38
Ad hoc networks Lecturer: Dmitri A. Moltchanov E-mail: dmitri.moltchanov@tut.fi http://www.cs.tut.fi/kurssit/ELT-53306/

Lecturer: Dmitri A. Moltchanov E-mail: dmitri.moltchanov ... · {How: Special Interest Group (SIG): 3Com, Ericsson, IBM, Lucent ... in piconet every node can reach every other node

Embed Size (px)

Citation preview

Ad hoc networks

Lecturer: Dmitri A. Moltchanov

E-mail: [email protected]

http://www.cs.tut.fi/kurssit/ELT-53306/

ELT-53305: Wireless networks D.Moltchanov, TUT, 2014

OUTLINE:

• Some history;

• Cellular vs. ad-hoc;

• Applications of ad-hoc networks;

• Technical challenges;

• Examples of special protocols:

– Data-link layer: MAC protocol for ad hoc networks (MACA);

– Network layer: location aided routing (LAR);

– Transport layer: split TCP.

Lecture: Ad hoc networks 2

ELT-53305: Wireless networks D.Moltchanov, TUT, 2014

1. HistoryThe idea of ad-hoc networks is the multi-hop wireless relaying of messages:

• 500 B.C. Darius I, King of Persia:

– the principle of multi-hop relaying was introduced;

– messages from the capital to remote provinces have been relayed using line of shooting men;

– more than 25 time faster than normal messengers available at that time;

– a lot of other ancient ad hoc systems: string of repeaters of drums, trumpets, or horns.

• 1970, Norman Abrahamson, ALOHAnet:

– aim: a network for University of Hawaii, to connect remote sites on islands;

– idea: fixed single-hop wireless packet switching with multiple access;

– extension: idea is applicable for multi-hop relaying also;

– outcome: stimulated the research in multi-hop relaying leading PRNET project.

Packet radio network (PRNET) sponsored by Defence Advanced Research Project Agency.

Lecture: Ad hoc networks 3

ELT-53305: Wireless networks D.Moltchanov, TUT, 2014

• PRNET project:

– Aim: developing packet wireless network for military applications;

– Idea: evolved from centralized control to distributed wireless multi-hop system;

– How: ALOHA and CSMA for access to a shared media, DSSS over the channel;

– Features: self-organization, self-configuration, detection of radio connectivity.

The main issues that the RPNET faced were:

– obtaining, maintaining, and utilizing the topology information;

– error and flow control over the wireless links;

– reconfiguration of path to handle path breaks;

– processing and storage capability of nodes;

– distributed channel sharing.

The successful demonstration of PRNET proved:

– feasibility and efficiency of infrastructure-less networks;

– applicability to civilian and military purposes.

Lecture: Ad hoc networks 4

ELT-53305: Wireless networks D.Moltchanov, TUT, 2014

• Just after PRNET: Survivable Radio Network (SURAN)

– What: extension of the DARPA PRNET project;

– Aim: providing efficiency in all aspects: size of devices, cost, scalability;

• In 1980th:

– What: military applications were extensively funded across the globe;

– Point: IETF created the WG called mobile ad-hoc network (MANET) group;

– Aim of MANET: provide standardized routing functionality for ad hoc networks.

• 1994: Bluetooth

– What: short-range, low-power, low-complexity radio;

– Aim: connectivity between heterogenous devices;

– How: Special Interest Group (SIG): 3Com, Ericsson, IBM, Lucent, Motorola, Nokia etc.

– Result: one of this first commercial realization.

Note: Bluetooth is not ad hoc itself! Just a possible platform for ad hoc networks.

Lecture: Ad hoc networks 5

ELT-53305: Wireless networks D.Moltchanov, TUT, 2014

• Bluetooth standardized two major nodes’ formations:

– Piconet:

∗ single point-to-point wireless links formatted in group of nodes;

∗ in piconet every node can reach every other node in a group within a single hop.

– Scatternet:

∗ formation created by several piconets;

∗ multi-hop routing protocol should be used (not standardized).

D

S

Figure 1: Piconets and scatternets in Bluetooth.

