network

network

Mon 01/26/08

for tomorrow: login to moodle

-------

Tue 01/27/08

History of netwrrking

---

before networks we had sneakernet .. taking punch cards, floppy disks, or tapes from one computer to another

and we still use it today with usb drives

2) mainframes .. at this pint, although we couldn't really connect the mainfframes, we had multiple dumb terminals

and we did have to xfer data to/from dumb terminals

3) "smart terminals" .. the advent of pcs .. got intelligence built into the smart terminals

and at theis point, we needed to amke them talk to each other

.. develop protocals

.. but we've got different types of smart terminals

PCs may be running different OSes .. and we need to make them communicate

we had vendor specific protocals

if you had an ibm pc, it could talk to another ibm pc but not a TI pc

this included software and hardware

since everybody was doing their own thing, nobody could really talk to one another

so we went from the homogeneous to the heterogeneous

only like devices could talk --> all devices can talk to one another

eventually we introduced the start of the Internet

and this is when we developed the protocal tcp/ip

arpanet

Why use a protocal

protocal: a set of guidelines or rules for communication .. think of it as a language

idea of a protocal is to setup a common language .. common way to communicate with each other

syntax and symantics that we use to communicate

why use a protocal?

1) vendor / platform compatability

.. so we don't have one major conglomerate control all communication

that's the primary reason for using hte protocall

one step fiurther ..

osi model

think about: why do we have this osi model?

rules:

1) each layer can only "communicate" with the layer directly above and directly below

2) sending layers form a peer relationship with corresponding layer at reciever side

going to look like the peer layers are sending info between each other .. not going to pay attention to anything else above or below

open systems interconnect model

layers:

Application layer

Presentation

Session

Transport

Nework Layer

Data Link control

Physical

what is the responsibility of each layer?

Application layer - application itself

some peices that are written in that are part of the presentation layer .. but this is really the app

common apps:

Knightline

email

web browsers

instant messegers

ftp

ssh

spyware

and this is theepart of the code that does the work

Presentation: syntax and symantics

how are we going to accept the data .. are we going to be watching for it .. is it going to be a request/response?

syntax: translation .. between .. data representation

ascii / epsodic / unicode

CR/LF

number representation

high bit vs low bit

encryption within the application

we can see encryption throughout the network within multiple layers

..

data compression ex: audio/video feed

written in app code but part of the presentation layer

session layer:

-------------

sets up and maintains the communication

.. open socket call

establish whether simplex vs half duplex, vs full duplex

symbplex: data one way

half-duplex: can go both ways but not at the same time

full duplex: can communicate back and forth at the same time

synchronization - making sure that we're both at the same page and creating sync points

authentication via passwords . ex: logging into a ssh sever .. enter in username and password .. that's part of the session layer

Transport - provides seamless communication between two end systems

.. not two end applications but two ends systems .. so taking place at the os level

most common thing that it does is servce point addressing

.. fancy way of saying port addressing .. determines which application

look at the packet and say "is it destined for the web server or the email server" what application do we want to deliver this data to

and what applicatio is the data from

- this is where we perform segmentation and reasembly

- if you have a large file that you ftp .. you don't send it all at oonce, you send it in chunchs

but then on the other end . you need to glue theepieces back together in th right order

connection control and setup

this is where we setup the connection between the two OSes

.. so that we can pass theedata through

flow control - end-to-end

error control - end-to-end

wanna make sure that all the segments arrive at their endpoint

wanna be able to say .. is there an error in this packet

Network Layer:

- - - - - - -

- source to destination delivery

.. what takes control of figuriong out how to get our packet from its source to its destination

- network addresses/ addressing

.. IP address in TCP/IP

and routing: figuring out based on the network addresses how to get to my destination

data link control

- communication between two directly connected nodes.

two nodes on the same local area network .. two nodes that are only separated by a switch or a hub

as part of the data link control we're gonna layout framing .. organizing the bits into frames

data stream of 0 and 1s .. how do we organize them into frames

physical addressing - MAC addresses .. hardware addresses that are burned into the nic .. idea is that they're relatively unique

.. as long as no so end up on the same lan, we're fine

flow control - making sure that our receiver is not getting overwhelmed

error control -

media access control - making sure everbody takes turns appropriately

physical layer -

electrical and mechanical specifications for transporting bits

how do we encode a 0 / 1

things that are important:

the data rate .. how many bits are we sending per second

block sychronization if we're using a syncronized tehcnoology

- physical trates of the media - what bandwidth it happenes to operate on .. what leve is 0 and what level is 1

physical topology .. layount of the network .. are devices connected to one central point? in a circle?

transmission mode - symplex, full dplex, or half duplex

you see a lot of things that reoccur throughout the osi model

finally, the encoding method for the zeros and 1s

how do we physicall represent a 0 and a 1

can't peak at a layer that is not above and below?

why are these rules in place?

sets up these very pseicif rules that everyone needs to follow

and error checking at multiple layers .. really important

and we have those extra chances to make sure the data gets through correctly

.. but the biggest thing is: if I change from wireless to wired .. we're changing the bottom two layers .. but the browser doesn't care

.. we can make changes in one layer and it only affects the layers above and below

.. we know technology changes over time .. this makes it a lot easier for us to replace one technology with another

why use this layered model?

1) if we change a layer, we only affect layers directly above and below

2) vendor compatibility - a lot easier to provide the vendor compatibility when you think .. os is really only at transport layer

.. on the network / data link layer .. i don't care what os .. just make it the same down here

3) easier to understand - taking this insane process .. incredibly complex! .. by breaking it into these 7 steps .. it makes it more manageable

easier troubleshooting: makes it easier to troubleshoot problems .. you know if there's a problem with network addressing .. you only need to examine the network layer

the physical media cant be messing it up .. doesn't have access

easier to troubleshoot whe weecan zero in on one layer

5) design and testing stages -

when we're implementing a new technology that has its own layers .. we can test to make sure in one module.

can write and test each individual layer

-------

Thu 01/29/08

TcP/ IP was invented by DARPA .. agency that had their scientists working on research for wars

what they wanted to do was establish a research network

.. back in the day, they were sending data just throug mail1

wanted a network taat was real time .. realatively real time .. within an hour .. not instantaneous

tcp/ip .. is a packet switching network

originally referred to as the arpanet

.. really was the start of the internet .. and the internet at one time was controled by a central agency -- the defense department

this outshoot explanded into the internet we know today

very first offshoot of the internet

salt lake city

burkley

and some other california university

the very first connection was between these three

and initially justa thoe couple .. all connected together

first developed: 1969.

should have an idea of did it happen in the 60s , 80s or 30s but don't worry about the date

tcp/ip model itself

- tcp/ip was developed before the osi model

not possible to actually implement the osi model .. but it was based on the tcp/ip

osi:

application

presentation

session

transport

network

data link control

pphysical

hw do they compare to the tcp/ip

the data link is going to be the same

the tcp model does not specify the physical .. it uses what's there ex: 802.11 or eathernet

it sits on top of that

tcp/ip has it's own netowrk model and the responsibility of the tcp/ip network model match up with osi

