Master of Computer Science & IT (MCIT)

Fall Semester 2013

Elec1ve Course (CT-531) Wireless & Mobile Communica:on

Lecture 6/7

Dr. Muhammad Mubashir Khan mmkhan@neduet.edu.pk

Department of Computer Science & IT, NED University of Engineering & Technology, Karachi

October2013

MANET A collec(on of wireless mobile hosts forming a

temporary network without the aid of any centralized administra(on or standard support services"

Ad-hoc network topology is dynamicnodes enter and leave the network con1nuously

No centralized control or fixed infrastructure to support network configura1on or reconfigura1on

Example scenarios for MANETs Mee1ngs Emergency or disaster relief situa1ons Military communica1ons Sensor networks & VANET

Mobile nodes have limited communica1on range Reduces ba\ery drain Enables spa1al reuse of limited bandwidth Increased network capacity

To connect all nodes in the network, each node is a packet source OR a packet sink OR a router

Nodes must route packets for other nodes to keep the network fully connected

In MANETs, rou:ng is very important and must be addressed!

A Vehicular Ad-Hoc Network, or VANET A technology that uses vehicles (moving cars) as nodes in a network to create a mobile network.

VANET turns every par1cipa1ng car into a wireless router or node

Allows cars approximately 100 to 300 metres of each other to connect and, in turn, create a network with a wide range.

As cars fall out of the signal range and drop out of the network, other cars can join in, connec1ng vehicles to one another

Possible applica1ons police and fire vehicles to communicate with each other for safety purposes.

WSN Sensor: is a device that measures a physical quan1ty and converts it into a signal which can be read by an observer or by an instrument.

A wireless sensor network (WSN) consists of spa1ally distributed autonomous sensors to monitor physical or environmental condi1ons, such as temperature, sound, vibra1on, pressure, mo1on or pollu1on and to coopera1vely pass their data through the network to a main loca1on.

The more modern networks are bi-direc1onal, enabling also to control the ac1vity of the sensors.

WSN Applica1ons Collec1on of data about temperature,

humidity, rainfall, and soil moisture from the environment

Air pollu1on monitoring Landslide detec1on Military security systems Temperature, humidity, vehicular movement, lightning condi1on, Pressure, soil makeup, noise levels, the presence or absence of certain

kinds of objects,

The current characteris1cs such as speed, direc1on, and size of an object.

Rou1ng Features in MANET Route-finding: Want to determine an op1mal way to find op1mal routes

Dynamic links Broken links must be updated when a node moves out of communica1on range with another node

New links must be formed when a node moves into communica1on range with another node

Based on this new informa1on, routes must be modified Frequency of route changes is a func1on of node mobility

Conven1onal rou1ng protocols for wired networks Distance vector rou1ng protocol Link state protocol

Distance Vector Rou1ng Protocol Each router has a table giving the distance from itself to all possible des1na1ons Each router periodically broadcasts its table to its neighbors

Neighbors will update their tables based on this informa1on to ensure they are using the shortest route to reach each des1na1on

To route a packet, router checks table to find next hop to des1na1on Tradeoff in how ogen rou1ng tables are exchanged

Too oHen: large amount of overhead, excess BW and computa1onal resources used Too infrequent: sub-op1mal or invalid routes (stale routes)

Can send rou1ng table updates whenever informa1on in the table changes due to a new link or a broken link Router Link Metric Router A to Router B 2 Router B to Router C 3 Router A to Router C 6 Router C to Router D 5

Generally, a metric is a number that is used as a standard of measurement for the links of a network. Each link is assigned a metric to represent anything from monetary cost to use the line, to the amount of available bandwidth.

Formula: M(i,k) = min [M(i,t) + M(t,k)]

Route finding (Example) M(i,k) = min [M(i,t) + M(t,k)]

Best path between two network points (M(i,k)) can be found by finding the lowest (min) value of paths between all network points.

Route from A to B to C is s1ll the best path: 5(A,C) = min[2(A,B) + 3(B,C)]!