Lecture: Ad hoc networks 6

ELT-53305: Wireless networks D.Moltchanov, TUT, 2014

2. Cellular and ad-hoc wireless networksCellular networks Ad-hoc wireless networks

Fixed infrastructure No infrastructure

Single-hop wireless links Multi-hop wireless links

Guaranteed CBR bandwidth (voice traffic) Shared radio channel (data traffic)

Initially, circuit-switched Initially, packet-switched

High cost and time of deployment Very quick and cost-effective

Reuse of frequency via channel reuse Dynamic frequency sharing

Bandwidth reservation is achieved easily Complex MAC layer

Nowadays applications: civilian, commercial Nowadays applications: military, rescue

High cost of network maintenance Maintenance operations are built-in

Low complexity of mobile devices Intelligent mobile devices are required

Widely deployed, evolves Still under development in commercial sector

Lecture: Ad hoc networks 7

ELT-53305: Wireless networks D.Moltchanov, TUT, 2014

3. Applications of ad-hoc wireless networksThere are a number of applications:

• military applications;

• collaborative and distributed computing;

• emergency and rescue operations;

• mesh networks;

• wireless sensor networks;

• hybrid cellular/ad-hoc wireless networks.

What makes ad hoc so attractive:

• quick deployment;

• inexpensive deployment and operation.

Lecture: Ad hoc networks 8

ELT-53305: Wireless networks D.Moltchanov, TUT, 2014

4. Technical challengesThere are many challenges in design, deployment, and performance of ad hoc:

• Medium access scheme;

• Routing and multicasting;

• Transport layer protocol;

• Pricing scheme;

• Quality of service provisioning;

• Security;

• Energy management;

• Addressing and service discovery;

• Scalability;

• Deployment considerations.

NO SINGLE SOLUTION TO ANY OF THESE PROBLEMS!

Lecture: Ad hoc networks 9

ELT-53305: Wireless networks D.Moltchanov, TUT, 2014

4.1. Medium access scheme

Why it is so important:

• MAC is responsible for shared use of the transmission medium;

• performance depends on MAC protocol (e.g. Token Ring vs. Ethernet).

Challenges of MAC protocol in ad-hoc wireless network are:

• Distributed operation;

• Maximum throughput;

• Minimum access delay;

• Fairness;

• Real-time traffic support;

• Power control capabilities;

• Use of directional antennas.

• Hidden terminal problem;

• Exposed terminal problem.

Lecture: Ad hoc networks 10

ELT-53305: Wireless networks D.Moltchanov, TUT, 2014

receiver senderhidden terminal

collisionpackets packets

Figure 2: Hidden terminal problem (transmission ranges are circular!).

sender receiverexposed terminal

packets

receiver

packets

Figure 3: Exposed terminal problem (transmission ranges are circular!).

Lecture: Ad hoc networks 11

ELT-53305: Wireless networks D.Moltchanov, TUT, 2014

4.2. Routing

The responsibility of any routing protocol:

• determining a feasible path to a destination based on a certain criterion;

• discovering, storing, and exchanging routing information;

• gathering information about a path breaks and updating route information accordingly.

Challenges for routing protocol in ad-hoc networks:

• Mobility;

• Bandwidth constraints;

• Resource constraints;

• Erroneous transmission medium;

• Location-dependent contention:

Lecture: Ad hoc networks 12

ELT-53305: Wireless networks D.Moltchanov, TUT, 2014

Requirements on a routing protocol in ad-hoc networks:

• Minimum route acquisition delay:

• Quick route reconfiguration:

• Loop-free routing:

• Distributed routing:

• Low overhead;

• Scalability;

• Privacy;

• Support of time-sensitive traffic.

Lecture: Ad hoc networks 13

ELT-53305: Wireless networks D.Moltchanov, TUT, 2014

4.3. Multicasting

Multicasting is an important feature in wireless ad-hoc networks:

• search and rescue operations: distribution of commands;

• military applications: distribution of commands.

Why not to adapt something from fixed networks (CBT, PIM, DVMRP):

• core base trees (CBT), distance vector multicast routing protocol (DVMRP), etc.;

• mobility of nodes changes the topology of the network! Trees are unstable!

There are following challenges in ad-hoc environment for multicasting:

• Fast recovery;

• Control overhead;

• Efficient group management;

• Scalability;

• Security.

Lecture: Ad hoc networks 14

ELT-53305: Wireless networks D.Moltchanov, TUT, 2014

4.4. Transport layer protocol

There are following major function of connection-based transport layer protocol:

• setting up and maintaining end-to-end connection;

• reliable end-to-end delivery of data packets;

• flow control;

• congestion control.

Why not to go with UDP:

• does not perform flow and congestion control and reliable end-to-end transfer;

• result: increase on contention → losses.