.. routing etc

tcp/ip transport layer

- flow control error control , connectoon control and setup

what allowed the operating systees to initiate the connection .. and port addressing

this is where the tcp port come in

and then at the top .. in the world of tcp/ip we just call it -- the top 3 -- the applicaton layer

.. implements the roles of presentation and session in the application

network .. is going to be your network devices

at the top, a message coming out of the application

at the transport layer, going to attach a header to that message .. include things such as the poot addressing, some error checking

and the thing in it's entirety is a user data gram

at the network layer, more headers

.. and now we're just ognna call it the data gram .. or just a packet

at the data link control layer, we are going to add a header to our diagram as well as a footer / trailer

what we have at theeend of the dta link layer is something that allows us to do error checking

when we put the entire thing together, we call it a frame

at the pysical layer, we just refer to it as bits

process of adding and subtracting all these headers and footers is what we call encapsalation .. and that's what really creates the effect of only being able to look at above and below

the protocal is essentially not allowed to look at the user datagram header .. as far as the network layer is concerned, it's just 0's and 1s

examples of tcp/ip:

apps: web, ftp, smtp, ssh, telnet, pop, dhcp, dns

transport layer:

- tcp -> process by which we send packets when we want to make sure they get there, udp

network: IP, which includes smaller protocals -> ARP .. address resolution protocal

ips mapped to mac addresses

rarp - reverse address resolution protocal .. not used much

icmp, igmp - management protocals for ip

.. "no route to host"

Transmission modes: defines the flow of a connecton

simplex: - one way .. unidirectional

one sender, one receiver

half duplex:

both devices have send/receive capabilities / can only go one direction at a time.

.. like a walkie talkie

.. up until about 7 years ago, half duplex was the standard

nowaways, the vast majority of our connections are full duplex .. both devices can send andreceieve at the same time

50Mbs full duplex

vs

100Mbs half duplex

.. at some point, you're going to be waiting for acknowledgements .. and it's better if you don't have to take a break and wait for those negociations

whenever you can, force to full duplex .. or half duplex

network categories

how do we categorize our networks?

four primary categorizations:

lan - local area network

.. what makes something a local area network

- single building or close proximity cluster of buildings

- typically high data rates

- same topology .. if we have token ring and eathernet , they can't be part of the same lan

- multi point architecture .. not a point to point connecton .. more than two devices

MAN - metropolitan area network -

- can span a city

these aren't really exact .. limits, btw

these can be publically or privately owned

often, cities have their own network for their government ops

lower data rates than lans .. 100Mb or below .. 10mb probably more common currently

examples:

city government

college campus

cable company

company officies .. within a city

WAN - wide area network

- can span the globe

typically wide area networks utilize public infrastructure

.. typically not setup by private companies .. leased from telco / satellite provider

slower speeds .. latency becomes an issue (ie, how long it takes to get a response)

LAN, the response is virtually instant, but over a wan, you start to noticde the latency of responses

PAN - Personal area network

- within a room

- speeds vary . if using bluetooth you may be able to get relatively high sppeds

if you use zigbee - that's a low data rate protocal

- typically wireless

- less than 30 devices

- typically ad hoc networks -> you don't have an administrator that sets it up .. just devices that find each other

examples: bluetooth, zigbee, infrared

topologies:

.. the general shape or configuration of connections in the network

simple example - mesh network .. all devices connect to all other devices

so for 4 devices, each has 3 connections

so for n devices, n(n-1) connections

not practical for a lab .. too many wires

what happens if one of these connections fails .. in a mesh network, nothing happens .. we can still get there

what happens if one of the devices fails .. well everyone can still talk to each other

so realability : awesome!

cost - horrible

star topology:

hub in the middle and devices hanging off the hub

* <-- that's the picture

n connections .. one per device

.. what happens if weelose a connecion -- only lose one device

if we lose a device, same thing .. only that device

but if we lose the hub, we're screwed

.. hub is our single point of failure

.. but there ar3e many waysthat we can make redondant hubs

-------

Fri 01/30/08

topologies:

mesh, star covered yesterday

mesh .. reliable, but too many connections

star: .. single point of failure .. the hub in the middle

but one connection pr device .. that's the advantage

tree -

star .. hub with all devices hanging off the hub

with a tree, you have multiple hub levels .. and devices hanging off of it

it's just a variation of a star .. everything else still holds .. the hubs are still single pointws of failure

ring topology ..

devices connected in a ring . each device has two connecdtions

connected in a closed loop

advantages: - if we have connection failure, wont disrupt connectivity f two-directional ring .. and you'd have to implement that by having one ring of connectons that go clockwise and one that goes counter-clockwise .. and that doubles the number of connections

disadvantages:

how does this work - FIDDI .. fiber loop

the idea here: if your connection breaks, these devices that now become endpoints essentially loop .. so we've still got a ring, but it just passes through the outside devices once

typically sending data one way

very easy to add and remove devices ..

generally goes in one direction at a time

if device 2 wants to send to device 1, it has to go all the way through 3 and 4 to get to device 1 .. so not necessarily the most efficient

one or two connections per device

if we lose a device - if it's one direction, the whole thing stops working

if we have a bidirectional ring - doesn't break, we'll start using both directions, the devices on the ends of the broken devices loop back.

token ring - you have what's called the token .. imaginary thing

whoever holds theetoken controlls the network -- can send data

so if device 1 has the token, passes it to 2 on the way to 3

3 (recever) may send an ack .. which get's bpassed on to 4 and then back to 1

1 then passes the token to 2

the beauty of the token ring setup is that it's fair .. everyone gets a turn becaase you just keepp passing hte token .

eathernet is not always fair

and the last topology is the bus topology

- logically what eathernet is

[] []

| |

*-----------------------*

| |

[] []

back in the day, there was just this cable that you clamped into with terminators on each end.

how does it compare?

loos of a device has no effect on the network

when we lose a connection .. if we lose the part to the device .. no problem

but if we lose a connection in the main line .. depending on how cleanly we lose our connecton, everyone else might still be able to communicate -- we just divide the network

if we break it in a manner that is unclean .. just partially broken .. that can garble everything

interfereance .. idea that if there's one bad bit , you throw out the whole packet

incredibly easy to add/remove devices

.. we used to just clamp them on in there

the big problem -> really a free for all ..

in token ring, we had a nice way of taking turns

in a bus topology , everyone is fighting to use the bus .. and we meed some algorithms to organize the caos

.. so cooridnation of who gets access

in the star , the hub took care of it .. but with the bus, everyone is just fighting

Ethernet -

--------

logically a bus topology

and used to be so physically as well

.. but now they look (physically) more like a tree or a star

types of signals:

analog and digital

analog - continous line/ infinite number of points

digital - discrete separate points

|. . . .

| . . . .

