1

CONTENTS

Digital Subscriber Line Technology

Cable Modems
Connecting devices
Repeaters
Hubs
Bridges
Switches
Routers

DIGITAL SUBSCRIBER LINE

Dial-up Connection for Internet access

Analog modems have limited data rate, 56 Kbps.
Digital Subscriber Line Technology provides high speed access to the
Internet over existing local loops.
DSL Technologies (xDSL): ADSL, VDSL, HDSL, and SDSL
ADSL (Asymmetric DSL)
Downstream bit rate: 500 Kbps to 8Mbps, Upstream bit rate: 64 Kbps to
1Mbps
Designed for residential users and is not suitable for businesses.
Uses existing local loops

DSLAM = Digital Subscriber Line Access Multiplexer

Local loop: twisted pair, theoretical 1.1MHz

Filters limits the bandwidth to 4 KHz.
Data rate is not fixed
Affected by distance, size of cable, signaling used, etc.

Bandwidth versus distance over category 3 UTP for DSL

Modulation technique
Discrete Multitone Technique (DMT)
QAM + FDM
The bandwidth of 1.104 MHz is divided into 256 channels.
Each channel uses a bandwidth of 4.312 KHz.

Channel 0 : Voice
Channel 1-5 : not used
Channel 6-30 (25 channels): upstream data and control; one channel is
used for control; 24 channels, each using 4 KHz with QAM (15 bits/baud )
provides = 24 x 4000 x 15 = 1.44 Mbps bandwidth.
Channel 31-255 (225 channels): downstream data and control; one
channel is used for control; 224 channels, each using 4 KHz with QAM (15
bits/baud ) provides = 24 x 4000 x 15 = 13.4 Mbps bandwidth.

SDSL (Symmetric DSL)

equal bandwidth for up and down streams

HDSL (High-bit-rate DSL)

An alternative to T-1 line
Use 2B1Q encoding, 2 Mbps dual-way up to 3.6 km
2 twisted-pair wires to achieve full duplex

VDSL (Very-high-bit-rate)
Use coax, fiber-optic, or twisted-pair for short distance
DMT with up to 50-50 Mbps downstream and 1.5-2.5 Mbps
upstream

CABLE MODEM

Traditional Cable Networks

unidirectional

Hybrid fiber-coaxial network

bidirectional

Coaxial cable, 5 to 750 MHz.

A TV channel occupies 6MHz

54 to 550 MHz: video band, 80 channels

550 to 750 MHz: downstream data band; divided in to 6 MHz channels;
QAM-64 or QAM-256; QAM-64 (6bits /baud), 2 bit for FEC and 5 bits for
data, 1 Hz for each baud, 5bits/Hz x 6 MHz = 30 Mbps (theoretical),
standard specifies only 27 Mbps.
5 to 42 MHz: upstream data band; divided in to 6 MHz channels, lower
frequencies are susceptible to noise and interference, QAM not suitable,
uses QPSK, 2 bits each baud, 2bits/Hz x 6 MHz=12 Mbps

Both upstream and downstream bands are shared by different users.

Cable Modem

Cable Modem Transmission System (CMTS)

A data transmission scheme: Data Over Cable System Interface

Specification (DOCSIS)
Upstream protocol
CMTS advertises itself with allocated up and down channels
CM starts ranging, estimating distance to CMTS
CM asks CMTS for an IP address
CM and CMTS exchanges messages for security matters
CM sends a unique ID to CMTS
Downstream protocol
No contention, CMTS just sends the packet with address of
destinations CM

NETWORK DEVICES:
REPEATER

Allows the connection of network segments

Extends the network beyond the maximum length of a single segment
Functions at the Physical Layer of the OSI model
Connects segments of the same network, even if they use different media

Has three basic functions

Receives a signal which it cleans up
Re-times the signal to avoid collisions
Transmits the signal on to the next segment
No traffic filtering

HUB

A multi-port repeater is known as a Hub

Functions in a similar manner to a Repeater
A central point of a star topology
Allows the multiple connection of devices
Can be more than a basic Hub providing additional services (Managed
Hubs, Switched Hubs, Intelligent Hubs)
Works at the Physical Layer of the OSI model
Passes data no matter which device it is addressed to
This feature adds to congestion
No filtering of traffic, collision, not scalability

