ad hoc networks - hscc.cs.nthu.edu.tw

Copyright © 2011, Dr. Dharma P. Agrawal and Dr. Qing-An Zeng. All rights reserved. 1

Chapter 13

Ad Hoc Networks


Outline

Introduction

Characteristics of MANETs

Applications

Routing

Table-driven Routing Protocols

Source-initiated On-demand Routing

Hybrid Protocols

Vehicular Area Network (VANET)

Security Issues in Mobile Ad hoc Networks (MANETs)

Network Simulators

Summary


Introduction

A Mobile Ad hoc Network (MANET) is an

autonomous system of nodes (MSs) (also serving

as routers) connected by wireless links

No infrastructure exists in a MANET

The network’s wireless topology may change

dynamically in an unpredictable manner since

nodes are free to move and each node has limited

transmitting power

Information is transmitted in a store-and

forward manner (peer-to-peer) using multi-hop

routing


Introduction (Cont’d)

Each node is equipped with a wireless

transmitter and a receiver with an appropriate

antenna

We assume that it is not possible to have all nodes

within each other’s radio range

When the nodes are close-by i.e., within radio

range, there are no routing issues to be addressed

At a given point in time, wireless connectivity in

the form of a random multi-hop graph exists

between the nodes


A Mobile Ad Hoc Network (MANET)

MS2

MS3

MS2

MS4

MS1

MS5

MS6

MS7 Symmetric link

Asymmetric link


A B

A D C B E

Direct Transmission versus Multi-hop

Energy Consumption

Time delay


Characteristics of MANETs

Dynamic topologies: Network topology may change dynamically as the nodes are free to move

Bandwidth-constrained, variable capacity links: Realized throughput of wireless communication is less than the radio’s maximum transmission rate Collision occurs frequently

Energy-constrained operation: Some nodes in the ad hoc network may rely on batteries or other exhaustible means for their energy

Limited physical security: More prone to physical security threats than fixed cable networks


Applications

Defense applications: On-the-fly communication set up for soldiers on the ground, fighter planes in the air, etc.

Crisis-management applications: Natural disasters, where the entire communication infrastructure is in disarray

Tele-medicine: Paramedic assisting a victim at a remote location can access medical records, can get video conference assistance from a surgeon for an emergency intervention

Tele-Geoprocessing applications: Combines geographical information system, GPS and high capacity MS, Queries dependent of location information of the users, and environmental monitoring using sensors

Copyright © 2011, Dr. Dharma P. Agrawal and Dr. Qing-An Zeng. All rights reserved.

Applications

Vehicular Area Network: in providing emergency

services and other information in both urban and

rural setup

Virtual navigation: A remote database contains geographical representation of streets, buildings, and characteristics of large metropolis and blocks of this data is transmitted in rapid sequence to a vehicle to visualize needed environment ahead of time

Education via the internet: Educational opportunities on Internet to K-12 students and other interested individuals. Possible to have last-mile wireless Internet access

9


Routing in MANETS - Goals

Provide the maximum possible reliability - use alternative routes if an intermediate node fails

Route network traffic through the path with least cost metric between the source and destination

Give the nodes the best possible response time and throughput


Need for Routing

Route computation must be distributed. Centralized routing in a dynamic network is usually very expensive

Routing computation should not involve the maintenance of a global state

Fewer nodes must be involved in route computation

Each node must care about the routes to its destination and must not be involved in frequent topology updates

Stale routes must be either avoided or detected

Broadcasts should be avoided (highly unreliable)

If topology stabilizes, routes must converge to optimal routes

It is desirable to have a backup route when the primary route has become stale

11


Routing Classification

The existing routing protocols can be classified as:

Proactive: when a packet needs to be

forwarded, the route is already known

Reactive: Determine a route only when there is

data to send

Routing protocols may also be categorized as:

Table Driven protocols

Source Initiated (on demand) protocols

Hybrid protocols


Table Driven Routing Protocols

Each node maintains routing information to all

other nodes in the network

When the topology changes, updates are

propagated throughout the network

Examples are:

Destination Sequenced Distance Vector

routing (DSDV)

Cluster-head Gateway Switch routing (CGSR).