----------

analog has more data .. has more points

so when we convert, we're always going to lose some data

analog: voice, light waves, radio

digital: 0's and 1's

Analog signal

- sine waves

- 3 characteristics

- 1) Phase .. position i the sine wave relative to time 0

360 degrees

0 90, 180, 270, 360

express the phase in degress

so if the phase is 90 deg, it starts at the top at time 0

180 deg phase shift, tarts at 0 but goes down first

2) Amplitude - the hight/strength of the signal

(y access has amplitude)

can be measured in volts, amperes or current

3) frequency or period.

on our time axis,

the frequency is the number of sine waves per second.

the period . the time to complete one sine wave

frequency is measured in hurtz

if we can fit two sine waves into 1 second, we have a 2 hz frequency

the peroid, then is .5 second

1/period = frequency

more data = higher frequ3ency or lower period

two ways to talk about them

we're looking at a time domain graph

can also look at a frequency domain graph

not going to get into composite waves

frequency domain graph .. frequencies on the x axis

y axis still has amplitude | |

| | |

| | |

--------------------

1hz 2hz 3hz

spectrum of a signal - basically the individual frequencies

.. listing of the different frequencies that the signal is made of

bandwidth .. tow sidderent types .. thise is the bandwidth of an analog signal

.. the highest frequency - the lowest frequency

so ex: 2, 3, 5 hz spectrum

bandwidth 5-2 hz

bandwidth of a voice grade line is from 0 to 3,000hz

sine wave .

5hz

period is .2

phase 0 degrees

amplitude - 2 volts

draw a sine wave with peroid of two seconds

phase shift of 270 degrees, 6 voltz

make sure you can convert between peroid and frequency

digital signals -

- discrete

- bit intervol .. time to measure one bit

... essentially equivalent to period

bit rate - the number of bits per second - similar to frequency

bps

how do we convert?

digital to digital - take data bits enconded into a digital signal with voltage levels

analog data ex: voice to digital

.. the process .. pcm -> voice over digital signal

digital to analog . digital data of zero's and ones's into sine waves

- computer modems

- fiber (to ligth waves)

analog to analog -- need to econde analog data ont an analog sygnal .. frequency modulation

- radio station

-------

Mon 02/02/08

how do we encode data?

digital data / digital signals

---------------

convert computer language (0's and 1's) into voltages

different ways that we can do this

ex: ethernet

coaxial cable

printer

usb

unipolar

-------

simplest form

1 is a positive voltage and 0 is zero volatage

1 bit = 1 symbol

the voltage is constant throughout the bit interval

example: string of bits

0011000000001010

what will it look like?

| ** *

|**--********-----

00++00000000+

bar like graphs .. becaase the voltage is constant throughout the interval

.. we can't really go straight from 0 to 1 .. but we'll pretend we can

advantages

----------

- don't need to deal with negative voltage

- simple! 1 is voltage, 0 is no voltage .. as simple as we can get .. that's the biggest advantage

disadvantage

------------

- is a DC component .. the average voltage is not 0

... so some media wouldn't be able to deal with this encoding scheme

- synchronization issues here with long strings of either 0 or 1

.. constant positive voltage .. don't know when I'm having a problem or just receiving lots of 1s

and we also don't know when to sense .. need to know if we have two ones or 3 ones.

so lets do better

.. get rid of dc component

Non return to zero levle (NRZ-L)

voltage is constant during a bit interval

- 0's + voltage

- 1's - voltage

so our string becomes

++--++++++++-+-+

.. drawn the same way as before (except different center of x axis :-D)

advantages:

----------

- no DC component - avg voltange is 0

- zero signals a problem

- still pretty simple!

disadvantages:

-------------

- long strrings of 0's and 1's are still a synchronization issue

non-return to zero level inverted - (NRZI)

- voltage constant during the bit interval

- 0 no transition at the start of the bit interval

- 1 transition at start of bit interval

.. start negative .. put it in paren because not really a bit

(-)--+---------++--

.. change the sign when you hit a 1

advantage:

---------

- long string of 1's no longer has synchronization issues

- still pretty simple

disadvantage:

------------

- we can no longer look at one bit at a time

- long string of zero's is still a problem

bipolar AMI (alternate mark inversion)

- 0 zero voltage

- 1 positive or negative voltage alternating

- start position

(+)00-+00000000-0+0

adv:

---

- long string of 1's is no longer a sync issue

- each bit can be determined individually

- some built in error checking ..

disadvantage:

------------

- strings of 0 still an issue

- loss of all voltage not detected

- slighty more complex\.. now we need to remember the previous 1 bit

bipolar ami with 8 zero substition (B8ZS)

bipolar ami scheme but a string of 8 0's is replaced with a string with 2 code violations

start out pos

(+)00000000

(+)000+-0-+

start out neg

(-)000-+0+-

that specific violation is interpeted as 8 zero's

(+)00-+000+-0-+-0+0

*T1 lines use this encoding scheme

Advantages:

- could detect a complete loss of voltage

- a long string of zero's .. no longer a sync problem (1's either)

- some built in error check (slightly diminished .. because we introduce our two specific violations)

disadvantage:

------------

more complex / less clean

.. gotta keep track of our last 1 .. and gotta keep track of 8 bits

Psuedotenary

------------

- 0 alternating + or -

- 1 represented with 0 voltage

start positive

(+)-+00-+-+-+-+0-0+

adv -

----

long string of 0's not a sync problem

- some built in error check

disadv -

------

- long string of 1's is a sync problem

- slightly more complex

essentially opposite of bipolar ami

Manchester encoding

- 0 is represented by a transition from positive to negative .. in the middle of a bit interval

ie, two transitions per bit interval

- 1 transition from minus to plus in the middle of the bit interval

+--+-++-+-+-+-+-+-+-+--++--++-

adv:

- built in sync .. garenteed a transition in the middle of the bit interval.

- error checking built in .. if we dn't experience a transition, I don't have an error

- zero voltage no longer a problem

- each bit is determined by just the two symbols

disadvantage:

-------------

twice as long .. cutting in half our ability to send data; two changes per bit

differential manchestor:

-----------------------

- always have a transition at mid bit interval

but 0 is prepresented with a transiton at the start of our bit interval

- 1 .. no transition at the start of the bit interval

0011000000001010

(+)-+-++-+

.. -+-+-+-+ .. +-

adv:

tranisitions

error checking

dis:

two times the modulation rate

examples: .. ethernet manchestor

diff manchestor is used in a token ring network

-------

Tue 02/03/08

multi-level schemes

-------------------

2B1Q .. 2 bits, 4 levels

standard .. 1bit, 2 levels

2B1Q is used in DSL

transmission imparement

- 1) attenuation -- loss of energy .. along the way

.. strengthh of teh signal erodes over distance

.. either resistance of rmedium or going through the air

what can we do to combat it .. use an amplifier .. strengthens our signal and allow it to travel a longer distance

2) distortion - many of the analog signals ans well as digital

analog are frequently made up of multiple signals .. all of them have different delay characteristics .. so when you signal gets to final destination , because of different delays, singla can change form

(different frequencies have different delays)

