Communication Networks

Download Report

Transcript Communication Networks

Networks
C
2
B
3
1
D
5
A
4
6
7
F
E
Router
End-node
link
Network:end-node and router
A, B, C, D, E, and F are all
end nodes and 1 through 7
are all routers. Each end-node
is attached to a router by a
link.
The end-nodes are actual
computers. The function of a
router is only to provide
switching facilities to move
the message from one node to
another node until they reach
their destinations. For
instance, message is
transmitted from source node
A to destination node D
through routers 4, 5, and 3.
Routing Techniques
Static routing
Routing
Dynamic routing
Static Routing: In static routing, for a given source, destinations
and routes are all predetermined. They are assigned during the
design of a network. For instance, node A can send message to node
D via 4,5, 3 . Node A can also send message to node D via 1, 2, 3.
However, node A cannot send message using the route
A 4 5
6
3
D
Advantage: easy to design and set-up cost is not high
Disadvantage: not very efficient, delay can be very long.
Dynamic Routing
Dynamic Routing: In dynamic routing, the routes are calculated
when they needed. The routes are not predetermined.
Advantages: more efficient, inherently more fault-tolerant.
Disadvantages: Set-up cost is high and it is fairly complex to design.
Network Topology: Physical & Logical
• The topology of a network defines how the nodes of a network
are connected.
• A network is defined by a physical topology and a logical topology.
• Physical topology defines how the nodes of the network are
physically connected .
• Logical topology defines how the nodes of the network are
logically connected. Dedicated connections between certain selected
source-destination pairs are established using the underlying
physical topology.
• In order to have an efficient system, the logical topology should be
chosen such that either the average hop distance or the packet
delay or the maximum flow on any link must be minimal.
• Another important issue to select the logical topology is the
simplicity of the routing.
Multi-hop Network
w
4,
w6
1
w 1, w 2
w 7, w 8
4
T
w 3, w 5
w1
w 1, w 7
Star
coupler
w 5, w 6
1
2
w 3, w 4
w 2, w 8
3
Physical Topology: 2 X 2 star
w4
2
w 2 w7
w6
3
w5
w 3
4
w8
Logical Topology: 2 X 2 torus
Physical and Logical topologies
• Node 1 can directly send information to node 2 and node 3 using
frequencies w1 and w2.
• For this reason, node 1 is directly logically connected to node 2
and node 3. Although, node 1 is not directly physically connected
to node 2 and node 3.
• Now, if node 1 wants to send message to node 4, it has to multihop through either node 2 or node 3.
• For this reason, node 1 is not directly logically connected to node
4.
• If node 1 communicates with node 4 via node 2, there will be a
wavelength conversion from w1 to w3 at node 2.
Communication Networks
Network topologies:
1
2
3
4
The Bus Network
Hub/Tree Network
Star Network
Ring Network
Bus Network
• In bus network, all stations are attached to a single cable.
• When a station sends a message, it is broadcast down on the cable in
both directions. Terminators at the end of the cable prevent the signal
from reflecting back to the sender.
• All stations on the cable constantly monitor for messages meant to
them. When a station detects a message meant for it, it reads the
message from the cable and the other stations will ignore it.
• Since all stations are sharing the same cable, some form of control is
needed to make sure which station will transmit when, otherwise there
will be a collision.
• Advantages: Easy to wire, quick response, less expensive, and if one
station dies, it has no effect on the total network. I uses less cable than
other topologies
• Disadvantage: Bad connection to the cable can short it and bring
down the entire network, limited number of devices can be attached,
difficult to isolate problems, sharing cable slows down response rates.
A bus network topology
C1
C5
C2
C3
C4
C6
Use: This topology is useful
in LANs. It does not rely on
central host.This network can
still function if one of the
computers malfunctions.
Other advantages: easy to
wire, quick response, less
expensive
Problem: The main disadvantage is bad connection to
the cable can bring down the entire network. Another
problem: collision occurs when 2 nodes send
messages simultaneously.
A Hub/Tree Network
Hub
End node
Hub
Hub
Hub
A Hub / Tree Network
Hub
Tree Network
• In hub or tree network, the wires that are used to connect different
nodes are collapsed into a central unit, called hub.
• Hub does not perform switching function.
• It consists of repeaters that retransmitted all the signals from
nodes to all other nodes in the same way.
• The multipoint nature of tree topology gives rise to several
problems.
• The access control is fairly difficult to determine the time slot for
each node. Another problem with multipoint is signal balancing.
• When two device exchange data over a link, the signal strength of
the transmitter should be adjusted.
• If it is low, after attenuation, it might not meet the minimum
signal strength requirement of the transmitter and minimum signal
to noise ratio. If it is very high, it can overload the circuitry of the
transmitter resulting in harmonics and other spurious signals.
A ring network topology
C1
C6
C2
C5
C4
Use: This topology is useful
in LANs. It does not rely on
central host.This network can
still function if one of the
computers malfunctions.
Problems:
C 3 The connecting wire, cable,
or optical fiber forms a closed