Wireless Routing Protocol (WRP)


Destination Sequenced Distance Vector Routing (DSDV)

Based on the Bellman-Ford algorithm

Each mobile node maintains a routing table in

terms of number of hops to each destination

Routing table updates are periodically

transmitted

Each entry in the table is marked by a sequence

number which helps to distinguish stale routes

from new ones, and thereby avoiding loops


DSDV

A new route broadcast contains: Destination address

Number of hops required to reach destination.

Sequence number of information received about the

destination

To minimize the routing updates: Either full dump carrying all available routing

information

Smaller incremental packets containing the change

in information since last full dump

15

14

15

11

10

13

8

4 7

12

5

9

6

1

3

2

DestinationID

SourceID

Dest NextNode Dist SeqNo

2

3

4

5

6

7

8

9

10

11

12

13

14

15

2

2

5

5

6

2

2

5

5

5

6

6

6

5

1

2

2

1

1

3

3

4

2

3

3

4

3

4

22

26

32

134

144

162

170

186

142

176

190

198

214

256

(a) Topology graph of the network (b) Routing table for Node 1

Figure 7.5. Route establishment in DSDV 16

14

15

11

10

13

8

4 7

12

5

9

6

1

3

2

DestinationID

SourceID

Dest NextNode Dist SeqNo

2

3

4

5

6

7

8

9

10

11

12

13

14

15

2

2

5

5

6

2

2

5

5

5

6

5

6

5

1

2

2

1

1

3

3

4

2

4

3

4

3

4

22

26

32

134

144

162

170

186

142

180

190

198

214

256

Figure 7.6. Route maintenance in DSDV

11

Node

Movement

17

DSDV (Cont.)

Advantages:

Route setup process is very fast

Make the existing wired network protocol apply to ad

hoc network with fewer modifications

Disadvantages:

Excessive control overhead during high mobility

Node must wait for a table update message initiated by

the destination node

Cause stale routing information at nodes

18


Cluster-head Gateway Switch Routing (CGSR)

CGSR is a clustered multi-hop mobile wireless network with

several heuristic routing schemes

A distributed cluster-head (CH) selection algorithm is used to

elect a node as the cluster head

It modifies DSDV by using a hierarchical CH to route traffic.

Gateway nodes serve as bridge nodes between two or more

clusters

A packet sent by a node is first routed to its CH and then the

packet is routed from the CH to a gateway of another cluster

and then to the CH and so on, until the destination cluster

head is reached

Frequent changes in the CH may affect the performance of

the routing protocol


1

3

2

4 7 10

5

6

8 9

11

12

Cluster Head

Internal Node

Gateway Node

Routing in CGSR from node 1 to node 12

CGSR (Cont’d)

CGSR (Cont’d)

Advantages:

Better bandwidth utilization

Easy to implement priority scheduling scheme

Disadvantages:

Increase in path length

Instability when cluster-head are high mobility

Battery-draining rate at cluster-head is more than a

normal node

Frequent changes in the cluster-head = multiple path

break

21


Source-Initiated On-Demand Routing

Ad hoc On-Demand Distance Vector (AODV)

Dynamic Source Routing (DSR)

Temporary Ordered Routing Algorithm (TORA)

Associativity Based Routing (ABR)

Signal Stability Routing (SSR)

Reactive protocol:

Dynamic Source Routing Protocol

(DSR)

Beacon-less: no hello packet

Routing cache

DSR contains two phases

Route Discovery (find a path)

Flooding RouteRequest with TTL from source

Response RouteReply by destination

If an forwarding node has a route to the destination in its

route cache, it sends a RouteREply to the source

Route Maintenance (maintain a path)

RouteError

23

Routing Discovery

14

15

11

10

13

8

4 7

12

5

9

6

1

3

2

DestinationID

SourceID

Figure 7.10. Route establishment in DSR.

Network Link

RouteRequest

RouteReply

Path1: 1-2-3-7-9-13-15

Path2: 1-5-4-12-15

Path3: 1-6-10-11-14-15

2011/12/8 24

Routing Maintain

14

15

11

10

13

8

4 7

12

5

9

6

1

3

2

DestinationID

SourceID

Figure 7.11. Route maintenance in DSR.