BRIDGE

Like a Repeater or Hub it connects segments

Works at Data Layer not Physical
Uses MAC (Medium Access Control) address to make decisions
Acts as a filter, by determining whether or not to forward a packet on to
another segment
Builds a Bridging Table, keeps track of devices on each segment
Filters packets, does not forward them, by examining their MAC address
It forwards packets whose destination address is on a different segment
from its own
It divides a network in to multiple collision domains so reducing the
number of collisions

SWITCH

A multiport Bridge, functioning at the Data Link Layer

Each port of the bridge decides whether to forward data packets to the
attached network
Keeps track of the MAC addresses of all attached devices (just like a
bridge)
Acts like a Hub, but filters like a Bridge
Each port on a Switch is a collision domain

ROUTER

Works at Network Layer in an intelligent manner

Can connect different network segments, if they are in the same building
or even on the opposite side of the globe
Work in LAN, MAN and WAN environments
Allows access to resources by selecting the best path
Can interconnect different networks Ethernet with Token Ring
Changes packet size and format to match the requirements of the
destination network

LAN DEVICES

10

11

CONTENTS

Networks Switched Vs Broadcast

Medium Access Control (MAC) ?
Peer-to-Peer Vs MAC
MAC Classification
Random Access Schemes
ALOHA, CSMA, CSMA/CD
Scheduling
reservation system, polling, token passing
Channelization FDMA, TDMA, CDMA

12

NETWORKS CLASSIFICATION
Networks are classified into two types: switched networks and broadcast
networks.
Switched Networks
consists of transmission lines, multiplexers, and switches.
information is transmitted from source to destination using routing tables.
uses hierarchical addressing scheme.
Ex: Telephone network, Internet
Broadcast Networks
stations shares a single medium
Information transmitted from source is received by all users.
no routing used
uses flat addressing scheme.
Ex: Local Area Networks Ethernet, Token Ring
collision of user information is problem in broadcast networks
therefore requires mechanisms that regulate users accessing shared medium
it is a data link layer Issue
Reference Model

PEER-TO-PEER Vs MAC

Both protocols are used to transmit user information despite transmission

impairments
Peer-to-Peer protocols
involves only two peer processes
main concern is: loss of packets, delay, resequencing of packets
control frames used to coordinate their actions
delay-bandwidth is an important parameter
MAC protocols
Uses some mechanisms to coordinate the access of channel

13

 need the coordination from all MAC entities, any one does not cooperate, the
communication will not take place.
Main concern is : interference from users
Delay-bandwidth is an important parameter

MULTIPLE ACCESS COMMUNICATION

Transmission from any station can be heard by any other stations as shown in
the fig above.
If two or more stations transmit at the same time, then collision occurs.
The different medium access techniques are shown in the fig below.

Static Channelization
transmission medium is partitioned into channels.
channels are dedicated to individual users, so that no collision takes place.
good for steady traffic and achieve efficient usage of channels

Dynamic medium access control

Minimize the incidence of collision to achieve reasonable usage of medium.
Good for bursty traffic.

14

 Scheduling
involves scheduling an orderly access to the transmission medium.
Good for heavier traffic.

Random Access
Try, if no collision, good, otherwise wait a random time, try again.
Good for light traffic.

Example networks Satellite network

Cellular network

Bus topology based network (Ethernet, IEEE 802.3)

15

 Ring topology based network (IEEE 802.5)

Wireless network (IEEE 802.11)

MEDIUM ACCESS CONTROL AND PERFORMANCE

Shared medium is the only means for stations to communicate;
Some kind of MAC technique is needed.
Like ARQs, which use ACK frame to coordinate the transmission and consume
certain bandwidth, the MAC will need to transfer some coordination information
which will consume certain bandwidth of shared medium.
Delay-bandwidth product plays an important role in the performance of MAC (as
in ARQs).
A Simple MAC protocol
If a station (node) has a frame (data) to send, it first listens to the channel to
see whether it is idle or busy. If it is idle, it transmits its frame. After it transmits, it
continues to observe the signal in the channel to see if incase the signal has
corrupted because of transmission from other station.

16

 At time t = 2tprop, both station would know that their frames have got collided
and are competing for the channel.
So, we need a mechanism to resolve the contention.
Let both stations know the value of tprop.
Stations measure the time from when they begin frame transmission to when
collision occur.
The station having the above calculated value greater than t prop /2 would win
and proceeds retransmitting the frame.
The station that loses remains quiet until the frame transmission from other
station gets completed.
For fairness, the winning station would wait 2tprop time after its frame
transmission, to allow the other station to capture the channel.