Performance degradation stems from:

• high error rate;

• frequent path breaks;

• presence of ’old’ routing information;

• network partitioning.

Lecture: Ad hoc networks 15

ELT-53305: Wireless networks D.Moltchanov, TUT, 2014

4.5. Quality of service provisioning

Effect of service performance determining the degree of satisfaction of a user of the service.

QoS includes a number of concepts including:

• traffic performance in the network;

• service support performance;

• service operability performance;

• service security performance.

To satisfy QoS:

• use values of traffic engineering variables that constitute the so-called Grade of Service (GoS).

Provision of QoS requires:

• negotiation between the host and a network;

• resource reservation schemes;

• priority scheduling;

• call admission control.

Lecture: Ad hoc networks 16

ELT-53305: Wireless networks D.Moltchanov, TUT, 2014

4.6. Self-organization

Self-organization is the main attractive property of ad-hoc networks.

To perform self-organization the following things are required:

• neighbor discovery:

– first phase when a node switches on;

– a node should gather network information (transmission of reception of discovery packets).

• topology organization:

– every nodes gathers information about the entire network (a part of);

– construct and maintain the network topology.

• topology reorganization:

– when links break, nodes switch off etc.

– requires periodic or aperiodic exchange of topology information.

Lecture: Ad hoc networks 17

ELT-53305: Wireless networks D.Moltchanov, TUT, 2014

4.7. Security

What makes ad hoc more vulnerable to attacks:

• lack of central coordination;

• shared wireless medium.

The attacks against ad-hoc networks are generally classified into:

• passive attacks:

– how: malicious nodes attempt to obtain information relayed in the network;

– damage: no damage to operation of the network, just capture if information.

• active attacks:

Active attacks disrupt the operation of the network and classified into the following types:

– external attacks: attacks executed by nodes outside the network;

– internal attacks: attacks executed by nodes belonging to the same network.

Lecture: Ad hoc networks 18

ELT-53305: Wireless networks D.Moltchanov, TUT, 2014

There are following security threads in ad-hoc networks:

• Denial of service:

– make the network resource unavailable for service to other nodes (i.e. flood packets).

• Resource consumption:

– consume the scarce resource in ad hoc networks including:

– Energy depletion:

to deplete the power of the node relaying the traffic through them.

– Buffer overflow:

fill the routing table with ’bad’ entries to consume the buffer space of the target node.

• Host impersonalization:

– act as an another node responding with control packets and interrupting data traffic.

• Information disclosure:

– disclose information destined for a certain node.

• Interference:

– just create some noise on a shared media!

Lecture: Ad hoc networks 19

ELT-53305: Wireless networks D.Moltchanov, TUT, 2014

4.8. Addressing and service discovery

No centralized control.

The following features are required for ad-hoc addressing scheme:

• Global unique address;

• Autoconfiguration of addresses;

• Duplicate address detection mechanism.

The following features are required for ad-hoc network to be meaningful:

• automatic service advertisement mechanism:

– should allow to identify the current location of the service;

– why current: it is not possible to assume static service locations in ad hoc networks.

• integration of service discovery protocols and routing protocols:

– may allow to easily find the necessary service in a network;

– why not so good: violate the traditional design objectives of the routing protocol.

Lecture: Ad hoc networks 20

ELT-53305: Wireless networks D.Moltchanov, TUT, 2014

4.9. Energy management

This may be done via following:

• shaping the energy discharge pattern;

• use routes with minimal total energy consumption;

• use special task scheduling schemes;

• proper handling the processor and interface devices.

Energy savings can be achieved by:

• Transmission power management;

• Battery energy management;

• Processor power management;

• Interface power management.

There will be no breakthrough in battery technology in the following decade!

Lecture: Ad hoc networks 21

ELT-53305: Wireless networks D.Moltchanov, TUT, 2014

4.10. Scalability

Testbeds and operational ad hoc networks made so far:

• contain only a limited number of nodes;

• may not be good examples of ad hoc performance.

What we may expect in real implementations:

• performance of ad-hoc network degrades drastically with the increase of the number of nodes;

• one may expect commercial realization of, at least, thousands of nodes.

4.11. Deployment

The deployment of ad-hoc network has the following benefits:

• Low cost: no cables, no configuration, no maintenance;

• Incremental: functioning starts immediately after minimum configuration is done;

• Short time: no cables, no configuration, no maintenance;

• Reconfigurability: no cables, no configuration, no maintenance.