Whereas the direct route from A to C could be: 6(A,C) = min[6(A,C)]!

Router Link Metric Router A to Router B 2 Router B to Router C 3 Router A to Router C 6 Router C to Router D 5

Link State Rou1ng Each router has a complete view of the topology of the en1re network Cost associated with each link Router monitors cost of link to each neighboring router

When cost changes, this informa1on is sent to all routers in the network Each router has a consistent view of the network

Uses Dijkstra Algorithm to compute op1mal paths

Each router can compute the shortest path from itself to the des1na1on

Incoming packet forwarded along this path

Link state protocols Converge more quickly than distance vector protocols Require more BW and compute power than distance vector protocols

Problems with using conven:onal rou:ng in MANET

Unidirec1onal links A can communicate with B does not always imply B can communicate

with A Redundant links

In wired networks, only one or a small number of routers connec1ng two networks

In wireless networks, may be several gateway nodes between a source and a sink due to wireless channel proper1es

Several redundant paths may be generated Waste of BW, computa1on, and storage

Periodic rou1ng updates Cost of sending rou1ng updates (BW, energy) is much greater in a

wireless network as compared to wired network Rou1ng updates required even if nothing has changed In highly-connected network, rou1ng updates may collide Mobiles cannot enter sleep state b/c need to hear all rou1ng updates

wastes energy

Problems with using conven:onal rou:ng in MANET

Dynamic nature of ad-hoc networks Wired networks rela1vely stable Occasionally links go down or come up or conges1on changes the cost of a link

Wireless network links con1nuously changing due to node mobility and environmental condi1ons Convergence to stable routes may be too slow to produce useful informa1on if rou1ng updates not sent ogen enough If rou1ng updates sent too ogen, wastes ba\ery of mobiles and channel bandwidth

Dijkstra algorithm

Rou1ng Protocols for MANET Route discovery and route maintenance Route discovery Ini1al discovery of valid route from source to des1na1on Source node can send a query for a des1na1on node Only des1na1on node responds back to the query If des1na1on located in sources transmission range, des1na1on responds and link established No periodic rou1ng updates needed Approach must be extended to case where des1na1on node not in source nodes transmission range

Want an approach that is simple and efficient !!!

Route Discovery One approach is to perform controlled flooding of the query Nodes receiving query will append their address to the route being recorded in the packet header and broadcast updated packet to all neighbors When a node receives a query, it checks to see if its address is already in the header (indica1ng this packet was already flooded by this node)

If address present, node drops packet Each query labeled with unique request ID

Each node keeps a cache with request IDs of packets it has already forwarded

Discards packets with request ID listed in nodes cache Avoids duplicate queries sent throughout the network

Node only propagates first copy of each route request packet it sees

Route Discovery Typically source and des1na1on will not be far away in an ad-

hoc network Can add a TTL (1me to live) to route request Each node reduces the TTL by one when it propagates the request If the TTL hits zero, the route request packet is dropped

When query reaches at des1na1on Des1na1on sends response back using route in packet header if available

Des1na1on sends response back to the node from which it received route request

Route discovery informa1on can be piggybacked onto data such as TCP connec1on packets Reduces latency (e.g., start-up 1me for TCP connec1on) Increases overhead, since packets flooded throughout network

Route Discovery Intermediate nodes can cache routes discovered Can use these routes if want to send a packet to a node listed in route

Reduces overhead in terms of number of route request packets required

Nodes can operate in promiscuous mode Listen to all packets exchanged on network Cache routes listed in packets

Route Maintenance Nodes can determine broken links through ACK/NACK included with most protocols

If link broken Node that detects broken link reports this informa1on back to sending node

Or node can try to fix the broken link by sending out its own route request to the des1na1on

If no ACK/NACK present in the link-layer protocol, nodes can listen to channel to determine if next hop transmits packet or not If do not hear forwarding of packet, assume link lost

