organised, adaptive and does not require system administrator.system administrator.

Infrastructureless-does not need fixed radio based Infrastructureless-does not need fixed radio based stations, wires or fixed routersstations, wires or fixed routers

Can take different forms (mobile, stand-alone, Can take different forms (mobile, stand-alone, networked)networked)

Routing information changes to reflect changes in Routing information changes to reflect changes in link connectivity (since nodes change on the fly)link connectivity (since nodes change on the fly)

Power consumption important (wireless networks Power consumption important (wireless networks rely highly on forwarding packets sent by other rely highly on forwarding packets sent by other nodes)nodes)

Why is routing difficult in Why is routing difficult in Ad Hoc Networks ?Ad Hoc Networks ?

Dynamic nature of the networkDynamic nature of the network No specific devices to do routingNo specific devices to do routing Limitations of Ad Hoc Networks Limitations of Ad Hoc Networks

likelike– high power consumptionhigh power consumption– low bandwidthlow bandwidth– high error rateshigh error rates

Challenges in ad hoc Challenges in ad hoc mobile networksmobile networks

Spectrum allocation Spectrum allocation and purchaseand purchase– ad hoc networks ad hoc networks

must operate under must operate under same form allowed/ same form allowed/ specifieed spectrum specifieed spectrum range and use of range and use of radio spectrum in radio spectrum in under control. under control.

– Spectrum frequency Spectrum frequency needs to be needs to be purchased - not sure purchased - not sure who should paywho should pay

Media accessMedia access– TDMA and FDMA schemes TDMA and FDMA schemes

not suitable since they do not suitable since they do not deal with host mobilitynot deal with host mobility

– results in scheduling results in scheduling frames for timely frames for timely transmission to support transmission to support QoS difficultQoS difficult

– sharing of media by sharing of media by mobile ad hoc nodes - mobile ad hoc nodes - access to common access to common channel must be made in channel must be made in distributed fashion distributed fashion through media access through media access control protocolcontrol protocol

RoutingRouting– nodes dynamically nodes dynamically

changes - links made changes - links made and break in and break in indeterministic fashion indeterministic fashion and changing topology and changing topology of networkof network

– existing distance-state existing distance-state vector and link state vector and link state based routing protocol based routing protocol not capable to adapt not capable to adapt frequent link changesfrequent link changes

– results: low results: low communication communication throughput and low throughput and low route convergenceroute convergence

– NEW ROUTING NEW ROUTING PROTOCOL NEEDEDPROTOCOL NEEDED

Energy efficiencyEnergy efficiency– most mobile devices most mobile devices

are operated by are operated by batteriesbatteries

– most network most network protocols do not protocols do not consider power consider power cunsumption an cunsumption an issueissue

– battery technology battery technology is still lagging is still lagging behind behind microprocessor microprocessor technologytechnology

– forwarding packets forwarding packets consume powerconsume power

TCP performanceTCP performance– enhancements enhancements

needed to ensure needed to ensure transport protocol transport protocol performs without performs without affecting end-to-end affecting end-to-end communication communication throughputthroughput

– current transport current transport protocols are not protocols are not designed for wireless designed for wireless ad hoc nwtworksad hoc nwtworks

– e.g:TCP is an end-to-e.g:TCP is an end-to-end protocol that end protocol that cannot distinguish cannot distinguish presence of mobility presence of mobility from congestionfrom congestion

Security and Security and privacyprivacy– ad hoc networks ad hoc networks

are intranets are intranets unless connected unless connected to the Internetto the Internet

– multiple nodes multiple nodes involve in ad hoc involve in ad hoc networks, relaying networks, relaying packets has to be packets has to be authenticated by authenticated by recognising recognising originator of the originator of the packetpacket

Existing Ad Hoc Routing Existing Ad Hoc Routing ProtocolsProtocols

Ad Hoc Routing Protocols

Table driven Source-initiatedon-demand

DSDV

CGSR

WRP

AODV DSR TORA ABR

SSR

Destination sequenced distance vector

Clusterhead Gateway Switch Routing

The Wireless Routing Protocol

Ad Hoc On-Demand Distance Vector Routing

Dynamic Source Routing

Temporally Ordered Routing Algorithm

Signal Stability Routing

Routing algorithm is important to Routing algorithm is important to compute a path from one node to compute a path from one node to another within a finite timeanother within a finite time

when multiple path exist, selection when multiple path exist, selection method needed- common way: method needed- common way: shortest path criteriashortest path criteria

shortest path=min number of hops or shortest path=min number of hops or least num of links used in the routeleast num of links used in the route