Lecture: Ad hoc networks 22

ELT-53305: Wireless networks D.Moltchanov, TUT, 2014

5. Example: data-link/network/transport

5.1. Data-link layer: MACA

MAC protocols for ad-hoc

Contention-basedContention-based

with reservation

Contention-based

with schedulingOther protocols

- MMAC;

- MCSMA;

- PCM;

- RBAR.

- DPS;

- DWOP;

- DLPS.

synchronous asynchronous

- D-PRMA;

- CATA;

- HRMA;

- SRMA/PA;

- FPRP.

- MACA/PR;

- RTMAC.

sender initiated receiver initiated

- RI-BTMA;

- MACA-BI;

- MARCH.

multiple channelsingle channel

- MACAW;

- FAMA.

- BTMA;

- DBTMA;

- ICSMA.

Lecture: Ad hoc networks 23

ELT-53305: Wireless networks D.Moltchanov, TUT, 2014

MACA: stands for MAC protocol for ad hoc networks.

What are major facts:

• contention-based without reservation and scheduling;

• MACA was proposed as an extension for CSMA/CA protocol;

• was further extended and adopted for IEEE 802.11.

CSMA operates as follows:

• the sender sense the channel for the carrier signal;

• if the carrier is present it retries to sense the channel after some time (exp. back-off);

• if not, the sender transmits a packet.

The following shortcomings are inherent to CSMA/CA:

• −: hidden terminal problem leading to frequent collisions;

• −: exposed terminal problem leading to worse bandwidth utilization.

MACA avoids hidden and exposed terminal problems using the RTS-CTS.

Lecture: Ad hoc networks 24

ELT-53305: Wireless networks D.Moltchanov, TUT, 2014

Neighbor Sender Receiver Neighbor

RTS

CTS CTS

Data

RTS

Data

Figure 4: Packet transmission in MACA.

• RTS and CTS packets carry the expected duration of transmission;

– a node near the sender: that hearing RTS do not transmit for a time to receive CTS;

– a node near the receiver: after hearing CTS differs its transmission;

– if the neighbor hears the RTS only:, it is free to transmit.

• if the data packet is lost, a node backs off for a random period of time T ∈ {CW}, CW = 2CW .

Lecture: Ad hoc networks 25

ELT-53305: Wireless networks D.Moltchanov, TUT, 2014

5.2. Network layer: Location Aided Routing (LAR)

Routing protocols for ad-hoc networks

Routing information update Temporal information for routing

HybridReactiveProactive Past history Predictions

- DSDV;

- WRP;

- CGSR;

- STAR;

- OLSR;

- FSR;

- HSR;

- GSR.

- DSR;

- AODV;

- ABR;

- SSA;

- FORP;

- PLBR.

- CEDAR;

- ZRP;

- ZHLS.

- DSDV;

- WRP;

- STAR;

- DSR;

- AODV;

- FSR;

- HSR;

- GSR.

- FORP;

- RABR;

- LBR.

Lecture: Ad hoc networks 26

ELT-53305: Wireless networks D.Moltchanov, TUT, 2014

Routing protocols for ad-hoc networks

Utilization of specific resource Topology information

FloodingGeographicalPower-aware Flat Hierarchial

- PAR. - LAR. - DSR;

- AODV;

- ABR;

- SSA;

- FORP;

- PLBR.

- CGSR;

- FSR;

- HSR.

Proactive Reactive

- OLSR. - PLBR.

Lecture: Ad hoc networks 27

ELT-53305: Wireless networks D.Moltchanov, TUT, 2014

What are basics of LAR:

• uses the location information (assumes the availability of GPS);

• reactive (on-demand) protocol.

LAR designates two zones for selective forwarding of control packets:

• ExpectedZone:

This is a geographical zone in which the location of the terminal is predicted based on:

– location of the terminal in the past;

– mobility information of the terminal.

There is no info about previous location of the terminal the whole network is the ExpectedZone.

• RequestZone:

This is a geographical zone within which control packets are allowed to propagate:

– this area is determined by the sender of the data packet;

– control packets are forwarded by node within a RequestZone only;

– if the node is not found using the first RequestZone, the size of RequestZone is increased.

Lecture: Ad hoc networks 28

ELT-53305: Wireless networks D.Moltchanov, TUT, 2014

8

5

6

7

2

9

4

1

3

Source

Dest.