MAC Efficiency
Suppose bit rate of medium is R, then number of bits wasted in access
coordination is 2tprop R.
If the average packet length is L, then the efficiency in use of medium is

a = tprop R / L i.e., the ratio of (one-way) delay-bandwidth product to the

average packet length.
Suppose
a = 0.01, then efficiency = 1/1.02 = 0.98
a = 0.1, then efficiency = 1/2 = 0.50
as a increases, the channel becomes more inefficient.
Examples
Ethernet (CSMA-CD):
Efficiency = 1/(1+6.44a) where a = tpropR/L.
Token-ring networks:

17

Efficiency = 1/(1+a ) where a = ring-latency in bits/L where ring-latency

contains:
The sum of bit delays introduced at each ring adapter.
Delay-bandwidth product where delay is the time required for a bit to circulate
around the ring.

MEDIUM ACCESS CONTROL PERFORMANCE METRICS

Delay
The time that elapses from when the first bit of the frame arrives at the source
MAC to when the last bit of the frame is delivered to the destination MAC.
Fairness
Does not give preference to any single node when multiple nodes are trying to
access the channel fair sharing of bandwidth
How about priorities in case of multimedia traffic?
Fairness in this case: ability to distribute bandwidth in proportion to their
allocation
Reliability
How robust is the MAC with respect to failures and faults?
Support for Multimedia
MAC protocols should support multimedia traffic : voice and video
Should treat packets from different applications based on QOS parameters
such as delay, bandwidth etc.
Scalability
How is the MAC scalable with respect to number of stations and user bandwidth
requirements?

MAC PERFORMANCE Delay Vs Throughput

Let M stations in a system generate frames at an aggregate rate of l
frames/second.
If length of the frame is L bits, then the average bit rate generated by all stations
is lL bits/second.
The normalized throughput or load is defined by r = lL / R.
What is important is lL < R i.e. r = lL / R < 1.
So, every MAC protocol has max. throughput less than R / L frames/second.
This is because some channel time is spent in collisions or in sending
coordination information.
So, r cannot exceed max. normalized throughput value rmax < 1.

18

Observations:

The average transfer delay increases with throughput. r is the load generated by
stations.
The average transfer delay Vs with load with a (one-way delay-bandwidth
product to the average frame length) is shown in fig above.
The average transfer delay Vs with load for different values of a is shown in fig
below.

19

RANDOM ACCESS SCHEMES

Random Access
station having a frame (data) would transmit at a full channel rate
R.
no priori coordination among stations.
if more than one station transmit, the collision takes place.

Random access schemes describes

How to detect collisions?
How to recover from collisions? (generally this is through delayed
retransmissions)

Collision

Different Random access schemes

Pure ALOHA
Slotted ALOHA
CSMA Schemes
CSMA/CD IEEE 802.3
CSMA/CA IEEE 802.11

PURE ALOHA

Designed by a research group at university of Hawaii to interconnect

different packet radio terminals.
Simple protocol and designed for light traffic.
Basic Idea:
let stations transmit whenever they have data to be sent.
When collision occurs, wait a random time and retransmit again.
Why random time?
To spread out the retransmissions from different stations and thus
reduce the likelihood of additional collisions.

20

Collisions in ALOHA is treated as transmission errors.

Regular errors Vs collisions
Regular errors only affect a single station
Collision affects more than one station
The retransmission may also collide again
Even the first bit of a frame overlaps with the last bit of a frame
almost finished, then two frames are totally destroyed.

Performance Analysis of ALOHA:

Let L: the average frame length, R: rate, frame transmission time X = L / R

Transmit a frame at t=t0 (and finish transmission of the frame at t0+X )
If ACK does not come after t0+X+2tprop or hear collision, wait for random time:
B
Retransmit the frame at t0+X+2tprop+B
Two modes: collide only from time to time and snowball effect collision.

Vulnerable period: t0-X to t0+X, (2X seconds), if any other frames are
transmitted during the period, the collision will occur.

21

Therefore, the probability of a successful transmission is the probability

that there are no additional transmissions in the vulnerable period.