Explicit rou1ng acknowledgements can also be used to determine the state of links

Proac1ve and Reac1ve Rou1ng Proac1ve rou1ng: nodes con1nuously evaluate and update

routes Periodic updates Triggered updateswhen a link changes Efficient if routes used ogen Large amount of overhead Similar to conven1onal rou1ng protocols

Reac1ve rou1ng: nodes evaluate and update routes only when they are needed When a node has a packet to send, it checks to see if it has a valid route

If no valid route known, node must send out a route-request message to obtain a valid route (controlled flooding of the network)

Data sent using valid route Efficient if routes not used ogen

Proac1ve Rou1ng Each node maintains consistent, up-to-date rou1ng informa1on in the form of a table with the next-hop to reach every node in the network

Changes in link state are transmi\ed throughout the network to update each nodes rou1ng table

Proac:ve rou:ng protocols DSDV (Des1na1on-Sequenced Distance-Vector) CGSR (Cluster-head Gateway Switch Rou1ng)

Des:na:on-sequenced distance-vector (DSDV) Each node maintains a table with

All possible des1na1on nodes Number of hops required to reach that node Next hop along route to that node Sequence number of the route

The route labeled with the highest (i.e. most recent) sequence number is used.

Sequence number used to dis1nguish new routes from old routes

Rou1ng table updates transmi\ed throughout network Full dump rou1ng updates

Large packet that carries all rou1ng informa1on Transmi\ed infrequently when li\le change in exis1ng links

Incremental rou1ng updates Contains only informa1on that has changed since last rou1ng

DSDV cont Route broadcast packets are sent to update tables at intermediate nodes Transmi\ed with Address of des1na1on Number of hops to reach des1na1on Sequence number of informa1on Sequence number of broadcast packet

If sequence number of update same as sequence number of informa1on in a nodes table, uses the route with the shortest path (or op1mal metric value)

If update is newer, node replaces its informa1on with the updated one

Cluster-head Gateway Switch Rou:ng (CGSR)

Cluster-head Gateway Switch Rou:ng (CGSR)

Clustering mul1-hop approach Cluster head controls nodes in cluster Cluster head selec1on algorithm within cluster Can perform minimum cluster changes to reduce overhead

E.g., only change cluster head if cluster head moves out of cluster or another cluster head moves into cluster

DSDV used as underlying rou1ng mechanism

Cluster-head Gateway Switch Rou:ng (CGSR)

All nodes that are in the communica1on range of the cluster-head belong to its cluster.

A gateway node is a node that is in the communica1on range of two or more cluster-heads.

In a dynamic network cluster head scheme can cause performance degrada1on due to frequent cluster-head elec1ons, so CGSR uses a Least Cluster Change (LCC) algorithm.

In LCC, cluster-head change occurs only if a change in network causes two cluster-heads to come into one cluster or one of the nodes moves out of the range of all the cluster-heads.

Cluster-head Gateway Switch Rou:ng (CGSR)

Each node maintains a cluster member table that has mapping from each node to its respec1ve cluster-head.

Each node broadcasts its cluster member table periodically and updates its table ager receiving other nodes broadcasts using the DSDV algorithm.

In addi1on, each node also maintains a rou1ng table that determines the next hop to reach the des1na1on cluster.

On receiving a packet, a node finds the nearest cluster-head along the route to the des1na1on according to the cluster member table and the rou1ng table.

Then it consults its rou1ng table to find the next hop in order to reach the cluster-head selected in step one and transmits the packet to that node.