Network Link

Selected Path

RouteError

Broken Link

25

DSR (Cont’d)

Advantage

No need to updating the routing tables

Intermediate nodes are able to utilize the Route Cache

information efficiently to reduce the control overhead

There are no “hello” messages needed (beacon-less)

Disadvantage

The Route Maintenance protocol does not locally repair a

broken link

There is always a small time delay at the begin of a new

connection

2011/12/8 26


Ad hoc On-Demand Distance Vector Routing (AODV)

AODV is an improvement over DSDV, which

minimizes the number of required broadcasts by

creating routes on demand

Nodes that are not in a selected path do not

maintain routing information or participate in

routing table exchanges

A source node initiates a path discovery process

to locate the other intermediate nodes (and the

destination), by broadcasting a Route Request

(RREQ) packet to its neighbors

AODV (Cont’d)

Every node has a routing table. When a node knows

a route to the destination, it sends a route reply to

the source node

The major difference between DSR and AODV

DSR uses source routing in which a data packet carries

the complete path to traversed.

AODV stores the next-hop information corresponding to

each flow for data packet transmission.

Message types

Route Requests (RREQs)

Route Replies (RREPs)

Route Errors (RERRs)

28

AODV (Cont’d)

RouteRequest packet carries:

SreID, DestID, DestSeqNum, BcastID, and TTL

DestSeqNum indicates the freshness of the route is

accepted

An intermediate node receives a RouteRequest packet. It

either forwards it or prepares a RouteReply if it has a valid

route to the destination

RouteReply packet:

A node receives RouteReply packet will record the

information as the next hop toward the destination

AODV does not repair a broken path locally

29

14

15

11

10

13

8

4 7

12

5

9

6

1

3

2

DestinationID

SourceID

Figure 7.12. Route establishment in AODV.

Network Link

RouteReply

RouteRequest

Path1: 1-5-10-14-15

Path2: 1-5-4-12-15

Cached Route: 14-15

30

Route Maintenance

14

15

11

10

13

8

4 7

12

5

9

6

1

3

2

DestinationID

SourceID

Figure 7.13. Route maintenance in AODV.

Network Link

Route for 1 -> 15

RouteError

Broken Link

31


AODV (Cont’d)

Advantage

Establish on demand

Destination sequences are used to find the latest path to

destination

The connection setup delay is less

Disadvantage

Intermediate node can lead to inconsistent route

Beacon-base

Heavy control overhead

2011/12/8 32


Hybrid Protocols

Zone Routing Protocol (ZRP)

Fisheye State Routing (FSR)

Landmark Routing (LANMAR)

Location-Aided Routing (LAR)

Zone Routing Protocol (ZRP)

Intra-zone routing protocol (Proactive routing)

It is only used in the routing zone.

It brakes all nodes in the routing zone into interior

nodes and peripheral nodes.

Each node maintain routing path to all nodes in the

routing zone by exchanging periodic route update

packets.

Inter-zone routing protocol (Reactive routing)

34

4

13

5

3

11

10

14

12

6

7

1

15

9

2

8

Figure 7.26. Routing Zone for node 8 in ZRP.

Routing Zone for Node 8

Network Link

Routing Zone with Radius = 1


35

ZRP (Cont’d)

When a node s has packets to be sent to a node d

It checks whether node d is with in its zone.

If d isn’t in the zone, s broadcasts (uses unicast routing) the RouteRequest to its peripheral nodes.

If any peripheral node finds d in its zone, it sends a RouteReply back to s indicating the path.

Otherwise, the peripheral node rebroadcasts the RouteRequest again.

The query control must ensure that redundant or duplicate RouteRequests are not forwarded.

The zone radius has significant impact on the performance.

36

4

13

5

3

11

10

14

12

6

7

1

15

9

2

8

Figure 7.27. Path finding between node 8 and node 16.

Routing Zone for Node 8

Network Link

RouteRequest


RouteReply

16

37

ZRP (Cont’d)

Advantage

ZRP reduces the control overhead employed in on-

demand approach and the periodic flooding of routing

information in table-driven.

Disadvantage

In the absence of a query control, ZRP tends to produce

higher control overhead.