3) noise (interference) - - cross talk .. other media that is also communicating, our singals may jump

b) EMI - electro-magnetic interference

used to have to keep wiring away from a/c in server room

it's always there .. we just hope it's not too big of a problem

c) Impulse noise - (lightning, power surges)

given a perfect sine wave

plus noise

.. add it on to the sine wave

the more tolerant of noise you need to be, the less you can encode

more terms-

throughput .. how many bits can you push through a single point in one second

how fat is your pipe?

different than latency! .. how long does it take to reach your destination? .. the time it takes a bit to go from source to destination

.. also refered to as proprogation top

jitter -- variation in latency

often better off with a larger latency than one that has a lot of jitter

analog data to analog signal:

three types of encoding

- amplltude modulation

- add signal you want to send to carrier signal

carrier signal, at carrier frequency .. looks like a sine wave

add them together to get -

** need to draw

to draw: draw carrier signal in dots and add data signal

ex: am radio

speach and music is 5Khz

just voice is 3khz

bandwddth required is twice the range of the modulating signal

so an am radio station requires 10Khz .. because the signal we want to send is 5khz

530-1700khz

need to allocate 10khz per signal

frequency modulation

.. how many times we can get the sine wave in in a second

- modify the frequency to indicate changes in amplitude

start out with a data signal

need to somehhow add that to carrier signal

carrier signal is ging to bbe at a sepcified frequency .. change it simehow to represent altitude

.. when I go up, higher frequency

as I go into negative voltage, decrease frequency

ex: FM Radio

- bandwidth is 10x the signal being transmitted .. significatnly less efficient

5Khz = 50khz

because of stereo, we essentially are sending a signal that's 15 khz, so we actually need 150khz bandwidth for each fm station

we also add a 50 khz between stations

range 88 to 108

88.1 ... 88.3

3) Phase modulation

digital data to analog:

1) amplitue shift keying

2) frequency shift keying

3) phase shift keying

in many respects, digital to analog is easier becaase we have discrete data

modulation/demodulation is the process of converting between digital data and an analog signal

modem

fiber is an analog signal, too

baud rate - number of signal units per second aka symbols per second

number of symbols per second will help us determine our bandwidht

but the bandwidth is our bit rade and the bit rate is the number of bits per second

sometimes equal sometimes not

baud rate - bit rate/divided by bits per symbol

bit rate = baud rate * bits per symbol

anology that is always useed: .. bauid rate as being a car and the bit rate is people in the car

for example:

4 bits in each symbol, and a baud rate of 2400, our bit rate is 2400 * 4 = 9600

highest baud rate to achieve over phone is 2400 baud

... how did we get faster?

9600 baud but more bits per baud

56 k is 6 bits per baud

-------

The 02/05/08

digital data and analog singals

- Amplitude shift key

- phase shift key

- frequency shift key

ASK

---

vary the stength / amplitude of the carrier signal to represent 0 or 1

1 may be represented by a sine wave represented by the formula

Asin(2π)... fix it .. where A1 may equal 5 vol

A2 may equal 10 vol

see attached

typically more than 1 sine wave per bit

baud rate: 5

bit rate: 5

frequency:15

baud and bit match .. incoding 1 baud per bit

disadvantages of ask:

amplitude is highly susecptable to noise and interference

because it's highly suseptable to interference, we typically only encode 1 bit per baud

advantage: simple

special type of ask called ook

on/off keying

1 = s(t) = A1sin)2π ...

0 = no voltage

see attached.

used in fiber optic

it's possible to do more, but we usually don't .. becaase of the noise factor

frequency shift keying

----------------------

binary values represented by 2 frequencies near the carier frequency

1 = s(t) = Asin(2π+f[1]t)0 = s(t) = Asin(2π+f[2]t)

example: frquency of 4 for 1

frequency of 8 for 0

see attached.

less suseptible to soise .. easier to tell peaks and troughs .. vs tring to descern the differences in amp

.. so this is less susceptable to noise than ask

big disadvantage - more bandwidth required

multi-level FSK

ex: encoding two beits into 1 baud

0 0 freq1 (1)

0 1 freq2 (2)

1 0 freq3 (3)

1 1 freq4 (4)

010011

see attached.

lets go wiith 4, 8, 12, and 16

baud rate: 4

bit rate: 8 .. 2 bits per baud

it is possible to do multi-level fsk . extends the bandwdith we need

a lat easier to tell how many waves rather taan different amp

our last one:

phase shift keying

.. shifting the phase to represent 0's and 1's

1 may be represented by:

= A sin(2π+ft + π) // π is phase shift

...

and then 0's may be just the standard carrier signal

= A

sin(2π+ft)

see attached!!!!

advantage of this? less susecptible to noise than ask

we do frequently have mult-level phase shift key

00 - 0

01 - 90

10 - 180

11 - 270

bit rate: 2* baud rate

you can definitely do more than just the 4 phases

QAM------

Quaiatude ampliturde moluuation

.. combination of ask and psk

.. helps increase the bit/baud ratio

00 could be a wave that looks like:

a1 amplitude and no phase shift

01 could be:

a1 amplitude but 180 degree phase shift

10 :

a2 amplitude and no phase shift

11 :

a2 amplitude and 180 phase shift

00 11 01

frequency 9

see attached.

constellation graph

attached.

dots on graph

amplitude by distance from the origin

phase - represented as the angle

16 dots -> 16 QAM

2 amplitudes and 8 phase shifts

16 qam

baud 9600

what is bit rate:

... 4 bits represented

9600 * 4.

38,400

2,000 baud

bit rate?

ook - 2,000 bps

binary fsk - 2,000

8 qam - 6,000

q fsk (ie, 4 values) - 4,000 (2 * 2,000)

8,000 bps using 4qam

baud rate: 4,000

analog signal -> data loss.

-------

Fri 02/06/08

discussion questins due monday

respond to online discussion!

analog data and digital signals

- going from continuous data, which has infinite values to discrete

.. we're garrenteed to have data loss

we're going to sample the continous data at set intervals and encode those values

the process we're using is called PCM .. pluse code modulation

four phases

- PAM .. pulse amplitude modification -- sample our data

- process quantization

- binary encoding - got the values now how do we encode them into binary data

- perform a digital to digital encoding .. could be any of them.

1) PAM - meausre the amplitude of the original signals at equal intervals

- we still have infinite values to represent .. not an infinite number of them, but the values themselves are infinite

given the attached graph

.. we sample at different points

.. at scheduled invervals, I"m gong to sample the value.. but I'm still going to have an infinite number of values

.. when we sample, we only do so to a certain accuracy

.. anything that goes on inbetween we completely miss

what determines our accuracy? our accuracy here is determined by the sampling rate.

the more frequently we sample, the more accurate our PAM is going to be

notes on accuracy:

according to Nyquest's theorum, in ourder to have a reasonable level of accuracy, we need to sample at 2x the highest freqency

voice goes from 0 to 4,000 HZ .. we need a sampling rate of 8,000 per second

translates into a lot of bandwidth

bits per sample ( really part of binary encoding)

(log based 2 of n) where n are the number of values for binary encoding

bit rate = samples per second * bits per sample

if we sample too infrequently, we're gonna have a mess when we reconstruct

the more frequntly we sample, the more accurate the PAM is going to be.