10

3

Figure 5: Example of ExpectedZone and RequestZone.

Nodes decide whether to forward or discard packets based on two algorithms:

• LAR type 1;

• LAR type 2.

Lecture: Ad hoc networks 29

ELT-53305: Wireless networks D.Moltchanov, TUT, 2014

LAR type 1 algorithm works as follows:

• the sender explicitly specifies the RequestZone in the RouteRequest packet;

• the RequestZone is the smallest rectangle that includes the source and the ExpectedZone;

• when the node is in ExpectedZone, the RequestZone is reduced to the ExpectedZone;

• if the ReouteRequest packet is received by the node within RequestZone, it forwards it.

8

5

6

7

2

9

4

1

3

Source

Dest.

10

3

Figure 6: Routing procedure in LAR type 1.

Lecture: Ad hoc networks 30

ELT-53305: Wireless networks D.Moltchanov, TUT, 2014

LAR type 2 algorithm operates as follows:

• the sender includes the distance to the source in the RouteRequest packet;

• intermediate nodes compute the distance to the destination:

– if this distance is less than the distance between source and destination packet is forwarded;

– otherwise the packet is discarded.

• distance in the packet is updated at every node with lower distance to destination.

8

5

6

7

2

9

4

1

3

Source

Dest.

10

11

12

13

14

15

16

18

17

Figure 7: Routing procedure in LAR type 2.

Lecture: Ad hoc networks 31

ELT-53305: Wireless networks D.Moltchanov, TUT, 2014

5.3. Transport-layer protocols: Split TCP

Transport layer protocols for ad-hoc wireless

networks

TCP modifications Other protocols

- ACTP;

- ATP.Split approaches

End-to-end

approaches

- Split-TCP. - TCP-ELFN;

- TCP-F;

- TCP-BuS;

- ATCP.

Lecture: Ad hoc networks 32

ELT-53305: Wireless networks D.Moltchanov, TUT, 2014

The following are two major problems with TCP:

• degradation of throughput with increase of the path length:

– short connections get more throughput than long connections.

Th

rou

gh

pu

t

Number of hops2 4 6 8

Figure 8: Throughput as a function of the number of hops.

Lecture: Ad hoc networks 33

ELT-53305: Wireless networks D.Moltchanov, TUT, 2014

• unfairness among TCP flows:

– MAC layer contention (channel capture effect):

– lengthy TCP flows – more points to contend.

Figure 9: Unfairness in TCP flows in ad-hoc networks.

Lecture: Ad hoc networks 34

ELT-53305: Wireless networks D.Moltchanov, TUT, 2014

Split-TCP provides the solution by splitting the TCP functionalities into two aims:

• congestion control;

• end-to-end reliability.

Why it is possible?

• congestion control: local phenomenon due to high contention for resources;

• end-to-end reliability: end-to-end phenomenon.

CongestionsEnd-to-end delivery

Figure 10: Congestion control and end-to-end reliability.

Lecture: Ad hoc networks 35

ELT-53305: Wireless networks D.Moltchanov, TUT, 2014

Split TCP: splits the connection into a set of concatenated TCP connections.

S

R

proxy node

segment 1 (zone 1)

segment 2 (zone 2)

Figure 11: Splitting of the TCP connection into segments (zones).

Proxy node is responsible for:

• terminating the connection from the sender/precessor proxy node;

• setting up a connection with receiver/successor node.

Lecture: Ad hoc networks 36

ELT-53305: Wireless networks D.Moltchanov, TUT, 2014

Proxy nodes are chosen using the distributed algorithm:

• simplest way: packet traversed n hops - behave as a proxy.

Transmission control at the TCP sender window is split into:

• end-to-end CW :

updated according to arrival of end-to-end ACKs.

• local CW : (local CW ≤ end-to-end CW )

updated according to arrival of local ACKs (LACKs) from the next node.

The proxy node behaves as follows:

• it maintains local CW governing transmission in a segment;

• when packet arrives from predecessor the LACK is sent back;

• arrived packet is buffered;

• the buffered packet is forwarded to the next node.

Lecture: Ad hoc networks 37

ELT-53305: Wireless networks D.Moltchanov, TUT, 2014

S

R

proxy node

segment 1 (zone 1)

segment 2 (zone 2)

TCP data flow;

LACK;

end-to-end ACK;

network link.

Figure 12: Flows in split-TCP.

Lecture: Ad hoc networks 38