COS 109 Monday December 7
• Housekeeping
– Lab 8 and Assignment 9 are posted
– Final exam – January 18 (Monday) at 7:30PM
In class, 3 hour exam
– A review session will be held on January 14 (time and location TBA)
– Reading assignment for next Monday (also posted on course home page)
https://static.newamerica.org/attachments/3421-riding-thewave/Riding%20the%20Wave_Final_and_Teaching_Note.d71b398fa011446ca
2739f5b0737dcfb.pdf
• Today’s class
– Wireless from radio to wifi to cellphones
– Data compression and error correction
Sound waves and frequencies
• Sound can be represented
by a graph showing
intensity plotted versus
time
• The graph is composed of
components that appear
regular (i.e. work at
certain frequencies)
• We can speak of the graph
as being composed of its
amplitudes (heights) at
various frequencies.
• We have Jean-Baptiste
Joseph Fourier (1768—
1830) to thank for this
Very simple view of what comes forward
• A radio process consists of a transmitter and a receptor
• Transmitter broadcasts signals at a certain frequency (or perhaps
many frequencies)
• Receptor is set to receive signals at a fixed frequency
• Signals decay over distance (based on their frequency)
• Underlying societal issue
– Who owns a frequency in a (geographic) area?
• Definition
– Frequencies are measured in cycles per second (called Hertz (Hz))
e.g. tune your radio to 770 KHz (an AM station) or 101.5 MHZ (an FM station) or
broadcast; Wifi may be at 5GHz; my cellphone uses 850MHz and 1900MHz
How does a radio work?
• Transmitter/receiver
• Tuned on a frequency
• Picks up that frequency and uses the amplitude to decode 1’s and
0’s from amplitude
• How a radio works
– How sound is made and transmitted (old time view)
– How sound is transmitted
• "modulate" the wave to impose information on it
– amplitude (AM): change the power level
– frequency (FM): change the frequency around nominal value
– digital: on/off
• received signal strength varies directly with power level
• received signal strength dies off with square of distance
• higher frequencies go shorter distances
On to WiFi
• A Wifi device involves two key components
– Antennae to receive signals as radio waves
– A port in the back to wire into the Internet (or LAN)
The wireless part
• Convert 0’s and 1’s into waves and transmit at frequency (2.4 or
5 GHz)
• Work according to standards
– Standards are defined by IEEE (Institute of Electrical and Electronic
Engineers)
• Transmit by various coding systems
– orthogonal frequency-division multiplexing (OFDM)
– complementary code keying (CCK) modulation
• The 802.11 standards
The 802.11 standards
•
•
•
•
•
802.11a (1997) transmits at 5 GHz and can move up to 54 megabits of data per
second. It uses OFDM to greatly reduce interference.
802.11b (1999) is the slowest and least expensive standard. For a while, its cost
made it popular, but now it's becoming less common as faster standards become
less expensive. 802.11b transmits in the 2.4 GHz frequency band of the radio
spectrum. It can handle up to 11 megabits of data per second, and it uses CCK
to improve speeds.
802.11g (2003) transmits at 2.4 GHz like 802.11b, but it's a lot faster -- it
can handle up to 54 megabits of data per second. 802.11g is faster because it
uses the same OFDM coding as 802.11a.
802.11n (2009) is the newest standard that is widely available. This standard
significantly improves speed and range. For instance, although 802.11g
theoretically moves 54 megabits of data per second, it only achieves real-world
speeds of about 24 megabits of data per second because of network congestion.
802.11n, however, reportedly can achieve speeds as high as 140 megabits per
second.
802.11ac came in 2013 to extend 802.11n. Channels are wider and will yield
higher bandwidth. Vendors are now building devices in this range.
Another view of wireless technology
A simple view of cell phones
• all phones are part of the public switched telephone network
• a cell phone is connected by radio instead of wires
• moves long distances, at high speed, appears out of nowhere
• shares a very limited radio frequency spectrum with others
• operates with low power because it uses batteries
• this makes life complicated
Public telephone
network
Switc
h
Switc
h
Base stations (== cellular
towers)
Cell
phones
Cells
(a very idealized picture)
• divide geographical area into cells (notionally hexagonal)
• each cell has an antenna, handles all cell phones in its area
• available radio spectrum is divided into channels
– two channels for one conversation, one for each direction
– competing carriers operate on
different frequencies
• each cell gets 1/7 of the channels
– adjacent cells can't use the same channels
because of interference
– non-adjacent cells can re-use channels
from www.howstuffworks.com
Not in my
back yard!
Camouflage
Camouflage
Camouflage
Meanwhile, in the southwest…
Can you hear me?