2) quantization:

match our PAM values to descrete values within a range.

so, take what we had from our PAM and going to round them to the nearest whole numbers

so now we lose all the data between our y values .. because we round to the nearest value

so here we lose accuracy on the y axis as we "round" to nearest value

Stage 3 .. binary encoding

assign binary values to each level

16 values -> 4 bits

how we assign -- it doesn't really matter .. we undo it on the other end

more efficient to just assign numbers from 1 to 16

no data loss at this stage! .. but we've already lost a lot of data

last stage:

digital to digital encoding - any method

nrz-l

0 +, 1 -

666665

6 -> 1101 -> ++-+

other end:

decode

first, we block into 4's because we needed four bits

digital to digital to binary

kinda lite our original wave .. depending on the numbers of samples

suppose we want to compress the bandwdith

.. lower the bit rate

.. we can reduce how many samples per second

.. we can also reduce the number of bits needed per sample

tecniques:

differential PCM:

instead of recording the full value each time, we're simply going to record the difference from previous sample

extreme example: delta modulation

.. simply record increase or decrease in the amplitude of the signal

up two upp two down 1 down 2

usually not a zero .. usually have to go either up or down

delta modulation .. simply record + or -

allows us to be more efficient with a lot less loss

transmission modes

- parallel -- one wire per bit and all bits are sent at once

.. common example is the bus in your computer

32bit bus -> 32 wires

serial transmition: .. 1 wire , one bit at a time

parallel .. advantages: it's faster because we send all bits at once

in some respects simpler -- don't have to take each bit individually and put them in theour appropriateeplace

disadvantage: need n wires, which isn't always practicle.

where do we see?

system bus

parallel ports

typically only a short distance

serial advantage: one wire! - simpler

disadvantage - 1 wire .,. lower bandwidth

breaknng into packet .. where doessone begin?

2 types of serial transmission:

simplest of the two:

asychroneous .. timing not important

.. bits are grouped into frames

send a stop bit at the end of the frame .. ex: 1

and continue sending stop bit until more data arrives

how do we know when more data arrives ? gonna set a start bit at the start

frame can be a fixed leength .. mostly the case

typically a minimum number of 1's we have to send between packets

advantages: easy .. whenever we see 0, grab the next frame .. and we go back to just waiting.

sychronization time is not required.

disadvantage: very slow:

a lot of overhead is getting in our way

.. slows us down .. really decrease our bandwidth

in ourder to send 8 bits, I have to have at least a 0 and 3 frames on the end .. taat's from overload

typically used for low band, where we're not worried about the extra load

sychorious -- relies on sychronized timeing SCL between different devices.

a common way of doing this - a syrchornization line

timing/idle bit of 1010101

see whwere the transitions are made and make sure tht I"m sensing at the right time periods

advatnage: - significantly reduce overhead .. don't need the start bit s and stop bits

.. we can eeave the midia idol .. and we can start sendding when we want to

disadvantage: - maintaine of timing .. need to make sure all devices are on the same time

- typically used for higher speed requirements (bandwidth ) .. worried about conserving it and getting as much out of it as we can

post discussion question!

-------

Mon 02/09/08

Multiplexing

------------

signals, smush, adn then pull them apart at the end of the line

combine many smaller data streams into one larger data stream

and then at the end - demultiplex .. the process of extracting the original, smaller data streams from one large data stream

data streams --> mux --> large data stream --> end --> demux --> extract out data stream 1, 2, 3

so to the devices on the other end, they should be completely oblivious that tis even happens

path -- physical path between te mux and demux

within the path, wee have what are called our channels.

essentially one channel per data stream

channel carries data between a specified pair of devices

m channels within the path

3 types of multiplexing

-----------------------

(1) frequency division multiplexing (FDM)

(2) wave divison multiplexing .. really just fdm using light waves (WDM)

(3) time division multipplexing , a whole nother animal (TDM)

Frequency division multplexing-

modulated carrier frequency to combine signals

simotaneously transmit at various frequencies

ex) FM radio - each radio station has their own frequency-- all transimitting at the same time, TV stations

- the bandwidth of the paht must be greater than the sum of channel bandwidths

- channels must be separated by some unused bandwidth to prevent interference

3 data streams

-----------

cf1

-----

buffer

-------

cf2

---

buffer

----

cf3

-----

individual lines .. channels ..

while thing is the path

10 mhz channel, 2mhz buffer

ultimate bandwidth .. 34Mhz

low bandwidth - Base band - uses only one carrier frequency, thus -- not multiplexing .. just a simple signal

high bandwidth - broadband - large specturm of frequencies -

wave division multiplexing:

multiple light waves going accross a single fiber

.. thin strand of glass or plastic

mux demux

/\ /\

__ __

.. prism! .. puts the light together and takes them apart.

.. integrity of the physical path is very important -- when we're talking fiber optics .. break in the glass is terrible

angles very particular

.. we tend not to need the buffer .. no interference.

.. frequencies that are exact multiples of each other can also interfere with each other

contributes to loss of efficiency

example: fiber optic, sonic rings

time divison multiplexing

------------------------

senders take turns sending on media

might do round robin

two different types

sychroneous TDM -

---------------

same time allotment for all devices

- we can utilize full bandwidth during one slot

---|m|

---|u| -- channel --

---|x|

in between .. divide each timeslot into n slots

[a][b][z][a][nothing][z]

process continues on

eventually we reach a pont where we have an a to send and devices 2 and n have nothig to send

advantages: simple, fair, efficient--> if channels have same or similar bandwidth

.. so if all devices are sending data at the same rate, this is a very fair way to send our data

the disadvantage: wasted bandwdith if devices have varied data generation rates/bandwidth

.. we can fix that a little

if a is producing data at twice the rate, he can get two slots per time period

modificaton to syc TDM

---------------

a a b ... z

this is good if a has 2 times the data generation rate of other devices

bit stuffing:

------------

- 10 Mbps, 10, 20, 17 mbps

1 2 3 3

device 4 .. can have tww time slots and essentially fill that extra 3 mbps with just 0's or 1's

so here, we essentially make this 20 mbps with adding zero's

mux and demux need to have this coordinated so that the demux knows what to chop off

timeslots have to be the same but not all fill fo balid data

but we still have the problem that the devicesshave to send data at a steady rate.