CGSR Nodes send data to cluster head Cluster head sends data to gateway nodes Gateway nodes route packet to new cluster head Packet goes from: Cluster-head Gateway Cluster-head un1l it reaches des1na1ons Cluster-head

Updates sent as in DSDV

Reac1ve Rou1ng Protocols Routes created only when needed Requires route discovery and route maintenance

Also called source-ini1ated on-demand rou1ng Goal is to minimize the amount of overhead compared with proac1ve rou1ng at the expense of latency in finding a route when it is needed

Reac1ve rou1ng protocols AODV DSR

Ad-hoc on-demand distance vector (AODV)

Node needing to find a route to a des1na1on broadcasts a route request (RREQ) to its neighbors

Neighbors broadcast RREQ to their neighbors un1l des1na1on found or intermediate node has recent informa1on about a route to the des1na1on

Each RREQ uniquely iden1fied by sources ID and a sequence number

Intermediate nodes keep track of neighbor from which RREQ came to establish valid reverse path

AODV

AODV Des1na1on or intermediate node sends (unicast) route reply RREP back to neighbor from which RREQ came

Intermediate nodes set up rou1ng informa1on based on nodes from which they receive RREP packet These routes contain 1mers and will expire if not used

AODV Route maintenance is required Source node moves Send new RREQ to des1na1on

Intermediate node moves Upstream neighbor propagates link failure no1fica1on message to source

Source can send new RREQ to find valid route

Hello messages used to inform nodes of neighbors Can be used to ensure link kept alive Can contain informa1on about a nodes neighbors to give informa1on about topology

Dynamic Source Rou1ng (DSR) Similar to AODV but en1re route maintained within packet header Intermediate nodes do not need to keep rou1ng informa1on to route packets

No periodic route adver1sements needed Intermediate nodes propaga1ng a route request append their ID to the route record in the packet header

When packet reaches des1na1on or node with valid cached route to des1na1on, route reply returned If des1na1on node, send route record in route reply If intermediate node, append cached route to des1na1on to route record in route reply

DSR Route reply returned along

Cached path to source if available Reverse route of route record if symmetric links assumed Piggybacked with a route request to the source node

Route maintenance Route error messages used to propagate informa1on about broken links

All cached routes with broken links are removed Nodes can use promiscuous mode to listen to all traffic on channel Can update their route caches based on informa1on from packets Can send gratuitous route reply to source if node knows a be\er route to des1na1on

If intermediate node determines that next hop in route record unreachable, can replace this route with a cached route

DSR DSR uses source rou1ng in which a data packet carries the complete path to be traversed.

However, in AODV, the source node and the intermediate nodes store the next-hop informa1on corresponding to each flow for data packet transmission.

Comparison of Protocols Proac1ve approaches More efficient when routes used ogen Assures routes ready when needed Requires periodic route updates (overhead) Node mobility affects en1re network as rou1ng update

Reac1ve approaches More efficient when routes used occasionally Require node to first find route before data can be transmi\ed

Periodic route updates not required Can have localized route discovery to deal with node mobility

Comparison Reac1ve protocols can deliver many more packets than proac1ve protocols when node mobility is high Proac1ve protocols have too many rou1ng updates triggered and routes cannot converge

Overhead of reac1ve protocols depends on node mobility DSR has less overhead than AODV Due to intermediate node caching routes and requiring fewer route requests to be sent throughout the network

Overhead of DSDV is constant if (periodic updates) When node mobility is low, DSDV performs much be\er than when high Reac1ve protocols s1ll have lower overhead and be\er packet delivery ra1os

Conclusion DSDV performs well only in low-mobility situa1ons Even then, overhead s1ll high

DSR and AODV perform well under all mobility situa1ons DSR has lowest packets overhead If overhead includes packet header (overhead in bytes), AODV performs be\er than DSR due to smaller headers

However, header overhead might be less costly than packet overhead due to cost to obtain channel access, MAC/PHY headers, etc.

Recent Research for MANET Rou1ng Hybrid proac1ve-reac1ve approaches QoS support How to assure certain performance in such a dynamic environment?

Resource-aware rou1ng Use metrics besides shortest-path for rou1ng decisions Low power rou1ng Try to avoid network par11oning

Loca1on-based rou1ng

Power Aware Rou1ng Where is energy needlessly expended in rou1ng protocols?