Associativity Based Long-Associativity Based Long-Lived Routing (ABR)Lived Routing (ABR)

Associativity is related to the spatial, Associativity is related to the spatial, temporal and connectivity of a mobile temporal and connectivity of a mobile hosthost

Specifically, associativity is measured Specifically, associativity is measured by one node’s connectivity relationship by one node’s connectivity relationship with its neighbouring nodeswith its neighbouring nodes

nodes association with its neighbours nodes association with its neighbours changes as it is migrating and its changes as it is migrating and its transition period identified by transition period identified by associativity ticks/ countsassociativity ticks/ counts

Associativity of a mobile Associativity of a mobile hosthost

Migration: after unstable Migration: after unstable period, exist a period of period, exist a period of stability (i.e., a node is stability (i.e., a node is constantly associated with constantly associated with certain neighbours over time certain neighbours over time without loosing connectivity without loosing connectivity with these neighbours)with these neighbours)

mobile host will spend some mobile host will spend some dormant time (mobile host dormant time (mobile host might/might not move) might/might not move) within a wireless cell before it within a wireless cell before it starts to break connectivity starts to break connectivity with surrounding neighbours with surrounding neighbours and move outside boundary and move outside boundary of existing wireless cell.of existing wireless cell. Time

Associativity Ticks

threshold

Dormant time

ABR Protocol DescriptionABR Protocol Description

Consists ofConsists of

– route discovery route discovery phasephase

– route route reconstruction reconstruction phasephase

– route deletion route deletion phasephase

when source node desires a when source node desires a route, the route discovery route, the route discovery phase is invoked: consists of a) phase is invoked: consists of a) broadcast query (BQ) and b) broadcast query (BQ) and b) await reply cycle (REPLY)await reply cycle (REPLY)

when links of established route when links of established route change due to source/ change due to source/ destination/ intermediate node destination/ intermediate node migration, route reconstruction migration, route reconstruction phase evokedphase evoked

when source node no longer when source node no longer desires the route, it initiates desires the route, it initiates the deletion phasethe deletion phase

Route Reconstruction Phase Route Reconstruction Phase (RRC)(RRC)

Although route selected using ABR Although route selected using ABR tends to be long-lived, there are tends to be long-lived, there are cases where association stability cases where association stability relationship is violatedrelationship is violated– eg:mobilityeg:mobility

RRC procedured invoked to cope with RRC procedured invoked to cope with mobilitymobility

ABR localise route repair operation ABR localise route repair operation intelligently to avoid excessive intelligently to avoid excessive control overhead and disturbing control overhead and disturbing unconcerned nodesunconcerned nodes

ABR route discovery is fast because it ABR route discovery is fast because it repairs broken route on-the-fly in repairs broken route on-the-fly in realtime (partial discovery)realtime (partial discovery)

ABR route maintenance ABR route maintenance phase consists of:phase consists of:

– partial route discoverypartial route discovery– invalid route erasureinvalid route erasure– valid route update andvalid route update and– new route discovery new route discovery

(worst case)(worst case)

When a discovered route is no When a discovered route is no longer desired, a route delete (RD) longer desired, a route delete (RD) broadcast will be initiated by the broadcast will be initiated by the source so that all INs will update source so that all INs will update their routing table entries their routing table entries

Simulation appletSimulation applet

SimulationSimulationFor each node desiring a route to Dest, the SRC will send a BQ. In the simulator, the user will input the SRC node number and the DEST node number and the ‘SEND BQ” button is clicked. Once the BQ expires i.e. timeout period reached the route formation stops for that particular BQ, but for other BQs, the same process continues.

Depending on the BQ timeout specified by user, the route will be transformed (if any) at every instant. Any number of BQs can be fired during the simulation period.

In the simulation, there is a table (n by n matrix) of AT for each node. The matrix shows the AT between 2 nodes, depending on the number of nodes decided by user. This matrix changes dynamically, updating at every instant.

