Wireless & Mobile Networking

1 Polytechnic UniversityM. Veeraraghavan

Wireless & Mobile Networking

Prof. Malathi VeeraraghavanElec. & Comp. Engg. Dept/CATT

Polytechnic [email protected]

Welcome to the first class of

EL604: Wireless & Mobile Networking


Outline

• Review of basic concepts in networking– Prerequisite: A first course on networking– Communication links and switches– Types of networks– Shared links: media access control (MAC)

• How does the “wireless” dimension change the networking problem?

• How does the “mobile” dimension change the networking problem? “Mobile” vs. “Portable”


What is a communication network?

• Simplest “network” – Single link between two pieces of end-user

equipment (e.g., PC, telephone)

End-userequipment

End-userequipment

– Types of communication links• Twisted pair

• Coaxial cable

• Optical fiber

• Wireless links– Radio frequencies

– Infra-red frequencies


What is needed to send data on communication links?

• Error control– Error detection:

• Parity checks, Checksum, Cyclic Redundancy Code (CRC)

– Error correction:• ARQ (Automatic Repeat reQuest) • FEC (Forward Error Correction)

• Flow control: handles rate mismatch between sender and receiver– x-ON/x-OFF– Window based flow control– Rate based flow control


Switches

• Connect multiple links and route traffic from one link to another

End-userequipment

End-userequipment

End-userequipment

End-userequipment

Switch


Why use a switch?

• If there are N endpoints (end-user equipment), then how many links are needed for full mesh connectivity?

• How many physical links are needed if these endpoints are connected through a switch?


Answers

• Number of direct links needed to connect N nodes is

• N links – since we only need one link from an endpoint to a switch

2)1( NN


Cost of using a switch?

• Switch cost

• Can all endpoints have full connectivity at all times to all other endpoints?– Yes, with multiplexing on the links


Concept of multiplexing

• Time division multiplexing– Allows data from different sessions to be

combined at different times on to the same line– How many DS0s in a T1?

• Wavelength division multiplexing– Difference between FDM (Frequency Division

Multiplexing) and WDM?– Relation between frequency and wavelength


Answers

• 24 DS0s in a T1

• Term WDM is the same as FDM at optical frequencies – see EM spectrum chart

• Speed of light c = f: wavelength; f: frequency


Transceiver rate

• Rate of transmission and reception at endpoints and the switch– Needs to be sufficient for “full mesh”

connectivity “all the time”– e.g., if DS0s used between endpoints in full

mesh network, then T1s can be used in 25 endpoint network with a switch for full mesh connectivity


Types of switches

• Circuit switches: Position-based switching– Switch consults a table to determine output port on which to send

data bits based on their arriving position• “Position”: Interface (space), time slot and/or wavelength

– Space division switch: switch based on input interface– Time division switching: interface + time slot– Wavelength division switching: interface + wavelength– No buffers

• Packet switches: Label-based switching– Switch consults a table to determine output port on which to send the

packet based on value of label (in packet header)– Label could be changed on outgoing port or could stay the same– Have buffers to hold packets


Network of switches

• Expand 1-switch network to a multi-switch network

• Why not build one gigantic switch?– Scalability limitations

End-userequipment

End-userequipment

Switch

Switch End-userequipment

Switch


Different types of networks

• A network is defined by its “switching mode” and its “networking mode”

• Circuit switching vs. packet switching– Circuit-switching: switching based on position (space, time, ) of arriving bits

– Packet-switching: switching based on information in packet headers

• Connectionless vs. connection-oriented networking:– CL: Packets routed based on address information in headers

– CO: Connection set up (resources reserved) prior to data transfer

Packet-switching

Circuit-switching

Switching modesConnectionless Connection-oriented

Networking modes

ATM, X.25IP, SS7MPLS

IP + RSVP

Telephone network, SONET/SDH, WDM


Types of data transfers

Sending end

Consuming endLive Stored

Live

Stored

Interactive/Live streaming

Recording

Stored streaming File transfers

An application could consist of different types of data transfers

— An http session has an interactive component, but could also have a non-real-time transfer


Types of data transfers

Sending end

Consuming endLive Stored

Live

Stored

Interactive/Live streaming

Recording

Stored streaming File transfers

An application could consist of different types of data transfers

— An http session has an interactive component, but could also have a non-real-time transfer