Overhearing transmission of packets not intended for the node Conten1on Overhead in route finding and route maintenance

Promiscuous mode Some protocols take advantage of nodes overhearing transmissions to other node to gather more informa1on

Improves rou1ng (more up-to-date rou1ng informa1on) Large amount of energy required to receive and decode en:re packet

Conten1on requires power Channel sensing ACK schemes Retransmissions

Power Aware Rou1ng Route finding and maintenance Overhead packets transmi\ed throughout network to find op1mal routes

Many useless packets transmi\ed Rou1ng metrics Most MANET rou1ng protocols use shortest-hop as op1mizing metric for determining good routes

Other metrics Minimum latency Link quality Loca1on stability Power of intermediate nodes

Power as a Metric Why use power as a metric? Can over-use intermediate nodes Causes nodes to run out of energy early Increases conten1on around overused node

Increases system life1me Reduces chance of par11oning network

Goal in power-aware rou1ng: to increase network life:me Problem similar to load balancing How should routes be chosen to ensure nodes overall energy dissipa1on kept low while not overly-u1lizing any individual node?

Rou1ng Metric that Incorporate Power Shortest-hop Fewer hops less energy dissipa1on Downside!!! energy load is not evenly distributed

Energy consumed/packet If li\le traffic, this is the same as shortest hop If large amount of traffic, energy for conten1on will ensure packet routed around high-load areas

If E(i,j) = energy to transmit packet from node i to node j (including recep1on energy), this metric tries to minimize:

Rou1ng Metric that Incorporate Power No extra packet delay recorded using power-aware rou1ng Even though routes are longer

avoiding conges1on fewer retransmissions needed, shorter wai1ng 1me to transmit packet, etc.

Power-aware rou1ng beneficial for Large networks

If the network is too small, there are not many alterna1ve routes to choose from

Moderate network loads If network lightly loaded, shortest path algorithm is fine if network heavily loaded, cost of conten1on outweighs any power savings

Dense networks These networks have more alterna1ve routes than sparse networks

Security A\acks Vulnerability of channels and nodes, absence of infrastructure and dynamically changing topology, make ad hoc networks security a difficult task

Applica:on Layer: Malicious code, Repudia1on Transport Layer: Session hijacking (exploita1on of a valid computer sessionsome1mes also called a session keyto gain unauthorized access), Flooding

Security A\acks Network Layer: Sybil (subverts the reputa1on system by crea1ng a large number of pseudonymous iden11es, using them to gain a dispropor1onately large influence.),

Black Hole (discard packets instead of relaying), Grey Hole (only forward selec1ve data). Worm Hole (an a\acker records packets (or bits) at one loca1on in the network, tunnels them (possibly selec1vely) to another loca1on, and retransmits them there into the network. ),

Link Spoofing (An a\acking link adver1ses links to non-neighbors, by faking links to the two-hop neighbors of S, A can become one of its MPR (Mul1point Relay) nodes, and then manipulate traffic.),

Link Withholding (A\acker does not adver1se a link to a specific node or group of nodes.),

Loca1on disclosure etc.

Security A\acks

Data Link/MAC: Malicious or Selfish Behavior Selfish Nodes that do not forward others packets, thus maximizing their benefit at the expense of all others.

Selfish nodes are assumed to always behave ra1onally, so they cheat only if it gives them an advantage.)

Physical: Interference, Traffic Jamming, Eavesdropping

References E. Royer and C.-K. Toh, "A Review of Current Routing Protocols for

Ad-Hoc Mobile Wireless Networks,"IEEE Personal Communications Magazine, April 1999, pp. 46-55.

J. Broch, D. Maltz, D. Johnson, Y.-C. Hu, and J. Jetcheva, "A Performance Comparison of Multi-Hop Wireless Ad Hoc Network Routing Protocols,"Proc. Mobicom '98, Oct. 1998.

S. Singh, M. Woo, and C. Raghavendra, "Power-Aware Routing in Mobile Ad Hoc Networks,"Proc. Mobicom '98 , Oct. 1998.