Let S be the arrival rate of new frames in units of frames/X seconds,

S is also the throughput of the system.
Let G be the total arrival rate in units of frames/X seconds, G contains the
new and retransmissions and is the total load.
Assume that aggregate arrival process resulting from new and
retransmitted frames has a Poisson distribution with an average number of
arrivals of 2G frames/2X seconds, i.e.,

(2G)k
P[k transmissions in 2X seconds] =
k!
Therefore, the throughput of the system

is:

e-2G

, k=0,1,2,
k=0,1,2,

S=GP[no collision] =GP[0 transmission in 2X seconds]

(2G)0

=G

e-2G =G e-2G

0!

from the fig above, the peak value can be seen at G=0.5 with S=0.184
for any given S, there are two values of G, corresponding to the two
modes: occasional collision mode with S G and frequent collision mode
with G >> S

22

SLOTTED ALOHA

Synchronize the transmissions of stations

All stations keep track of transmission time slots and are allowed to
initiate transmissions only at the beginning of a time slot.
Suppose a packet occupies one time slot
Vulnerable period is from t0-X to t0, i.e., X seconds long.

the throughput of the Slotted ALOHA system is:

From the fig. above, the peak value is seen at G=1 with S=0.368 for
slotted ALOHA, double when compared with ALOHA.
In LAN, propagation delay may be negligible and uncoordinated access
of shared medium is possible but at the expense of significant wastage
due to collisions and at very low throughput.

23

Throughput of ALOHAs is not sensitive to the reaction time because

stations act independently.

CARRIER SENSE MULTPLE ACCESS (CSMA) SCHEMES

Problem with ALOHAs: low throughput because the collision wastes

transmission bandwidth.
Solution: avoid transmission that are certain to cause collision, that is
CSMA. Any station listens to the medium, if there is some transmission
going on the medium, it will postpone its transmission.

Suppose tprop is propagation delay from one extreme end to the other
extreme end of the medium.
When transmission is going on, a station can listen to the medium and
detect it.
After tprop, As transmission will arrive the other end; every station will
hear it and refrain from the transmission, so A captures the medium and
can finish its transmission.

CSMA Schemes:

p-persistent CSMA
1-persistent CSMA
Non-persistent CSMA

P-PERSISTENT CSMA SCHEME

Method:
Listen to the medium for any activity. If there is no activity, transmit;
otherwise, continue to monitor the medium.
When the medium becomes quiet, transmit with a probability p.
Otherwise, wait for the next time slot (probability 1-p) and repeat the
above steps.

24

Special Cases:
p is the probability that a station will transmit after the medium becomes
quiet.
If p = 1, we call it 1-persistent CSMA, which means it always transmits
when the medium is quiet.
If p = 0, we call it 0-persistent CSMA, which means it always waits for one
time slot.

NON-PERSISTENT CSMA SCHEME

Method:
A station checks for activity when it wants to transmit.
If the medium is idle, it transmits. Otherwise, it waits for the next time slot
and repeats the above steps.
The station does not continue monitoring the medium when it is waiting.

25

from the fig. above it is clear that for the different values of G, nonpersistent CSMA scheme performs better than other schemes.
When G approaches to 1, the 1-persistent CSMA performs as poor as
ALOHA schemes.

CARRIER SENSE MULTIPLE ACCESS WITH COLLISION DETECTION

(CSMA/CD)

The vulnerable period in CSMA schemes compared to ALOHA is reduced

from 1 or 2 frame transmission times to a single propagation delay tprop.
When the transmitting station detects a collision, it stops its transmission
immediately, does not transmit the entire frame which is already in
collision.
The time required for a transmitting station to detect a collision is 2tprop.

Method: when a station wants to transmit a frame, it senses channel, If

idle, it transmits, If it is busy, uses one of the CSMA algorithms (i.e., 1persistent, non-persistent, and p-persistent schemes). The station
continues to sense the channel during transmission. If a collision occurs
and was sensed, station stops its left transmission of the current frame
and a short jamming signal is transmitted to ensure other stations know
that a collision has occurred and backoff algorithm is used to schedule a
future re-sensing time.

From fig. below, Station A requires 2tprop seconds (reaction time) to

confirm that it has successfully captured the channel.