Matching applications & networks

Data transfers

Non-real-time(stored at sender and receiver ends)

Real-time(consumed or sent live)

Interactive (two-way)(consumed and sent live)

e.g. telephony, telnet, ftp, http

Streaming (one-way)(consumed live;

sent from live or stored source)e.g. radio/TV broadcasts

Recording (one-way)(stored at receiver end;

sent from live source); e.g. Replay

Short transfers(e.g. short email)

Long transfers(e.g. large image,

audio, video or data)

Ideal networks

Connectionlessnetworks Circuit-switched

networks

Packet-switched CO networks


Congestion control

• What is it?– The purpose of a network is to allow sharing of

resources– This means if demand is high, there could be

competition for resources from multiple users– What are network resources:

• Link capacity (bandwidth)

• Switch buffer space (only in packet switches)


Congestion control

• In CO networks– Congestion control: mostly preventive

– Connection Admission Control (CAC)• Check availability of bandwidth and buffer resources before

admitting a connection

• CS CO networks: congestion will not occur once circuits are admitted

• PS CO networks: congestion can occur after connection is admitted if connection admission is based on statistical multiplexing

– Have some supplemental reactive congestion control scheme


Congestion control

• In CL networks– Have packet switches detect congestion and

send reactive messages asking sender to slow down

– e.g., datagram routers in SS7 networks send such messages; SRP (Spatial Reuse Protocol) switches in 802.17 MANs send such messages

– IP routers implement Explicit Congestion Notification (ECN) procedures


End-to-end path

• Transport protocols– Ensure reliable transfer across a communication

path consisting of many links (“zero” loss)– OR ensure delay-controlled path across a

communication path consisting of many links– Error control and flow control– Delay control (e.g., RTP)– Congestion control and connection control –

special in TCP


Applications

• Most Internet applications are client-server based

Web serverEnd-userequipment

Network

Web clients(Usually runson fixed hosts)

Network Network

End-userequipment

Email-sending clients(outlook, messenger)

Outgoingemail servers(pop, imap)

Network

End-userequipment

Email-receiving clients(outlook, messenger)

Incomingemail servers

(smtp)


Protocol Stacks

• OSI model: two more layers between AL and TL– Session layer and presentation layer

• PHY: Physical; DLL: Data Link Layer; NL: Network Layer; TL: Transport Layer; AL: Application Layer

DLL

NL

TL

AL

PHY

Endpoint Switch

NL

DLL

PHY PHY

DLL DLL

IP

TCP/UDP

AL

PHY

EndpointSwitch

NL

DLL

PHY PHY

DLL


Example protocols

• AL protocols: http, smtp, ftp, PCM voice

• TL protocols: TCP, UDP, RTP

• NL protocols: IP, ATM

• DLL protocols: PPP, HDLC

• PHY protocols: DS0, DS1

• Ethernet: PHY+DLL+NL


Functions of protocol layers

• PHY: sends bits across a link• DLL: error control and flow control on a

link• NL: switching (routing), multiplexing,

congestion control• TL: error control and flow control on an

end-to-end basis• AL: Functions specific to the application


Congestion control and connection control in TCP

• IP routers did not implement ECN until recently– TCP performs congestion control

– Senses whether network switches (routers) are congested or not

– Adjusts rate accordingly

– Slow start and congestion avoidance

• Concept of a “connection” at the TL– End hosts maintain state information regarding a TCP connection

to track sequence numbers and ACKs

– Connection open (SYN) and close (FIN) procedures

– Contrast with a “connection” at the NL, where each switch maintains state about the connection


User plane, control plane, and management plane

• Management plane: consists of all the protocols needed to “configure” data tables for the operation of the network– For example, protocols for routing data dissemination (distributed or

centralized)– Other functions: performance, fault mgmt., accounting, security

• Control plane: – Connection control protocols

• in CO networks, this includes connection setup at each switch (connections at the network layer)

• in CL networks, this includes connection setup only at the endpoints (connections at the transport layer, if the TL protocol is reliable)