The decision on the zone radius has a significant impact

on the performance of the protocol

38

Location-Aided Routing (LAR)

With the availability of GPS, the mobile hosts knows their physical locations

Expected Zone: the destination node is expected to be presented

Request Zone: the path-finding control packets are permitted to be propagated

LAR1: the source node specifies the request-zone in the RouteRequest packet

LAR2: source node includes the distance between itself and the destination node

39

Expected Zone & Request Zone

4

13

5

3

11

10

14

12

6

7

1

15

9

2

8

S (X1, Y1) (X2+r, Y1)

(X1, Y2+r) (X2+r, Y2+r)

D (X2, Y2)

rNetwork Link

RoutReply

RoutRequest

RequestZone

ExpectedZone

Figure 7.16. RequestZone and ExpectedZone in LAR1.40

LAR Scheme

41

4

13

5

3

11

10

14

12

6

7

1

15

9

2

8

Destination Node

D (X2, Y2)

Network Link

RoutReply

RoutRequest

Figure 7.17. Route establishment in LAR2.

Source Node

S (X1, Y1)

LAR (Cont’d)

Advantage

Reduce control overhead

Increase utilization bandwidth

Disadvantage

Depend heavily on availability of GPS

43


Protocol Characteristics (1/2)

Routing

Protocol

Route

Acquisition

Flood for

Route

Discovery

Delay for

Route

Discovery

Multipath

Capability

Effect of Route Failure

DSDV Computed

a priori

No No No Updates the routing

tables of all nodes

WRP

Computed

a priori

No

No

No

Ultimately, updates the

routing tables of all

nodes by exchanging

MRL between

neighbors

DSR

On-

demand,

only when

needed

Yes.

Aggressive

use of caching

may reduce

flood

Yes

Not explicitly.

The technique

of salvaging

may quickly

restore a route

Route error

propagated up to the

source to erase invalid

path

AODV

On-

demand,

only when

needed

Yes.

Controlled

use of cache to

reduce flood

Yes

No, although

recent research

indicate

viability

Route error

propagated up to the

source to erase invalid

path


Protocol Characteristics (2/2)

Routing Protocol

Route Acquisition

Flood for Route Discovery

Delay for Route Discovery

Multipath Capability

Effect of Route Failure

TORA On-demand, only when needed

Basically one for initial route discovery

Yes. Once the DAG is constructed, multiple paths are found

Yes Error is recovered locally

ZRP Hybrid Only outside a source's zone

Only if the destination is outside the source's zone

No Hybrid of updating nodes' tables within a zone and propagating route error to the source

LAR On-demand, only when needed

Reduced by using location information

Yes No Route error propagated up to the source


Multipath Routing

Multipath provides redundant paths between source and destination.

Routes are disconnected frequently in ad hoc networks due to mobility or poor wireless link quality

Multipath routing could lead to out-of-order delivery, resequencing of packets at the destination and increased collision

Can aid in secured routing against denial of service

Various unipath protocols can discover multiple paths


On-Demand Multipath Routing

Extension of DSR protocol

Route discovery by flooding the network query: two possible extensions

First extension: destination responds to a set of query packets- source has multiple routes

Second extension: destination replies to all intermediate nodes along primary paths- giving alternate disjoint routes to all those nodes

47


Multipath Routing

Route construction and maintenance in On Demand

Multipath Routing Protocol

S

P1

D

P3

P4

L1 L2 L3 L4 Lk n1 n2 n3 n4 nk+1


Ad Hoc On-Demand Distance Vector-Backup Routing

AODV-BR constructs routes on demand

Uses alternate path if primary path disrupted

Utilizes mesh arrangement to provide alternate paths

Two phases: Route Construction and Route Maintenance

Route construction: Source initiates route discovery by

flooding:

Intermediate nodes stores previous hop and source node information upon receiving non-duplicate path request

Mesh construction and alternate paths established during route reply phase

Node chooses the best route among multiple route responses

When route response reaches the destination, primary route is established


Multipath Routing (Cont’d)

Route Maintenance and Mesh Routes

Primary path used unless failure

In case of route failure, one hop data broadcast is performed

Neighbors having entry to destination in alternate route table send unicast packet

A node on primary path detects a route failure, sends a route error packet to source

50


Multipath Routing: AODV-BR

1 2 4

3

5 6

7 Alternate route

used when

primary

disconnects

1 2 4

3

5

6 7

Multiple routes from

Node 1 to Node 7

Primary route

Alternate route



52

Basic objective is to find some relevant local information, such as close

by gas stations, restaurants, grocery stores, and hospitals

Primary motivation is to obtain knowledge of local amenities


VANET (Cont’d)

53

Hello beacon signals are sent to determine other vehicle in the vicinity

Table is maintained and periodically updated in each vehicle

Vehicle in an urban area move out relatively low speed of up to 56 km/hr while

Speed varies from 56 km/hr to 90 km/hr in a rural region

Freeway-based VANET could be for emergency services such as accident, traffic-jam, traffic detour, public safety, health conditions, etc.