26

The channel is CSMA/CD can be in 3 states: busy in transmitting a frame,

idle, or in contention period where stations attempt to capture the channel

Throughout performance of 1-persistant CSMA/CD

Assume that the time is divided into minislots of length 2tprop seconds so
that stations can always detect a collision.
The contention interval is 2tprop seconds.
We are interested to calculate the mean time required for a station to
successfully capture the channel.
Suppose n stations are contending for the channel and each station
transmits during the contention minislot with probability p.
The probability of a successful transmission is that a probability the one
station transmits and is given by
Psuccess = np(1-p)n-1
- (1)
probability that a given station transmits and n-1 would not is p(1-p)n-1 and
since there is n possible ways to select a station that transmits, it is
multiplied by n.
To determine the max. achievable throughput, take derivative of eq (1)
with respect to p and set it to zero top find the value of p.
The value of p is 1/n.
The max. probability of a success is given by
= n1/n (1-1/n)n-1 = (1-1/n)n-1 -> 1/e - (2)

27

max

If the probability of success in one minislot is Psuccess , the average number

of minislots that elapses until a station successfully captures the channel
is 1/
max
If n takes a large value, then 1/ Psuccess= 1/e
Therefore, the average number of minislots until a station successfully
captures the channel is max
1/ Psuccess = e = 2.718 minislots
- (3)
Max. throughput in CSMA/CD system occurs when all of the channel time
is spent in frame transmission followed by contention intervals.
Each frame transmission take X seconds followed by a period t prop during
which stations find out that the frame transmission is completed and then
a contention period of average duration e x 2 t prop

When a is small, i.e. tprop << X, the CSMA-CD performs best and all
CSMAs are better than ALOHAs.
When a is approaching 1, CSMAs performs worse than ALOHA.
ALOHAs are not sensitive to a because they do not depend on reaction
time.
28

PROBLEMS:
P1: Suppose that the ALOHA protocol is used to share a 56 kbps satellite
channel. Suppose that packets are 1000 bits long. Find the maximum throughput
of the system in packets/second.
Soln.:
Given, R = 56 x 103 bps, L = 1000 bits
We know that the max. throughput achieved by ALOHA (Pure ALOHA) =
0.184
Max. throughput for the given channel of 56 kbps = 56000 bits/sec x 0.184
= 10304 bits/sec
Max. throughput in packets/sec = 10304 / 1000 =10.304.
P2: Suppose a group of N stations share a 56 kbps pure ALOHA channel. Each
stations outputs 1000 bit frame on an average of once every 100 sec, even if the
previous one has not yet been sent (e.g. the stations can buffer outgoing
frames). What is the maximum value of N?
Soln.:
Given, R = 56 x 103 bps, L = 1000 bits
Each station generates 1000 bit frame on an average of once every 100
sec i.e. at the rate of 10 bps.
We know that the max. throughput achieved by ALOHA (Pure ALOHA) =
0.184
Max. throughput for the given channel of 56 kbps = 56000 bits/sec x 0.184
= 10304 bits/sec
Since each station requires 10 bps, N = 10304 / 10 = 1030 stations.
P3: Let G be the total rate at which packets are transmitted in a slotted ALOHA
system. What proportion of slots go empty in this system? What proportion of
slots go empty when the system is operating at its maximum throughput?
Soln.:
Proportion of slots that go empty in Slotted ALOHA system = P[0
transmission] = [G0/0!]e-G = e-G
Max. throughput for is observed when G = 1, so, S =0.368
Proportion of slots that go empty when the system works at max.
throughput = e-1 = 0.368
P4: Measurements of a slotted ALOHA channel with an infinite number of users
show that 10% of the slots are idle.
a) What is the channel load, G?
b) What is the throughput?
c) Is the channel underloaded or overloaded?

29

Soln.:
a) Proportion of slots that go idle in Slotted ALOHA system = P0 = P[0
transmission] = [G0/0!]e-G = e-G, therefore G = -lnP0 = -ln0.1 = - (-2.303) = 2.303
b) S = Ge-G =2.303 x 0.1 = 0.2303
c) If G > 1, then the channel is said to be overloaded, since G = 2.303, the
channel is overloaded.
P5: M terminals are attached by a dedicated pair of lines to a hub in a star
topology. The distance from each terminal to the hub is d meters, the speed of
the transmission lines is R bits/second, all packets are of length 12500 bytes,
and the signal propagates on the line at a speed of 2.5 x10 8 meters/second. For
the four combinations of the following parameters {d = 25 meters or d = 2500
meters; R = 10 Mbps or R = 10 Gbps}, compare the maximum network
throughput achievable when the hub is implementing:
Slotted ALOHA;
CSMA/CD.
Soln.: Given, L = 12500 x 8 = 100000 bits, v = 2.5 x108 meters/second,