The use of AT will enable us to get a route which is stable, and since in mobile ad hoc networks, mobility is the main issue, we don’t want routes to break off as soon as we get a link to the destination. For comparison study, I have carried out the simulation without the feature of AT. This protocol (XAT), uses the feature of minimum hop to get to destination. For this version of the protocol, I did not include the feature of AT, thus route established may or may not be stable.

Stability here refers to routes which can maintain its link for a minimum of 2 units of time in the simulation. However, in 90% of the cases, routes tend to break at or before 1 unit of time.

The lifetime of a particular route is dependent on the speed and direction of movement of all the nodes involved in the route.

Simulation resultsSimulation results Stable routes means routes can

maintain or remain long enough for data packets to be transmitted to

the destination node.

with increasing number of nodes, the percentage of unstable routes increases with the protocol without the use of AT (XAT). This is because nodes move out of connectivity or because routes are constructed on migrating nodes, thus making the routes break. Depending on the type of message to transmit, both, the protocol with AT and XAT can be useful.

Comparison of route instablility (routes lasting less than 3 units of time)

0

20

40

60

80

100

25 30 35 40

number of nodes

% of

unsta

ble ro

utes

Protocol without AT Protocol using AT

with increasing number of nodes, the protocol with the use of AT seem to have more number of hops compared to the protocol without. In mobile ad hoc networks, a route with shortest path alone doesn’t determine that the route will last and be stable. Thus, having more number of hops when the route is surely stable, can be helpful than having to reconstruct the route if it breaks. The process of reconstructing routes may lead to more time and a high cost.

Number of hops comparison

0

0.5

1

1.5

2

2.5

3

3.5

4

4.5

25 30 35 40

Number of nodes

mea

n nu

m o

f hop

s

Protocol w ithout use of AT

Protocol w ith AT

Percentage of established routes with varying transmission power

0

10

20

30

40

50

60

70

80

90

100

0 10 20 30 40 50 60

transmission power (pixel)

% o

f ro

ute

s e

sta

blis

he

d

nodes=20

nodes=25

nodes=30

nodes=35

nodes=40

nodes=45

n=45

n=40n=35n=30

n=25

n=20

The increasing pattern of successful routes tends to increase linearly with all combination of nodes. With more nodes in the network, the probability of establishing routes would be more, and with lesser nodes, to get a high probability of routes, a bigger transmission power would be needed.

Average time to establish route and max number of hops

3 4 6 8

0

10

20

30

40

50

60

70

80

90

100

25 30 35 40

Num of nodes

Un

its o

f tim

e

Max num of hops

Ave units oftime todestination

The result shows that with lesser nodes, it takes more time to reach DEST. It is almost likely to happen that the protocol might not work at all in a sparse environment. With more nodes in the network, the average time it takes to reach DEST is lesser, since there are more intermediate nodes to participate in the network. Also it shows that with more number of nodes, the maximum hops increases, thus showing

that more nodes participate to reach DEST.

Percentage of successful routes with varying motion speed

0

10

20

30

40

50

60

70

80

90

100

0 10 15 20 25 30 35 40 45 50 60

motion speed limit (pixel/time)

% o

f suc

cess

ful r

oute

s es

tabl

ishe

d

nodes=25 nodes=30 nodes=35 nodes=40

with varying motion speed limit, the number of routes increases. Motion speed 0 indicates that the nodes are not at all in motion. Since the positioning of the nodes in the simulation is random, there are times when the intended source node and destination are directly connected to each other, thus making it possible for

a direct link. when the motion speed limit increases till half the transmission

power size, the number of possible routes increases almost linearly. However, when the motion speed increases to 30 and above, the percentage of routes established drops linearly and when the speed reaches 60 pixel/time, with a transmission range of 50, there are no routes constructed, since there will be

no value for AT at this speed.

In conclusion, if there is no feature to determine a stable road (for my simulation, the use of AT), then the protocol might not be as efficient. Transiting nodes may be chosen to relay data packets thus resulting in link breakage very

often. The use of ABR with its AT promises long lived, stable

routes, which is important to carry big data packets in the network such as video or audio files or for web browsing. However, with small data packets, where the duration of the call is just intended for less than 3 units of time, then, we may not need a protocol with the use of AT since that

might be a waste in bandwidth and signal strength.

Further WorkFurther Work A constrained environment where

mobility models can differ from one area to another will be modelled

Different kinds of mobility models will then be incorporated in the simulation model to evaluate the protocol and obtain results from it.

Queing theory models will be included