Early VANET used 802.11-based ISM band


VANET (Cont’d)

75 MHz has been allocated in 5.850 - 5.925 GHz band

Coverage distance is expected to be less than 30 m and data rates of 500 kbps

FCC has allocated 7 new channels of in 902 - 928 MHz range to cover a distance of up to 1 km using OFDM

It is relatively harder to avoid collision or to minimize interference

slotted ALOHA does not provide good performance

Non-persistent or p-persistent CSMA is adopted

54



55

Characteristic Urban Area Rural Area Freeway-based

1. Connectivity High Sparse Unpredictable

2. Application Streaming media; emergency

information; geographical information

Geographical

information

Emergency use

3. Mobility Low; slow changes in connectivity Low medium High-speed; rapid changes

in link topology

4. Mobility

pattern

Random road Most likely fixed

path

Fixed

5. Routing Geographic Geographic Connectivity-aware

Routing

6. Area of

communication

Small region Small area Large space

7. Delay Mostly acceptable Acceptable Not acceptable

8. Type of

Information

Nearby grocery stores, restaurants, gas

stations; and hospitals; rarely for

emergency; safely for pedestrian or

cyclists

Nearby amenities;

notifying emergency

of a vehicle

Congestion; detour; accident; traffic

jam; emergency; road geometry

warning; rail-road crossing;

overweight vehicle

9. Volume of

Information

Low to medium Low: infrequent

message

Large: frequent data

10. Data

Delivery mode

Push Push Pull or Push

11. Security

Requirements

Short term Short term Relatively long term


Security Issues in MANETs

Missing authorization facilities hinders the usual practice of distinguishing nodes as trusted or non-trusted

Malicious nodes can advertise non-existent links, provide incorrect link state information, create new routing messages and flood other nodes with routing traffic

Attacks include active interfering, leakage of secret information, eavesdropping, data tampering, impersonation, message replay, message distortion, and denial-of-service (DoS)

Encryption and authentication can only prevent external nodes from disrupting the network traffic

Internal attacks are more severe, since malicious insider nodes are protected with the network’s security mechanism

56


Disrupting Routing Mechanism by A

Malicious Node

Changing the contents of a discovered route

Modifying a route reply message, causing the packet to be

dropped as an invalid packet

Invalidating the route cache in other nodes by advertising

incorrect paths

Refusing to participate in the route discovery process

Routing mechanism attacked by:

Modifying the contents of a data packet or the route via

which that data packet is supposed to travel

Behaving normally during the route discovery process but

drop data packets causing a loss in throughput

Generate false route error messages whenever a packet is

sent from a source to a destination 57


Attacks by A Malicious Node

Can launch DoS attack

A large number of route requests due to DoS attack or

a large number of broken links due to high mobility

Can spoof its IP and send route requests with a fake

ID to the same destination

Routing protocols like AODV have many

vulnerabilities

Authority of issuing authentication is a problem as a

malicious node can leave the network unannounced

58


Intrusion Detection System (IDS)

Automated detection

Subsequent generation of an alarm

IDS is a defense mechanism that continuously monitors the network for unusual activity and detects adverse activities

Capable of distinguishing between attacks originating from inside the network and external ones

Intrusion detection decisions are based on collected audit data

59

Security Approaches


Security Approaches (Cont’d)

Intrusion Response Mechanism (IRM)

Depends on the type of intrusion

Likely responses include: reinitializing communication channels between nodes, identifying the compromised nodes, and staring a re-authentication process among all nodes

60


Requirements for an Intrusion

Detection System IDS must effectively detect and classify malign and

benign activity correctly

IDS should detect a large percentage of intrusions

IDS must be capable of recovering from system crashes

Intrusion Detection in MANET

o IDS is limited to observing only the traffic coming in and out of the node

Six functional components of an IDS agent are:

61


Distributed IDS

Anomaly detection procedure:

o Select or divide audit data

o Perform appropriate data transformation according to the entropy measures

o Compute the classifier using training data

o IDS architecture for a wireless ad hoc architecture

o Apply the classifier to test data

o Post-process alarms to produce intrusion reports

62


MOBILE AGENTS

Mobile agents are agents that move around the network

Eliminates the need for moving large volume of data

If some portion of an IDS get destroyed, mobile agents can

still continue to work

63


Local Intrusion Detection System

A common communication framework to facilitate all external and internal communication with LIDS

Several data collecting agents for different tasks, such as:

o A local LIDS agent is in charge of local intrusion detection and response

o Mobile agents collect and process data from remote hosts any additional investigation

o A mobile agent should also be able to protect itself from malicious mobile agent

o MIB variables for mobile and LIDS agents are obtained from local MIB agent

o An SNMP-based agent allows optimized updates and retrieval of the MIB variables used by intrusion detection

64


IDS based on Static Stationary

Database

An IDS agent runs on each node as two parts

Mobile IDS agent resides on each node

o Five parts: a local audit trial, a local intrusion database (LID), a secure communication module, anomaly detection modules (ADMs), and misuse detection modules (MDMs)

Stationary secure database

o Signature files of known attacks, established patterns of users on the network, and the normal traffic flow of the network

65


Cluster-based Intrusion Detection

System

MANETs can be organized into a number of clusters

A cluster head (CH) is selected that has connections to all 1-hop members

CH assignment must be fair and secure

Should detect Blackhole, packet drop attack, maximum sequence number attacks, etc.

o Blackhole attack is suction attack where a malicious node uses the routing protocol to advertise itself as having the shortest path to the node whose packets it wants to intercept

o Then drops the entire traffic

o Statistics like number of others packets forwarded, number of packets originated etc. are collected to monitor the activity of a node

66


Cluster Formation

Formed by dividing the network into manageable

entities

CH also communicates with other clusters for

cooperative detection and response

Cluster management responsibility is rotated among

the capable members of the cluster for load balancing

and fault tolerance and must be fair and secure

Can be achieved by conducting regular elections

The proposed election process does not require the

clique computation or the neighbor information

67


Cluster Head Selection

Initial cluster head setup round composed of two steps: Clique and Cluster head Computation

A clique is a group of nodes where every pair of members can communicate through a direct wireless link

Once the protocol is finished, every node is aware of its fellow clique members

The clique requirement can be relaxed right after the CH has been identified with has direct links with all members

A count is maintained to remember many times an elected node has refused to respond

68


Cluster-based Intrusion Detection

CH provides an opportunity for launching collaborative

intrusion detection

Detects intrusions under various attacks such as Blackhole,

routing loop, selfishness, and sleep deprivation in a MANET

environment

At CH, packet analysis of nodes’ traffic analysis reduces

processing at each node

IF CH finds some malicious, it informs its members and the

neighboring clusters to take certain set of actions

IDS can be either host-based or network based

The techniques to detect intrusion can be anomaly detection or

misuse/signature detection

69


The IDS can be categorized as misuse detection system or anomaly detection system

Misuse detection or signature detection system is generally used for known patterns of unauthorized behavior

Anomaly detection system identifies intrusions using ‘normal’ activity baseline

Disadvantages of Misuse Detection:

o Misuse detection system often fails if the database of attack signatures is not up to date

o The bulk of database cannot be handled due to memory constraints

70

Cluster-based Intrusion Detection ctd..


Logging Module of CHs

CH captures all the traffic in the promiscuous mode

Keeps the data related to traffic such as number of packets sent, received, forwarded or dropped in a database

Intrusion Information Module

o Every node must maintain a database such as "intrusion interpretation base“

o Anomalous behaviors must also be well defined with upper and lower threshold values

Intrusion Detection Module

o Detected by analyzing and comparing the traffic patterns with normal behavior

o Packet monitoring level can be increased

Intrusion Response Module

o Response may be local to the cluster or global

71


Network Simulators

ns-2

o Utilizes discrete event-driven mechanism to simulate all kinds of activities in networks

o Four schedulers available in ns-2: linked-list, heap, calendar queue, and real-time

o Split-language programming

o Open source

o Visualization

o Support of emulation

o Support of mobility models

OPENT Modeler

QualNet

OMNeT++

72


Homework

13.1, 13.7, 13.12 (Due: Dec. 19)

Practice at home: 13.2, 13.11, 13.13, 13.20

73

ad hoc networks - hscc.cs.nthu.edu.tw

Documents