d = 25 meters, R = 10 Mbps
Slotted ALOHA = 0.368
CSMA/CD = 1/ (1 + 6.4a) 1/ (1 + 6.4 (tprop / X))
= 1/ (1 + 6.4 [(25/ 2.5 x108) / (100000 / 10 x 106)]

d = 25 meters, R = 10 Gbps
Slotted ALOHA = 0.368
CSMA/CD = 1/ (1 + 6.4a) 1/ (1 + 6.4 (tprop / X))
= 1/ (1 + 6.4 [(25/ 2.5 x108) / (100000 / 10 x109)]

d = 2500 meters, R = 10 Mbps

Slotted ALOHA = 0.368
CSMA/CD = 1/ (1 + 6.4a) 1/ (1 + 6.4 (tprop / X))
= 1/ (1 + 6.4 [(2500/ 2.5 x108) / (100000 / 10 x106)]

d = 2500 meters, R = 10 Gbps

Slotted ALOHA = 0.368
CSMA/CD = 1/ (1 + 6.4a) 1/ (1 + 6.4 (tprop / X))
= 1/ (1 + 6.4 [(2500/ 2.5 x108) / (100000 / 10 x109)]

Homework Problems:
HW1: 10,000 airline reservation stations are competing for the use of a single
slotted ALOHA channel. The station makes 18 requests/hour. A slot is 125 sec.
What is the approximate total channel load?
HW2: A large population of ALOHA users manages to generate 50 requests/sec,
including both originals and retransmissions. Time is slotted in units of 40 msec.

30

a) What is the chance of success on the first attempt?

b) What is the probability of exactly k collisions and then a success?
c) What is the expected number of transmission attempts needed?
HW3: Consider four stations that are all attached to two different bus cables.
The stations exchange fixed-size packets of length 1 sec. Time is divided into
slots of 1 sec. When a station has a packet to transmit, the station chooses either
bus with equal probability and transmits at the beginning of the next slot with
probability p. Find the value of p that maximizes the rate at which packets are
successfully transmitted.
HW4: A channel using random access protocols has three stations on a bus with
end-to-end propagation delay t. Station A is located at one end of the bus, and
stations B and C are together located at the other end of the bus. Frames arrive
at the three stations and are ready to be transmitted at stations A, B, and C at the
respective times tA = 0, tB = t/2, and tC = 3t/2. Frames require transmission
times of 4t. In appropriate figures, with time as the horizontal axis, show the
transmission activity of each of the three stations for ALOHA, Non-Persistent
CSMA, Non-Persistent CSMA/CD

SCHEDULING

Random access schemes are suitable for light traffic.

When traffic is heavy, random access schemes results in large variations
of frame delays.
Scheduling schemes provide orderly access to transmission medium for
stations.
Good for heavier traffic.

Scheduling schemes:

Reservation systems (used in GPRS)

Polling systems (used in HDLC)
Token passing systems (used in IEEE 802.5)

RESERVATION SYSTEMS

Stations take turns in transmitting its frame.

Transmissions from stations are organised into cycles that can be of
variable length.
As shown in fig. below, reservation interval consists of M minislots, one
minislot per station.
Stations announce their interest in transmitting their frame by broadcasting
reservation bit during their minislot.

31

Transmission takes place in the order specified in reservation interval.

Maximum achievable throughout

Assume,
The propagation delay is negligible.
Frame transmission times are X = 1 unit times
Reservation minislots are v time units, v < 1.
One minislot used for one frame transmission.
Each frame transmission requires 1 + v time units.
Maximum throughput occurs when all stations are busy and is given by

If the propagation delay is not negligible, reservations would not take

effect until some fixed number of cycles.
Length of the cycle corresponds to the number of stations having frame to
transmit.
If the cycle length is greater than round-trip propagation delay, the
reservation would take effect in second cycle as shown in fig. below.

Maximum achievable throughout

Stations can reserve more than one slot per frame transmission per
minislot.
Suppose a minislot can reserve up to k frames.
Maximum cycle size occurs when all stations are busy and is given by Mv
+ Mk time units.
Maximum throughput is given by

32

If M (number of stations) become very large, Mv would form a significant

overhead.
This would become worst when stations transmit infrequently and simply
reservation minislots are incurred every cycle.
Do not allocate a minislot to each station, let stations contend for
reservation minislot.
If slotted ALOHA is used, the each successful reservation require 1 /
0.368 = 2.71 minislot on average.
The maximum achievable throughput is
max

1
1 2.71v

POLLING SYSTEMS

Stations take turns in accessing medium.