Sent on behalf of Jay Dominick, Vice President for Information
Technology and Chief Information Officer
Dear Colleague,
OIT is currently working with Verizon to supplement the
cellular wireless signal in and around the E-Quad. As part of
that process, Verizon has established a Cellular transmitter on
the roof of the J Wing of the E-Quad. The equipment being
used will transmit in licensed frequency bands (746Mhz 787Mhz, 835Mhz - 894Mhz and 1710Mhz - 2120Mhz) at
between 10 and 40 Watts.
How it works
• when a phone is turned on, it broadcasts its ID (“registration”)
– nearest base station notices, validates with home system
registration uses encryption for fraud prevention
– phone keeps broadcasting enough to keep in touch
• when the phone is called, the home system knows where it is
– home system contacts base(s) where phone is
– bases broadcast to where phone was last seen (“paging”)
• phones talk to base with strongest signal
– base and phone communicate over 2 agreed-upon channels (up, down)
– phones continuously adjust power level to signal strength at base
uses less battery, creates less interference for other phones
• phones move from base to base and from system to system
– base initiates handoff when signal gets weak
– phone picked up by base with strongest signal
– elaborate protocols at all levels
How it works, continued
• multiple frequency bands
(different in different parts of the world)
– divided into channels (frequency multiplexing)
digital phones multiplex several calls on one channel (GSM)
or spread calls out over the whole spectrum (CDMA)
– phones usually support multiple bands
• channels carry both voice and control information
(including data)
– digital speech is highly compressed (~1 bit/speech sample)
– elaborate coding & error correction for speech & control information
– power turned off when nothing is being sent
• phones store user info on removable flash memory card
– SIM (Subscriber Information Module)
– may be able to replace card to use in a different environment
• most of the world uses GSM
– in USA, AT&T & T-Mobile use GSM; Verizon & Sprint use CDMA
Radio waves – how are frequencies used
• Might be different channels to offset frequencies
• Device is set up to receive the frequency like a radio
–
–
–
–
WCBS – 660KHz
WPRB – 103.3MHz
My home landline – 900MHz
iPhone6 –
850,900,1800, 1900,2100 MHz for Cellular
802.11n (2.4GHz and 5 GHz for wifi)
– WNBC-TV
67.25 MHz (video signal)
71.75 MHz (audio signal)
• Frequency enables decoding
Sample frequencies
•
•
•
•
•
•
AM radio - 535 kilohertz to 1.7 megahertz
Short wave radio - bands from 5.9 megahertz to 26.1 megahertz
Citizens band (CB) radio - 26.96 megahertz to 27.41 megahertz
Television stations - 54 to 88 megahertz for channels 2 through 6
FM radio - 88 megahertz to 108 megahertz
Television stations - 174 to 220 megahertz for channels 7 through 13
•
Wireless devices in the house
– Garage door openers, alarm systems, etc. - Around 40 megahertz
– Baby monitors: 49 megahertz
– Radio controlled airplanes: Around 72 megahertz, which is different from...
– Radio controlled cars: Around 75 megahertz
– New cordless phones operate at DECT 6.0 (1.9 GHz)
– Bluetooth: between 2.402 and 2.480 gigahertz
– My slide advancing remote ~ 30 PHz (30 million GHz) (on infrared spectrum)
•
Other frequencies
– Wildlife tracking collars: 215 to 220 megahertz
– MIR space station: 145 megahertz and 437 megahertz
– Air traffic control radar: 960 to 1,215 megahertz
– Global Positioning System: 1,227 and 1,575 megahertz
– Deep space radio communications: 2290 megahertz to 2300 megahertz
– Ezpass – 915 megahertz
The full spectrum
From http://electronics.howstuffworks.com/radio-spectrum.htm
RF spectrum
Selling frequency
• see an auction
• Looking at a particular auction
• Seeing the results
Technology meets politics again
• should texting while driving be illegal (and enforced)?
– how about just talking on a phone while driving?
• where determines where cell phone towers are permitted?
– property rights versus eminent domain
• should cell phone jammers be legalized?
– in theatres, trains, etc.
• location tracking and surveillance
–
–
–
–
FCC mandates that cell phone can be locatable within 125 meter radius
should real-time location info be available to law enforcement, etc.?
how should this evolve as GPS becomes universally available?
who can have access to what cell phone records under what
circumstances?
• if you use your cell phone to make a call or use wifi on your
phone to Skype, what is the difference?
GPS (Global Positioning System)
• 31 satellites, each broadcasting time & its location
– altitude ~ 20 km, frequency ~ 1575 MHz
– at least 6 are visible at any time
• receiver calculates its position using distances to 3 or more
satellites
– distances computed by careful measurement of time
– accuracy typically within 15 m for civilian systems
– additional inputs or use of encrypted info reduces this to < 1 m
Web site of the day
• The useless web
Compression; Error detection & correction
• compression: squeeze out redundancy
– to use less memory or use less network bandwidth
– encode the same information in fewer bits
• some bits carry no information
• some bits can be computed or inferred from others
• some bits don't matter to the recipient and can be dropped entirely
• error detection & correction: add redundancy
– to detect and fix up loss or damage
– add carefully defined, systematic redundancy
– with enough of the right redundancy,
can detect damaged bits
can correct errors
Compressing English text
• letters do not occur equally often
• encode frequent letters with fewer bits, less frequent things
with more bits (trades complexity against space)
– e.g., Morse code, Huffman code, ...