loop. Data are passed along the
ring from one computer to
another.
Ring Network
• Ring network consists of nodes that are joined by point-to-point
connection to form a closed loop or ring.
• The transmitted signal is regenerated at each node.
• It is a shared-access network and it has the capability of
broadcasting messages.
• It needs some form of access control to determine which node
will transmit when.
Advantages
• The most important advantage of this network is its point-topoint connection. The electronics and maintenance of point-topoint links are much simpler than multi cast links.
• No collisions
• No terminators required
Ring Network
• The structure of this network is very simple and it is very easy to
insert and delete any number of nodes. For this reason, ring
network is easily scalable and highly modular.
• Fault isolation and recovery are much simpler than tree.
Disadvantages:
• A break in the ring will bring it down. It can bypass the faulty
route for bidirectional ring, otherwise it offers only one route for
unidirectional ring.
• Requires more cable than a bus.
• Another important drawback of ring network is its large
diameter. If diameter of a network is high, the number of edges
involved in each communication will also be very high resulting
in high signal attenuation and network blocking probability.
A star network topology
C1
C6
C2
Host
CPU
C3
C5
C4
Use: This topology is
useful for applications
where processing must
be centralized and some
can be performed locally.
Problems:
All communication pass
through the central computer.
The communications in the
network will stop if the host
computer stops functioning.
Star Network
• In star network, each station is connected via a point-to-point link
to a central point.
• This central point is called hub, multi port repeater, or
concentrator.
• The central point may be “passive”, “active”, or “intelligent”.
• A passive hub simply connects the arms of a star, no signal
regeneration is performed.
• An active hub is like a passive hub, except that it regenerates
signals.
• Intelligent hubs not only regenerate signals but also perform
activities such as intelligent path selection and network
management.
Star Network
•
•
•
•
•
Advantage:
The main advantage of the star architecture over the bus is
its small excess loss, since in case of n-user star the excess
loss increases logarithmically with n, while in case of nuser bus it increases linearly with n.
Since all information in a star network goes through a
central point (Hub), stars are easy to troubleshoot.
Easy to add devices as network expands.
One cable/node failure does not bring the entire system
down. Easy to find device/cable faults
Easy to find device/cable faults
Star Network
Disadvantages:
• Requires more cable than ring or bus networks
• Since all information in a star network goes through a
central point, stars are more vulnerable. Failure of the
central bud can bring down the entire system down.
• Costs are higher (installation and equipment) than most
bus networks.
Complex Networks: De Bruijn graphs
000
001
100
010
no. of nodes
=2^3=8
101
Advantages: diameter
= max hop no. = 3
011
110
111
Complex Networks:ShuffleNet
0
4
0
1
5
1
2
6
3
7
2
3
A ( 2, 2 ) Shuffle Net.
A (p,k) ShuffleNet consists of N = kpk nodes, k is the number of
columns, each contains pk nodes, and the kth column is wrapped
around to the first to form a cylinder.
Complex Networks: Manhattan Street Network
0,0
0,1
1,0
1,1
2,0
2,1
3,0
3,1
0,2
0,3
1,2
1,3
2,2
2,3
3,2
3,3
A 4 x 4 Manhattan Street Network
An NxM Manhattan Street Network consists of N x M nodes.There
is an even number of rows and columns with two links arriving at
and the two links leaving each node. It performs best if the number
of rows and columns are the same.
Complex Networks:Binary Hypercube
010
000
011
001
111
110
100
101
An 8-node binary hypercube
Hypercube architecture was originally proposed for multiprocessor
architecture, but it can be used as the logical topology for multi-hop
Network. Hypercube is popular due to : symmetry, small diameter,
strong connectivity, relatively small link complexity.
Fault-tolerant network
Fault-tolerant network provides
• reliable operation in faulty networks.
• alternate ways of communication.
Faults in optical networks
channel fault
link fault
node fault
Channel fault
• Why and when it occurs?
A channel fault occurs when a single wavelength channel on a link
between two nodes has failed due to the failure of the laser or
receiver for that wavelength channel, or due to cable disconnections.
• How to manage?
Thus fault can be managed by routing the traffic to a spare channel
on the same physical link and bypassing the faulty channel.
Link fault
• Why and when it occurs?
A link fault occurs due to fiber cut or due to noise
(particularly those running over long distance).
• How to manage?
It can be managed by using a bundle of protection fibers in
addition to the working fiber. The performance or survivability
is limited by the number of links in each bundle.
Node fault
• Why and when it occurs?
A node fault occurs due to power outages or catastrophic failure.
How to handle?
It is complicated to handle since when this fault occurs, a part of
the conversion capability of the network as a whole is lost and
leaves an open circuit and can result in deadlock of all routing
networks.
One solution is to introduce redundancy in the internal
connections of the networks rather than in the connections of
the nodes..
Survivable route graph
G
G
A
F
A
F
B
B
E
C
D
A Network.
E
C
D
Survival route graph.