.. but we send bursty traffic

.. and this is bad for bursty traffic

and even our bit stuffing is still bad for birsty traffic

other approach -

asychroneous TDM

in each timeslot, we need to ID which device belongs to each bit of data

statistical TDM

timeslots are filled by any device who has data to send

- must specify which device each slot is destined for

ex:

4 devices .. send 8 bytes at a time

each one of my slots is 8 bytes .. but we need a header . that needs to be 2 bits in this case -- four devices

with this , we're creating an inefficiency

64 bits of data + 2 bits overhead = 66 bits

64 = 64/66 efficient

device 1 aaaaaaaa

device 2 b

device 3 cc

device 4 dddd

we'll do 1 byte per slot, 2 bits of overhead

00-a 01-b 10-c 11-d

that goes over to demux and it decides what goes where

00-a 10-c 11-d 00-a

....

advantages and disadvantages:

- always using the slots .. in that way, this is efficient .. for bursty traffic

disadvantage:

-------------

- headers themselves are inefficient, specifically if we have steady rates of traffic

slot sizes are usually larger than synchroneous slots

.. because each slot has a header .. sothe more data we can associate with a header, the less overhead

but we can have wasted space if we don't fill the wle slot

.. overall bandwidth does NOT need to exceed the sum of bandwidths .. so that's the other advantage: path bandwidth does not need to exceed sum of channel bandwidth

-------

Tue 02/10/08

spread spectrum:

--------

send data by jumping between frequencies in a set pattern

FHSS - frequency hopping spread spectrum

send one bit per frequency

send another bit , switch frequency

f1, f2, f3

and want to send 10111

send first bit at frequency 1, 0 at frequency 2

there's going to be a pattern to this .. the pattern is not necessarily in order.

keep repeating the pattern .. and to make things interesting, I coul also have another device

that wants to send something

device 2 may be using f2, f3, f1

so I can have both of the devices happen at the same time as long as their patterns dont overlap

helps with interference and eavs dropping

constantly changing freqencies, so if one freqency has a lot of interference, depending on how we aare enconding things, you may still be able to recover the signal

if we have just one frequency that has a lot of interference on it, but w'ell have a lot more than 3 freqencies .. if using ecc, we can survive that

eavs dropping -- if you're just watching one frequency, you won't get intelligble data

you'd need to know the negotiated pattern ... you can't intercept data without the frequency hopping pattern

.. originally developped to make jamming and evesdropping more difficult

.. eliminates problems of jamming

.. you'd have to jam all the frequencies

another type of spread spectrum .. direct sequence spread spectrum (DSSS)

- encode a single bit onto multiple frequencies

If i want to send a 0

f2f4f3f1

in order to send the 0, I"m going to start sending and I"ll send it at f2f4f3f1 .. sending pieces of the zero to all four places

these four bits put together form a chip sequence

advantage: - can recover from interference

- disadvantage: less efficient

another device can fill in the pattern

speed of information divided by number of frequencies .. gotta do four of these just for one bit

chapter 9(?)

telco service

.. business class

leased lines from the telco

cost of these depends on the distance of hte lease line and the speed

sometimes referred to as just t1 lines

at any rate -- t lines

requres csu and dsu devices

channel service unit

data service unit

.. terminating devices

your side of the csu/dsu or their side?

the csu deals with maintenence of the channel

- creates the connection to the other side

idea is that anything that happens between the csu and csu is out of your control.

provides the ability to do a loopback

.. takes the line and connects it to yourself .. a good way to tell if its your probbem or the telco's

the csu prevents excessive 1 bits

.. uses and encoding technique that can prevent excessive 1s

dsu .. translates between digital representations

T lines -

t1 .. ds-1 signal

bit rate on the t1 is 1.544 Mbps

essentially 24 T0 lines

T0 - really just a plane telephone line .. 1 voice quality, 64k connection

each frame will have 8 bits of data plus 9 sync

8,000 frames per T0

+ additional 8k of overhead .. tells us how they have the lines broken out

you can use the T1 for 24 voice or 1.544 mbps data service

t2 - don't hear about

t3 - uses a DS-3 signal .. bit rate here: 44.736 Mbps

.. eesssentially the equivalent of 672 T0 lines

1.368 Mb of overhead - wow.

.. not the most efficient

.. but these have been around for years and years and years

run over the standard twisted pair lines

these opperate via time division multiplexing

Optical cirucits from the phone company

- leased lines from the telco that run over optical fiber

what do our signals look like here?

carrier signal bit rate # voice channels

oc1 sts1 51.84Mbps (50Mbps data + 1.7 Mb overhead)

the first 3 bytes of every row is always overhead

idea is that if you aave a sudden birst of noice, you don't destroy all of your header .. easier to recover from problems

oc-3 sts-3 155Mbps 6.8Mbps of overhead .. 148.6 mbps data

oc-12 sts-12 622Mbps

oc-24 sts-24 1.24Gbps

oc-48 sts-48 2.488Gbps

other type

oc-3c - concatinated service

.. can't break it out into separate oc-1s

one single cirucit that can't be divided .. just have to treat it as just one gigantic data circuit

.. typically run on a sonnet ring

digital subscriber lines (dsl)

xDSL

ADSL - asymetrical .. higher download bandwidth

uses the existing telco lines

the actual bandwidth achieved is determined by two things: distance to the telco office

line quality

how do we encode dsl data?

.. dmt .. discrete miltitone technique

uses a combo of fdm and 15qam

fdm to originally devide the bandwidth into channels

256 channels

0 is voice

1-5 buffer

6-30 upstream data

31-255 downstream data

typically , the upstream data is 64k to 1-Mb

downstream 500k to 8meg

in here we'll use qam to encode our data

how do you make sure your telephone isn't hearing your data, etc

your dsl model also has a filter to filter out the voice

how DSL works

Telco has a big pipe for my neighborhood

and all my neighbors have dsl

lots of wires in the long pipe, but there's one wire per house essentially

so the end result is that virtually everyone has own connection back to telco

.... it doesn't amtter what the neighbors are doing on their dsl modems

sdsl -- symetric dsl .. division is different .. for euql number of upstream and downstream

high bit rate dsl

and very high bit rate dsl .. 50-55meg down, 1-1.5meg up

-------

Thu 02/12/08

Cable modem

-----------

- coaxial cables

- FDM (frequency division multiplexing) to divide badnwidth

5Mhz to 870Mhz

first part - upstream data -

then analog cable tv channels to 550mhz

then

up to 750 mhz .. downstream data

750-850 .. digital cable tv

upstream data - because this is at low frequency , noise is more of an issue

upstream data - QPSK - 4 phase shifts .. 0, 90, 180 270

.. can encode 2 bits into each one of these

we also use FDM to further devide the bandwidth into 6 channels, 6Mhz each

you can only use 1 channel at a time

one of the things to note about the cable modems, you have your cable company

they have a wire that comes out to a hub/switch .. depending on how antiquated the system is

and then from the hub it goes to the houses .. so we're sharing one wire

we have 5 channels but everyone's fighting for it

.. in cable service, you're sharing with your neighbors

if you use cable during prime time, you'll suffer more

for the upstream data, you're sharing these 6 channels, you're essentially competing to use the 6 channels

typically, you negociate and be assigned a channel and then compete with everyohne else assigned to that channel

the maximum bandwidth is 2 bits * 6MHz = 12Mbps .. in the optimal world .. usuaaly a lot less than that

and the cable modem can throttle your service

and note that the 12Mbps is the shared bandwidth of the neighborhood

analog cable tv:

80 + 6Mhaz channels

uses FDM to break out into the separate channels

Downstream Data:

----------------

- FDM to divide this into 33 channels

.. less susceptible to noise, so we can use 64QAM or occasionally 256QAM

.. usues phase shifts and amplitude shifts

.. encodes 6 bits (2^6=64)

5 bits data and 1 bit ECC (error correcting code)

6Mhz * 5bits = 30Mbps

but this 30Mbps is shared by the neighborhood

the way in which it is shared is interesting

one device sending to everyone on the neighborhood .. so no media contentin

data is sent via broadcast.

essentially all the data is sent out to all the cable modems

cable modem filters out data for other destinations

sounds like a crazy way of doing things

physical layer - the electrical and mechanical specifications for transporting bits .. getting data from a to b

when we are choosing a physical media for our network, what are some concerns?

cost: install, per foot, terminate

bandwidth: how fast can it go?

interference: - some media are more succeptible than others

distance: how far can you go? - attenuation / loss of strength in the signal

security - fiber optics are significantly more secure than a wireless network

ease of installation: wireless is incredibly easy to install than fiber optics .. particularly if you have to terminate

latency - the delay of getting there

guided vs unguided: aka wired vs wireless

twisted pair wiring -

- most common type.. ex: eathernet (cat 5 and 6) and phone lines (cat 3)

different types:

shielded : has a protective shield around it

unshielded does not

shielded twisted pair wiring doesn't seem to exist

common problems with unshielded:

low resistance to electric current gives an advantage - the signals can travel longer distances without repeaters

but we're also suseptible to interference

problems:

emi -- electromagnetic interference

cross talk - signals fro parall wires jumping tt each other

primary sources of emi

-

- electrical runs and the ever-present air conditioner .. which used to be a huge problem in computer rooms

terror of the centry

unshielded twisted pair:

- cat 1 - primitive phone wires

- cat 2 - voice grade or up to 4 MBps

- cat(category) 3 - standard voice or 10Mbps

- cat 4 - up to 16mbps .. old ethernet cables

- cat 5 - up to 100Mbps (standard)

e .. a couple of tweaks specific to eathernet

four pairs of wires twisted at different rates

- cat 6 - data networks up to 10Gbps

- guided media.

on average, cheap (cat 6 more expensive)

bandwidth .. 4 mbps to 10Gbps

- interference issues - cross talk and emi

- higher levels of intereference than some of the others

distance - depends on speed / the further you go, the lower your bandwidth

can get up to 2km at a low bandwidth

.. at higher speeds 300ft is more common

security - better than unguided but not as good as fiber

.. relatively easy to wiretap

- very easy to install!

easy to manage

.. you can bend them and they'll still work

- latency is not an issue

cat 6 .. higher quality wires .. but very similar

-------

Fri 02/13/08

PHysical layer criteria

----------------

- cost

- bandwidth

- interference

- distance

- securiy

- ease of installation/maintenence

- latency

- guided/unguided

- durability

coaxial cable

- copper wire

- a single, thicker wire surrounded by protective sheath

looks like:

middle: wire

outside .. protective sheath

insulltion

further outside - braided metal shield

another layer of insullation . may or may not inclucdde another plastic sheath.

3 types:

thin coax

thick coax

sometimes used for data networks

cable tv coax .. kinda inbetween thin and thick

connectors used for coax are called BNC connectors .. really hard to get those on.

rj-45 is the connector for twisted pairs

cost of coax: still relatively low cost

typically more than unshielded twisted pair

bandwidth .. 1 mb to 1 gb

typ8ically you don't see speeds up to 1gb

generally 1 to 2 mb typically

it used to be the case that people used coax in networks where they needed highhspped .. but now unshilded twisted pair is faster

less susecptible to interference than unshielded twisted pair

interference is still an issue, though .. but less so than what we had with the unshield twisteddpair

distance - is less than what we could do for UTP .. 1km

security - about the same as unshield twisted pair

ease of install .. harder than utp but not too difficiult

cables are more bulky and less plyable

latency - not an issue

guided/unguided -- guided .. there is a physical wire

durable: more than the utp because guarded edges

typical uses for coax:

cable tv, cable internet (which really are the same thing)

_old_ data networks

optical fiber

- uses light waves instead of electricity

- travel accross miniature stands of: glass or plastic

a lot of different grades of optical fiber

0's and 1's are econded using OOK (on off keying)

use either LED or lazers to send the signals

on the other end, we have a light sensative transistor to detect light waves .. decode our signal

LED's are cheaper but less distance

therefore, lazer's are more expensive and typically used for long distance.

2 types of optical fiber

1. multimode -cheaper, less advanced version .. thicker strand of glass / plastic .. and the light waves essentiallly bounce their wave through.

as they bounce, at different freuqencies there will be different delays

so when the signals go in .. because of the delays, the signal looks more rounded (distorted) on other end.

because of this, cheaper to manufacture (than single mode) can only go shorter distance

2. snigle mode -

have a very thin high quality piece of glass that enables our light signal to essentially travel in a straight line so we dont' have reflextions taking place

so less distorted on the other end.

in order to get the single mode to work, the radious of the peice of class has to be exactly equal to the wave length

.. more expensive

can go longer distance

cost - very expensive .. the most expensive by far of all of our guided media .. expensive to install and manufacture

send and receive .. two wires!

fiber itself is expensive

and the modems on either end are expensive

bandwidth - 10Mb - 100Gb

.. probably been surpassed

interference: essentially immune to interference .. no emi because no electrical signals

distance: up to 50k

and can maintain relatively high data rates even at that distance

- but you need single mode fiber, with laser and glass

however, if you've got that, you can push shome pretty good data rates

security -- the most secure.

technically, with the copper, you could get really close to them and detect if there were signals

with optical fibers, unless you can get in to see that light, you have no way of dectecting what's going on in there

furthermore, you can tap into any of the eithernet wiring

.. but if you try that with optical, you'll destroy the signal

so most secure

ease of installation - if you just get the wires, have the modems and plug them in .. that's easy

but if you wanted to do a run, you need to be trained, the equipment is expensive and very precise ... have to polish the fiber

professional installation requred to terminate

this ishnt' something that you'll just go and install in your house

it is light weight, you have to be careful not to bend it too much, but they're pretty resilliant

plyable but don't bend them too much

latency - are you kidding??? we're going at the speed of light

guided or unguided .. obviously guided

durability - need to be careful not to break glass / plastic strand

also, a lot of optical is done in the ground .. construction crews cut the lines , then they're never the same.

whenever you have a break in optical fiber, the best solution is to replace it .. because the quality is degraded whe you try to repair

difficult to repair.

examples of where used: long runs / long distance high speed data runs

network backbones

between buildings

may have it between floors

cable and telco companies

.. internal services often optical fiber

unguided media:

--------------

sending antenna that radiates the signal in the air

and a receiving antenna that accepts it

inBetween, nothing but air

two main types:

directional - where you send the signal in a specific direction and you have to align the sender and reciever

signal is point to point

in contrast to the omnidirectional .. we send out the signal in all directions.

directional - more secure but not all that secure

-------

Mon 02/16/08

exam: Thur 2/26 .. 6pm

everything through physical layer .. dll will not be part of the exam

cof 108

for tomorrow's lab: form a company name

unguided media:

radio:

communicating via radio waves .. .. ranges of frequency from 3Khz to 1Ghz .. about ..

keep in mind, there really isn't a lot of difference between microwave and radio other than their freuqncies

microwave .. above 1ghz

but kinda a gray dividing line

going to use antennas to communicate through the air.

speeds .. 1 to 10 Mbps

higher speeds when we use microwaves

and this can typically penetrate walls

walls decrease the properties of the signal .. the more walls you have to go through the shorter distance you can do

and what tthe walls are made of is also important .. stix buildings .. pretty easy to penetrate the walls

typically when installing wireless, shcools are the hardest .. all the brick walls

criteria:

- - - - -

- cost

- bandwidth .. how fast can you go?

- distance

- interference

- security

- ease of installation / maintenence, ie ease of use

- latency / propagation delay

- durability

when we talk about radio waves, what can we say:

cost: a lot cheaper to add wireless to an existing building, but often cheaper to add wired media to a new building

- .. not necessarily cheaper for an all-brick wall building like cofrin

bandwidth - radio waves 1-10Mbps, microwaves .. higher

distance - really really really depends on interference

typically, you're not going to get more than a city block

esp if you want high bandwidth

examples of radio wavews:

- AM/FM radio ..

- UHF, VHF TV

- dtv

.. depends on how many trees, walls you need to go through

interference - huge issue

any electric equipment can interfere

and depends on the power of your signal

walls, trees, buildings, objects in general

.. it can openetrate the objects, but it degrades the quality of your signal and lessons your distance

.. so not impossible to get through these objects, just that they degrade your signal and don't let you have as high of data rates

security: the worst security of all!

easy to use and install.

the more difficult part with a wairelss install is figuring out how far your signal is going to go .. how many acess points are you going to need

latency, durability .. not an issue

one more thing for the interference - multipath interference

.. at least as big if not bigger than having other euqipment in the way

.. when you have an attena that is sending out a radio signal .. because it goes in multiple directions, when it hits the mountain , it bounces .. and they travel different distances ..and the reflection causes different travel distances and thus signals arrive at slightly different times

microwaves .. really just higher freqency radio waves .

bce some slightly different properties particularly at the higher end of the spectrum

really no clearcout difference at the lower frequencies

but when we get to the higher frequencies we get to see a ery distitnct different

the range we have here is 2ghz - 300ghz

ex: 2.4 GHz .. but it's a poor example, becaase it shares just as muuch with the radio waves as it dos with the microwaves

anyway, these can carry more data than radio waves

.. higher freqency = higher amounts of data

ex: wireless aka wifi .. different standards .. 802.11b was the original

.. a is business class

.. std top speed .. 11Mbps .. accross the entire network! not per device

802.11g was an update to 802.11b .. this can do upwards of 54Mbps

802.11n 540Mbps .. but not officially standardized yet .. there are defvices that are considered 802.11n devices but they say in their fine print it says it follows the proposed standard .. speeds vary significantly

we don't care the exact data rate, but it's about 10 times the speed of 802.11g

other uses of microwave

- - - - - - - - - - -

cordless phones

bluetooth .. 802.15.1 (high data rate)

802.15.4 .. low data rate personal area network

.. battery operated devices .. sleep most of the time and only wake up when they need to

^^^^^ 2.4.Ghz

other examples: point to point data .. non-2.4Ghz

distance is essentially unlimited

- cellular networks

pagihng networks

gps

cost: cost depends

2.4Ghz .. stuff is relatively inexpensive

.. higher frequencies are more expesnive

ease of use:

higher frequency point to point is difficult to install .. because the higher frequencies are typically point to point

why difficult to install - the antennas need to be carefully aligned

.. we typically have:

ex: the dish

or funny looking __ thing .. all the signals come in and get reflectef down to reciver

these antennas are aimed at eahh other and it goesout some .. butmore intended to be aimed at each other

distance: 10 to 100 kilometers without interference

interference: huge issue .. higher freqencies can't penetrate objects as well.

typically need line of site

higher frequencies = longer distances

so , again, on a continuum

rainfall causes inteference

we need 2 point to point signals for full duplex

otherwise, if we try sending back and forth on the same freuqency , you clobber your own data

security -- better than the radio waves, but probably worse than any of the guided media

latency / durability .. not going to wear out

latency only an isuse when we get to satellite

-------

Tue 02/17/08

using microwaves for satellite communication

Satillite

--------

uses micrwaves

.. use a satellite to relay information

.. advantages

- fewever obstructions

- can cover longer idstances

sattellits link two or more ground based receivers

.. can be either point to point .. ex company's data network

or multipoint

- broadcast

.. satellite connection -- uplink is basically a dialup connection

the downlink is satellite

common bands

cBand

if you waat full diplex, yo need two separate connections

cBand -- 17.7 to 21 Ghz uplink and the downlink 27.5Ghz

kuBand - 11.7 -12 Ghz

14- 14.5 Ghz

satellites need to be in orbit

simplest .. geosychroneous orbit

.. orbit at the same rate as the earth rotates .. so the same position relative to the earth at all times

.. this makes life a lot easier if we know that the satellites i the same place

needs to be 122, 236 miles above the equator .. and they need to be placed apart

.. other type - low earth orbit satellites

and these are not always in the smae position

.. can only be used when in view.

in order to compensate for this, we typicall use arrays of LEOS so that we always have at least one in view at alll times .. one or more

Iridium Satellites

- .. iridium is the 77th element and this was intended to hve 77 satellites

ultimately , becuae its extremely expensive to lauch , they only lauched 66.

oidea .. that we can see at least one at a time

originally used for pager system

the other example of the low earth satellites is the gps system

- has 24 satellites .. so they say

gps knows where the satellites should be .. sends a signal and waits for it to come back .. by that it can tell how far it is from the satellite .. three can draw cricles .. and by that ythey can pinpoint your location .. the more satellites you can see, the closer you can get

other uses for stellites:

- satellite tv

- weather can cause reception issue to satellite tv

- millitary uses -

- satellite phones .. popular with peope who go on around the world boating trips

- closed circuit broadcasts

- satellite internet -

- sateliite radio

criteria

- latency is an issue! becaase your signal needs to travel up to a satellite and back down, it will take some time

cost - very expensive

not easy to install

.. expesive to rent, maintain

distance - can cover the globe

bandwidth - .. remember, longer distance - lowever bandwidth .. 1-10 mb

security -- not very secure - anyone in the vicinity is going to be able to get the signal

ease of use - difficult to work with -- at the mercy of some other company --