– Call control protocols

• User plane: protocols for the actual flow of data


Routing protocol in all three types of networks - Phase 1

Host AHost B

I

IV V

III

II

Routing protocol

Routing protocol

Routing protocol

Dest. Next hop

III-* IVDest. Next hop

III-* III

Dest. Next hop

B B

Routing tables

• Routing protocols exchange topology/loading/reachability information

• Routes to destinations are precomputed and stored in routing tables


Signaling protocol for NL connection setup in a PS CO network - Phase 2

• Connection setup consists of each switch on the path– Route lookup for next hop node to reach destination

– CAC (Connection Admission Control) for buffer and BW

– Writing the input/output label mapping tables and programming the scheduler

Host AHost B

I

IV V

III

II

Connection setup

Connection setup

a

b

c

a

bc

d

d c

a

b

INPort /Label

OUTPort/Label

a/L1 c/L2IN

Port /LabelOUT

Port/Label

a/L2 c/L1

INPort /Label

OUTPort/Label

d/L1 b/L3Connection setup (B)

Connectionsetup

Virtual circuit


Signaling protocol for NL connection setup in a CS CO network - Phase 2

• Connection setup consists of each switch on the path– Route lookup for next hop node to reach destination– CAC (Connection Admission Control) for BW (note: no buffers)– Writing the port/timeslot/mapping table

Host AHost B

I

IV V

III

II

Connection setup

Connection setup

a

b

c

a

bc

d

d c

a

b

INPort /Timeslot

OUTPort/Timeslot

a/1 c/2IN

Port /TimeslotOUT

Port/Timeslot

a/2 c/2

INPort /Timeslot

OUTPort/Timeslot

d/2 b/1Connection setup (B)

Connectionsetup

Circuit


TL connection setup in a CL PS network - Phase 2

• Notion of transport layer connections– Exchange initial sequence numbers end-to-end to allow for ARQ

(Automatic Repeat reQuest) based error correction, i.e., retransmissions in case of errors

Host AHost B

I

IV V

III

II

Dest. Next hop

B IIDest. Next hop

B III

Dest. Next hop

B B

Routing tables

SYNSYN

ACK


User-plane packet forwarding in a PS CO network - Phase 3

• Labels are VPI/VCIs in ATM• Labels are translated from link-to-link

Host AHost B

I

IV V

III

II

a

b

c

a

bc

d

d c

a

b

L2

L1

L1

L3

INPort /Label

OUTPort/Label

a/L1 c/L2


User-plane actions in a circuit-switched network - Phase 3

• Bits arriving at switch I on time slot 1 on port a are switched to time slot 2 of port c

Host AHost B

I

IV V

III

II

a

b

c

a

bc

d

d c

a

b

1 2

1 2

1 21 2

OUTPort/Timeslot

INPort /Timeslot

a/1 c/2


User-plane packet forwarding in a CL PS network - Phase 3

• Packet headers carry destination host address (unchanged as it passes hop by hop)

• Each CL packet switch does a route lookup to determine the outgoing port/next hop node

Host AHost B

I

IV V

III

II

a

b

c

a

bc

d

d c

a

b

B

B

B

B


Addressing

• Where are endpoint addresses used:– In CL PS networks, endpoint addresses are

carried in packet headers– In CO networks, be it PS or CS, endpoint

addresses are carried in connection setup messages


Summarized addresses

• What are summarized addresses?

• Why summarize addresses?


Summarized addresses

• What are summarized addresses?– An address that represents a group of endpoint

addresses– e.g., all 212 numbers, 128.238 IP addresses

• Why summarize addresses?– Reduces routing table sizes – hold one entry for a

summarized address instead of a large number of individual addresses

– Reduces routing message lengths that convey reachability information


Examples of signaling protocols

• SS7 (Signaling System No. 7) network (with its SS7 protocol stack) carries signaling messages to set up and release circuits in a telephone network


Examples of routing protocols

• In the Internet:– Link-state routing protocols, such as Open Path

Shortest First (OSPF)– Distance-vector based routing protocols, such

as Routing Information Protocol (RIP)

• In telephone networks:– Real-Time Network Routing (RTNR)


Examples of addressing schemes

• Internet– 4-byte IP addresses

• Telephone networks– 8-byte E.164 address (telephone number)

• ATM networks– 20-byte ATM End System Address (AESA)