At any time in the system, only one station would transmit.
When a station completes it transmission, the turn is given to another
station by some mechanism.

In fig. above, central station sends polling message to a particular station.

Station sends its inbound frame and indicates its completion using goahead message.
Polling may be round-robin or any other predetermined order.

33

Frequency-division duplex (FDD): one frequency band for transmission of

outbound frames and another frequency band for transmission of inbound
frames
Time-division duplex (TDD): same frequency band for transmission of both
inbound and outbound frames but transmission alternates.

No central controller.
All stations can receive the transmission from all other stations.
All stations agree upon a polling order list using some protocol.
After a station completes its transmission, it polls next station using polling
list.

Fig. above describes a polling system.

Walk time is the time that elapses while polling message propagates and
until polled station begins transmission.
Total Walk time is the sums of the walk times in one cycle.
The walk time between two consecutive stations t depends on several
factors:
propagation time for the signal to travel from one station to other.
time required for a station to begin transmitting after it has been
polled.
time required to transmit the polling message.
Cycle time Tc is the total time that elapses between start of two
consecutive polls of the same station.

34

It is equal to sum of the M walk times and the M station transmission

times.
Average cycle time E[Tc] ?
Let /M frames/sec be the average arrival rate of frames for transmission
from a station.
Let E[Nc] be the average number of message arrivals to a station in one
cycle time.
If we assume that all messages that arrive in a cycle time are transmitted
E[Nc] = /M E[Tc]
Assume that all stations have same frame transmission time, X.
Therefore, time spent at each station is E[Nc]X + t , where t is the walk
time.
The average cycle time is M times the average time spent at each station.
E[Tc] = M {E[Nc]X + t } = M {/M E[Tc]X + t }
Solving for E[Tc],

Under light load, no stations have message to transmit and cycle time is
required only for polling.
Under high load, cycle time can increase without bound.

TOKEN PASSING RINGS

A token ring system is shown in the fig. above. A station can transmit data
only after it acquires a token. A token would be circulating round the ring.
If the time to circulate around the ring < time to transmit a frame, then the
arriving information at the sender station corresponds to its own bits
transmitted.
If the time to circulate around the ring > time to transmit a frame, then
more than one frame may be present in the ring.

35

Frame inserted into the ring must be removed

Destination station would remove
Source station would remove (acts as acknowledgement)

Token ring systems differ based on token reinsertion approaches used.

Multitoken
Single token
Single frame
Ring latency: the number of bits that can be simultaneously in transit
around the ring.

Multitoken: the free token is transmitted immediately after the last bit of the
data frame.
Several frames may be in transit in the ring.

Single token: the free token is transmitted after the last bit of the busy
token is received back and the last bit of the data frame transmitted.

Single frame: the free token is inserted after the transmitting station has
received the last bit of its frame.

36

Token ring systems limit on the time a station can transmit

Number of frames
Duration of transmission
The number of frames transmitted per token would affect the maximum
achievable throughput.

Maximum achievable throughput.

Suppose one frame can be transmitted per token.

Let be the ring latency (in seconds) and a be the ring latency
normalized to frame transmission time.

Where is the total propagation delay around the ring, b is the number of
bit delays in an interface.
Maximum throughput occurs when all stations transmit a frame.

Multitoken:
Total time taken to transmit the frames from the M stations is MX + . The
maximum throughput is

Single token:
Effective frame transmission is the maximum of X and
The maximum throughput is

37

Single frame:
Effective frame transmission is always X +
The maximum throughput is

From Fig. above it is clear that achieves better throughput for different
values of a when compared to single token and single frame operations.

