CS3502:CS3502:
Data and Computer NetworksData and Computer Networks
Local Area Networks - 1Local Area Networks - 1
introduction and early broadcast introduction and early broadcast protocols protocols
CS3502 , LANs. Objectives CS3502 , LANs. Objectives
1. describe LAN topologies/transmission media
2. describe MAC protocols -> in detail
3. compare/contrast different LANs
4. verify basic LAN protocols
5. describe and compare LAN throughputs
6. describe and analyze bridges/LAN switches
7. describe basic router function, differentiate from bridge.
local area networks : general local area networks : general infoinfo
limited geographical area
relatively high transmission rates
simple topologies and routing
mostly baseband -- single channel
usually owned by 1 organization
characterized by topology, medium, and MAC protocol
LANs : classes, topologiesLANs : classes, topologies broadcast (contention); bus or wireless
Aloha, CSMA, CSMA/CD (802.3) *, wireless LANs (802.11)
broadcast (controlled) bit map protocol, token bus
ring 802.5 token ring *, FDDI (token), slotted rings
star ATM LAN
local area networks : broadcastlocal area networks : broadcast all nodes connected by ONE channel if more than 1 node transmits simultaneously,
signals interfere (collision): the message is lost
thus, the transmission medium is always in 1 out of 3 possible states:
(1)
(2)
(3) example: classroom? .... 2 channels
LANs : ALOHA (pure) LANs : ALOHA (pure)
radio frequencies OR any broadcast medium
U of Hawaii, early 1970s. Prof.. N. Abramson, funded by ARPA.
simplest possible protocol
a station with a message simply transmits it to completion. If no collision, message gets through, otherwise wait random time and retransmit.
LANs : ALOHA (pure)LANs : ALOHA (pure) works for when transmissions are rare; but
quickly degenerates as load increases
performance analysis, based on assumed Poisson distribution, shows max utilization of 18%. (following slides) .
User Load- the amount of traffic attempting to get through the channel
Throughput - amount of traffic getting through the channel
Utilization - is the fraction of time that the channel is transmitting data (throughput/max throughput)
LANs : Aloha performance LANs : Aloha performance analysis analysis
based on several assumptions: 1. Transmission attempts are generated by an infinite number of users.2. transmission attempts follows a Poisson distribution.3. fixed packet size
Def: Let X be a random variable, representing a nonnegative integer. X is a poisson random variable if
p(i) = P[X=i] = (e - i )/i!
LANs : Aloha performance LANs : Aloha performance analysis analysis
note: Poisson distribution (discrete RVs) and exponential distribution (continuous RVs) are closely related.
the mean, or “average” of the poisson dist. is E [X]lso note --
P[X=0] = p(0) = e - , and P[X=1] = p(1) = e -
( come from plugging 0, 1 into the formula)
LANs : Aloha performance LANs : Aloha performance analysisanalysis
Let S = number of successful packet transmissions per packet time (equals channel utilization)
G = average number of attempted transmissions per packet time(user load+retransmissions).Then,S = P[successful transmission]
= P[1 attempt] P[no other attempts]= P[X=1] P[X=0]= G x e -2G, pure Aloha
Q : what is the maximum utilization? (take the derivative, set to 0, plug back in)
LANs : performance analysis LANs : performance analysis
derivative : ( G x e -2G)’ = (1)(e -2G ) + G(e -2G )(-2)
setting to 0, e -2G - 2G e -2G = 0 => 1 - 2G = 0 => G = 0.5
I.e., utilization is max at G = 0.5. Plugging this into the original formula,
S = G x e -2G
yields a max value of 0.18.
LANs : ALOHA (slotted) LANs : ALOHA (slotted)
how can ALOHA be improved?
need to reduce collisions
slotted ALOHA : restrict transmissions to time slots divide time into “slots” station waits until next time slot to transmit slots must be synchronized, somehow
how much will utilization improve?
LANs: ALOHALANs: ALOHA
when should station retransmit after a collision?
show why throughput should double with slotted Aloha over pure Aloha
what is the worst-case time a station will have to wait until getting a successful transmission?
how can Aloha be improved? hint: what if we could use 2 power levels?
LANs : ALOHA, 2 power levels LANs : ALOHA, 2 power levels
idea: when station transmits, flip a coin. Heads, use low power level. Tails, used high power level.
high power clobbers lower power; if same power, collision as before.
can be added to either pure or slotted. Improves max throughput to 26% (pure) or 52% (slotted) under same Poisson assumptions.
LANs : ALOHA summaryLANs : ALOHA summary
simple communications (simple is good)
relatively cheap, simple to implement
good for sparse, intermittent communication.
not a good LAN protocol because of
poor utilization
potentially infinite delay
stations have listening capability, but don’t fully utilize it
LANs: CSMALANs: CSMA
corrects the obvious flaw in Aloha (blindly transmitting without first checking the medium)
CSMA(carrier sense multiple access) protocol:
(1)sense the carrier; {LISTEN}
if no signal detected then transmit message to end; {TALK} if collision occurred, then wait random time, go to (1) else END.
else {carrier is busy} go to (1).
LANs: CSMALANs: CSMA
basic CSMA is “persistent,” or “1-persistent” -- it transmits as soon as it detects the open carrier.
suppose another station is transmitting; when will the station start to transmit?
what effect does propagation delay have on this protocol?
note that whenever transmission occurs, the whole message is sent: no way to abort
LANs: CSMALANs: CSMA
what are 2 ways that collisions can occur in CSMA? What is their likelihood?
Will CSMA improve throughput over Aloha?
nonpersistent CSMA: when carrier is busy, wait a random time.
Will this improve utilization?
LANs : CSMA, p-persistent LANs : CSMA, p-persistent
variation of CSMA; generalizationfor parameter p : real, in (0,1], ---
(1)sense the carrier;
if no signal detected then transmit message to end with
probability p ; else {probability 1- p} wait random time, goto (1);
if collision occurred, then wait random time, go to (1) else
END; else {carrier busy} go to (1).
LANs : CSMALANs : CSMA
will this improve utilization?
for low values of p, maximum utilization is highest; what about user friendliness?
Since even if the channel is free one defers transmission
with p=> 0 the average wait time =>
LANs : Aloha/CSMA , summaryLANs : Aloha/CSMA , summary
Compare approximate utilization of the protocols pure Aloha slotted Aloha 1-persistent CSMA p-persistent CSMA nonpersistent CSMA
Compare the throughput as a function of user Load