• run-length encoding
– encode runs of identical things with a count
– e.g., World Wide Web Consortium => WWWC => W3C
• words do not occur equally often
• encode whole words, not just letters
– e.g., abbreviations for frequent words
Letter frequencies in King James bible (4.1M chars)
800000
SP
700000
600000
500000
E
400000
T
H
300000
A
O
N
I
200000
S
R
D
L
F
100000
U
M
W
Y
G
C
B
P
V
K
J
Z
X
Q
J
Z
X
Q
0
SP
E
T
H
A
O
N
I
S
R
D
L
F
U
M
W
Y
G
C
B
P
V
K
Word frequencies (790K words)
70000
the
60000
and
50000
40000
of
30000
20000
to that
in
he shall
untofor i his a
lordthey be is
him notthem it w ith
all thou
thyw asgod
w hichmy me saidbut ye theirhavew illtheefrom
as
10000
as
th
ee
ha
ve
ye
sa
id
m
y
go
d
th
y
al
l
it
no
t
is
th
ey
a
i
un
to
he
th
at
of
th
e
0
are
Zipf's Law
(adapted from Wikipedia)
• frequency of any word is inversely proportional to its position
in the frequency table
70000
60000
– empirically, the most frequent word will occur approximately twice
as often as the second most frequent word, which occurs twice as
the
often as the fourth most frequent word, etc.
and
50000
40000
of
30000
20000
to that
in
he shall
untofor i his a
lordthey
be is him notthem it
w ith all thou
thyw asgodw hichmy me saidbut ye theirhavew illthee
from as
10000
as
th
ee
ha
ve
ye
sa
id
m
y
go
d
th
y
al
l
it
no
t
is
th
ey
a
i
un
to
he
th
at
of
th
e
0
are
Lempel-Ziv coding; adaptive compression algorithms
• build a dictionary of recently occurring data
• replace subsequent occurrences by (shorter) reference to the
dictionary entry
• dictionary adapts as more input is seen
– compression adapts to properties of particular input
– algorithm is independent of nature of input
• dictionary is included in the compressed data
• Lempel-Ziv is the basis of PKZip, Winzip, gzip, GIF
– compresses Bible from 4.1 MB to 1.2 MB (typical for text)
• Lempel-Ziv is a lossless compression scheme
– compression followed by decompression reproduces the input exactly
• lossy compression: may do better if can discard some information
– commonly used for pictures, sounds, movies
JPEG
(Joint Photographic Experts Group)
picture compression
• a lossy compression scheme, based on how our eyes work
• digitize picture into pixels
• discard some color information (use fewer distinct colors)
– eye is less sensitive to color variation than brightness
• discard some fine detail
– decompressed image is not quite as sharp as original
• discard some fine gradations of color and brightness
• use Huffman code, run-length encoding, etc., to compress
resulting stream of numeric values
• compression is usually 10:1 to 20:1 for pictures
• used in web pages, digital cameras, ...
JPEG images
• (16"x14") 1800 bytes:
(87"x118") 360K Bytes:
MPEG
(Moving Picture Experts Group)
movie compression
• MPEG-2: lossy compression scheme, based on human perceptions
• uses JPEG for individual frames (spatial redundancy)
• adds compression of temporal redundancy
–
–
–
–
–
–
look at image in blocks
if a block hasn't changed, just transmit that fact, not the content
if a block has moved, transmit amount of motion
motion prediction (encode expected differences plus correction)
separate moving parts from static background
...
• used in DVD, high-definition TV, digital camcorders, video games
• rate is 3-15 Mbps depending on size, frame rate
– 15 Mbps ~ 2 MB/sec or 120 MB/min ~ 100x worse than MP3
– 3 Mbps ~ 25 MB/min; cf DVD 25 MB/min ~ 3000 MB for 2 hours
– regular TV is ~ 15 Mbps, HDTV ~ 60-80 Mbps
MP3
(MPEG Audio Layer-3)
sound compression
• movies have sound as well as motion; this is the audio part
• 3 levels, with increasing compression, increasing complexity
• based on "perceptual noise shaping":
use characteristics of the human ear to compress better:
•
•
•
•
– human ear can't hear some sounds (e.g., very high frequencies)
– human ear hears some sounds better than others
– louder sounds mask softer sounds
break sound into different frequency bands
encode each band separately
encode 2 stereo channels as 1 plus difference
gives about 10:1 compression over CD-quality audio
– 1 MB/minute instead of 10 MB/minute
– can trade quality against compression