PROBLEMS
P6: Consider the star-topology network in problem 5 when the token-ring
protocol is used for medium access control. Assume single-packet operation,
eight-bit latency at each station, M = 125 stations. Assume a free token is three
bytes long.
a. Find the effective packet transmission time for the four combinations of d
and R.
b. Assume that each station can transmit up to a maximum of k packets per
token. Find the maximum network throughput for the four cases of d and
R.
Soln. Given,
M = 125, Lpacket = 12,500 bytes, Ltoken = 3bytes,
b = 8bits, v = 2.5 108 m/sec
The distance from each terminal to the hub is d meters. Assume the distance
traveled inside the hub is 0. The total distance around the ring is then M2d.
Xeff = token transmission time + packet transmission time +
ring latency

38

The maximum throughput occurs when all stations transmit k packets per token.
After completing the transmission of k packets, each station then transmits a free
token into the ring.

d = 25 meters, R = 10 Mbps

d = 25 meters, R = 10 Gbps

d = 2500 meters, R = 10 Mbps

39

d = 2500 meters, R = 10 Gbps

P7: A wireless LAN uses polling to provide communications between M

workstations and a central base station. The system uses a channel operating at
25 Mbps. Assume that all stations 100 meters from the base station and that
polling messages are 64 bytes long. Assume that packets are of constant length
of 1250 bytes. Assume that stations indicate that they have no packets to
transmit with a 64-byte message.

Soln. Given, d = 100 m between the base station and the stations, v = 3 x
108 m/sec, tprop = 100 / (3 x 108) = 0.33 msec,
R = 25 Mbps, Xpacket = 10000/25M = 400 msec, Xpoll = 512/25M = 20 msec,
R = 2.5 Gbps, Xpacket = 10000/2.5G = 4 msecXpoll = 512/2.5G = 0.2 msec
Xend = Xpoll

What is the maximum possible arrival rate that can be supported if stations are
allowed to transmit an unlimited number of packets per poll?

What is the maximum possible arrival rate that can be supported if stations are
allowed to transmit N packets per poll?

40

Repeat parts (a) and (b) if the transmission speed is 2.5 Gbps.

P8: A token-ring LAN network interconnects M stations using a star topology in

the following way. All the input and output lines of the token-ring station
interfaces are connected to a cabinet where the actual ring is placed. Suppose
that the distance from each station to the cabinet is 100 meters and that the ring
latency per station is eight bits. Assume that packets are 1250 bytes and that the
ring speed is 25 Mbps.
Soln.

41

What is the maximum possible arrival rate that can be supported if stations are
allowed to transmit an unlimited number of packets/token?

What is the maximum possible arrival rate that can be supported if stations are
allowed to transmit 1 packet/token using single packet operation? Using multiple
token operation?

Repeat parts (a) and (b) if the transmission speed is 2.5 Gbps.

42

HW4: Suppose that a LAN is to carry voice and packet data traffic. Discuss what
provisions if any are required to handle the voice traffic in the reservation, polling,
token ring, ALOHA and CSMA-CD environments. What changes if any are
required for the packet data traffic?
CHANNELIZATION

In a network consisting of M stations, the transmission medium is divided

into M channels and allocated for transmission of information from each
channel.
Suitable for steady flow of information, not for bursty data traffic.

Channelization Techniques:

FDMA ( Frequency Division Multiple Access)

TDMA ( Time Division Multiple Access)
CDMA ( Code Division Multiple Access)

FDMA

Transmission medium is divided into M separate frequency bands.

Guard bands are used to reduce the channel interference.
If the total available bandwidth is W (R bps), then each station can
transmit at a rate of R/M bps (ignoring guard bands).

43

TDMA

Stations take turns in accessing medium.

Transmission from stations is separated in time.
Stations transmit information in their slots allotted in each TDMA cycle.
Stations transmit at R bps for 1/M of the time.
Unlike TDM system, it uses guard times since each station may
experience different propagation delays.
Also, a preamble signal used for synchronization.

CDMA

Transmission from different stations occupy entire frequency band at the

same time.
Transmissions are separated by using different codes for producing
signals from stations.
The receiver uses these codes to recover the signal from the desired
station.
User information is transmitted at R bps.
Each user bit is transformed into G bits by multiplying user bit value (+1 or
-1) by G chip values (sequence of +1 and -1).
Chip values is a unique binary pseudorandom sequence.
The result is modulated and transmitted.
G is called spreading factor and is selected such that it occupies entire
frequency band.
44

Ex. Suppose transmitter wants to transmits +1, then it is multiplied by chip

sequence c1, c2, cG (each taking either +1 or -1). At the receiver, the
received signal is correlated with know chip sequence. i.e.
c1.c1+ c2.c2 + +cG.CG = G.
If correlated with any other chip sequence the result would be zero or
close to zero.

45

Orthogonal sequences to eliminate interference between channels.

Orthogonal spreading using Walsh functions.

46

Walsh-Hadamard matrix provides orthogonal spreading sequences of length n =

2m.

47