Broadcast links

• Wireless

• Copper: ethernet hubs

• Optical fiber: Passive star couplers

Ethernet hubor

WDM Passive Star Coupler

Blind broadcast

Ethernet switch(packet switch)

Dest: A

A


MAC protocols

• Medium Access Control (MAC) protocols are used in broadcast links to allow a node to access medium and send information

• As if “switch” is in endpoints

• Wasteful of resources because all endpoints receive all packets

End-userequipment

A

End-userequipment

B End-userequipment

C

To B

To B

B’s MAC layer checks destination address todetermine whether the packet should be “switched” to

the application or dropped

C’s MAC layer checks destination address todetermine whether the packet should be “switched” to

the application or dropped


Outline


How does the “wireless” dimension change the networking problem?

• How does the “mobile” dimension change the networking problem? “Mobile” vs. “Portable”


The “wireless” dimension

• Naturally broadcast medium

• Poor link quality

• Low power


Protocol impact

• Need a MAC protocol to share the “naturally broadcast” wireless medium

• High error rate because of the poor link quality– Need link-level error correction

• Low power causes high error rate


Outline


• How does the “wireless” dimension change the networking problem?

How does the “mobile” dimension change the networking problem? “Mobile” vs. “Portable”


Portable vs. mobile

• A portable end user device is one that can be removed easily from its network point of attachment and connected to a different point in the network– usually nodes that run “client” software not

server

• A mobile end user device is one that can be moved while engaged in a communication session


Impact of “portable” nodes

• When a node connects to some switch (router) of a network, it should be assigned an endpoint address derived from the network address of switch– Allows for routing information in the rest of the network to not require an

update – e.g., DHCP (Dynamic Host Configuraion Protocol)

@Home

Poly network

R My laptop24.3

R

get address

address: 24.3.5.10yahooweb

serverDest: yahoo; Src: 24.3.5.10

My laptop

Dest: 24.3.5.10; Src: yahoo

128.238

get address

address: 128.238.24.5Dest: yahoo; Src: 128.238.24.5

Dest: 128.238.24.5; Src: yahoo


Node: client or a server

• If the node that is portable is a client, i.e., no other host tries to reach this node, then this solution of it obtaining a temporary address works.

• But in cellular telephone network, this is not applicable because all endpoints can be “called”

• “Server” receives calls/session requests• “Client” only makes calls, initiate sessions


What happens if theserver (called end) moves?

• The server is known to all nodes in the network by some address

• If the server changes its point of connection to the network, its location should be “managed”

• This is the problem of location management.

• Any ideas for how to solve this problem?


Location management problemin IP networks

128.238.42

Host A.110R1R2

R4Host B

110.54.66 .11

Host C

Dest Next-hop128.238.42110.54.66

R2R4

R3

R5

Host A

128.238.42.110

To A

• How is the IP packet destined to A delivered?


Location management in cellular telephone networks

• How is a call delivered to A?

S1S2

S4

415-555

Dest Next-hop718-261415-555

S2S4

S3

S5

718-261-5678

718-261-5678

415-555-1234

setup(718-261-5678)718-261


A user moves while in a connection

• If a user moves while communicating – how does the network reroute data destined to

the user?– this is called handoff management

• Examples– In cellular networks, a user traveling in a car

while communicating keeps changing his connection point to the network (base station)


Cellular network handoffs

S1S2

S4

S3

S5

718-261-5678718-261

718-261-5678Have to preserve the connection as the user moves


Handoffs in a data network

• In an IP network, there is a similar need for a handoff if a mobile wireless user moves while the user’s laptop is engaged in a TCP connection (say for email download or web file download)


Big difference between these two types of handoffs

• In CO networks, every switch on the path has been configured with translation tables specifying how to route bits arriving on each connection– Handoff involves a rewriting of this data

• In CL networks such as IP, with TCP as a CO transport protocol, state information is only held at the end hosts– Handoff is a lot simpler


Mobility management

• Mobility management includes both– Handoff management– Location management


Summary

• Reviewed networking concepts• To support wireless links, need

– MAC protocols– Location management– Handoff management

• To support mobile servers on wireless links, need– Location management– Handoff management

• To support mobile clients on wired or wireless links (portables), need support for dynamically allocated addresses

Documents

Wireless & Mobile Networking