Transcript slides

Review
Layers
• Physical layer – sending bits from one place to
another, ensuring an okay BER
• Data link layer – encapsulate information bits into
frames, and send frames from one node to
another neighboring node
• Network layer – send packets from one place to
another, may travel multiple hops
• Transportation layer – (TCP) sending segments
from one end to the other end, making sure that
there is no error, etc.
Physical Layer
• The goal – sending bits from one place to
another
• Two limiting factors, bandwidth and power.
• Shannon’s theorem.
Problem
Physical layer
• If bandwidth is abundant, no special things
needs to be done. Like 10Mbps Ethernet. Just
pulling voltage up and down.
• In other cases when bandwidth is not so
much, like wireless and telephone lines, have
to do something.
Physical Layer
• Sine waves. Send a sine wave, the other end will
receive a sine wave also. It will be on the same
frequency, but the amplitude will be different
(attenuation) and phase will be different (propagation
delay).
• Typically you are allowed to use only a frequency band.
Wireless LAN – 802.11g – 2.4G-2.48G. Divided into
channels, each channel is 20M. Cellphone – maybe in
the 900M band, or the 1.8G band, etc. GSM divides the
band (50M) into channels, each channel being 200k.
Each channel shared by 8 users in a time division
fashion.
Physical Layer
• You have to send signals *modulated* on a
sine wave on the frequency you are allowed to
use.
• Modulation techniques. Frequency.
Amplitude. Phase.
• OFDM (in 802.11g and 802.11a).
The basic transmission/reception
diagram
• Information bits baseband waveform
modulated waveform.
• Received waveform baseband waveform
information bits
I, Q channels
• It is possible to pack two sine waves on the
same frequency and send them out, one is
sine and the other is cosine, each carrying one
baseband waveform.
• I channel and Q channel.
Problem
Error Correction and Detection
• By introducing redundancy, the receiver can do
some calculation to correct or detect errors.
• We talked about Hamming Code in details.
• The key is, every error will result in a unique
syndrome. You find the syndrome you find the
error. If there is a syndrome, there is an error.
• Cyclic codes. Any cyclic shift of a codeword is still
a codeword. Code generated by doing polynomial
calculations.
Problem
Data Link Layer
• Sending frames from one end of a link to the
other end of the link.
• The problem is: the receiver may be slow, the
link may lose frames, both the data frames
and the control frames you want to use.
Data Link Layer
• Simplest correct protocol. Stop and Wait.
• Transmission delay and propagation delay.
• If propagation delay is small, like in a distance
of 1km, Stop and Wait is perfectly fine.
• Otherwise, not efficient.
Data Link Layer
• Go-back-N.
• The sender is allowed a window. The frames in
this window can be sent without ACK.
• Every time a timeout happens, retransmit
every thing in the current window.
• Selective Repeat. Only retransmit the frame
that was time out.
MAC
• In Ethernet and wireless LAN, nodes share a
medium.
• How to design a protocol to allow nodes share
the medium without causing collisions.
• Collision means two frames were sent at the
same time.
MAC
• You can do polling, but the traffic of
computers are random.
• Contention-based. Nodes know better when
they want to send.
• ALOHA – send when want to, no carrier sense.
Ethernet
• Ethernet – CSMA/CD. 1-persistent.
Exponential backoff.
• Sense the medium before sending, if busy,
don’t send. When medium is free, send. If
noticed a jam signal, stop, backoff. The time to
backoff is picked randomly from [0,CW-1]. If
collided again, double CW. Until CW is 1024.
give up when tried 16 times.
802.11
• CSMA/CA. non-persistent. Exponential
backoff.
• Sense the medium before sending, if busy,
don’t send. When medium is free, backoff a
time picked randomly from [0,CW-1]. If still
free, send. If no ACK received, double CW,
repeat the process. Give up when tried 1
times.
Problem
Problem
Network Layer
• Network Layer – moving packets from one
place to the other
• First problem – routing.
• Shortest path algorithm.
• Link State Algorithm. Distance Vector
Algorithm.
IP
• IP address. A,B, C, D class. CIDR. The allocation
of IP addresses.
• Internet Protocols. BGP, OSPF, DHCP, ARP.
• NAT.
Problem
Problem
Problem
Routers
• Lookup to find next hop link.
• Input-buffered switches and scheduling
algorithm.
Transportation Layer
• TCP – send segments from one end to the
other end reliably.
• TCP connection set up. Three-way handshake.
Initial sequence number.
• TCP connection close.
TCP
• TCP congestion control. AIMD. Why is it used.
• TCP optimizations. Fast retransmit and fast
recovery.
Network Security
• Three tools. Hash function. Block cipher. RSA.
• Hash functions. Two properties.
• Block Cipher. Pseudo random permutation.
AES. Cipher modes.
• RSA.
Network Security
• Applying the tools to realize functions.
• Digital Signature.
• Certificate.
Problem
Problem