Download Network Fundamentals: Information Transmission, Connectivity, and Switching Techniques and more Lecture notes Communication in PDF only on Docsity!
TCOM 370 NOTES 99-
NETWORKING AND COMMUNICATIONS
Communication Networks
Allow Exchange of Information between Users
- telephone network for voice communication
- interconnected computers and peripherals, a computer network
User:
- Human, application program in a PC, printer, file server
Network:
- Network consists of point-to-point links between certain nodes.
- All node pairs not connected with dedicated point-to-point links; links can be shared
- Nodes are terminal (user) nodes or communication (network) nodes
A network of interconnected nodes for information exchange
Hub
switch (bridge) PC
PC
server
PC PC
printer
Hub
In telephony , a sampling rate of 8 kHz (80000 samples per sec) is used with 8 bits for each numerical sample value, giving a bit rate of 64 kbps
General Result:
Sampling Theorem: If waveform does not have frequencies beyond a maximum frequency of F Hz., then samples at a minimum sampling rate of 2F samples/sec contain all the information in the original waveform
011
110
111
100 101
000
000 000
100
011 011
001
101
001
011
001
110
011 010
001
010
101
000
010
001
000
100
101
Time between samples = T
Example of Sampling/Quantization, Sampling rate 1/T, 3-bit or 8-Level Quantization
time
signal amplitude
Connectivity Between Terminal Nodes
Circuit Switching:
Common example: Public Switched Telephone Network (PSTN).
Dedicated communication path established between two terminal nodes (e.g telephones) through set of network nodes, during call set-up.
Resources are reserved for signaling (at 64 kbps in telephony) between the terminal nodes through this path.
There is a call set-up delay overhead.
Network Node
Terminal Node
Circuit Switching
= dedicated path
- Numerical Example of Store-and-Forward Switching
4 hops (point-to-point links) between two terminal nodes; 3200 Message Bits; Transmission rate 9600 bps on all links; 24 overhead bits [Header + Trailer] for each packet; 1024-bit fixed packet size; 1 ms. (0.001 sec) per- hop signal propagation delay.
1 sec. call set-up time for circuit switched connection across 4 hops.
What is total time to send the complete message using circuit switching and packet switching? What is the transmission delay for each bit between the terminal nodes?
Circuit Switching:
3200 bits at 9600 bps ⇒ 0.333 sec. message duration. total propagation delay 0.004 sec., Total time for message is 0.333+0.004 +1 = 1.337 sec.
Transmission delay is 0.004 sec. , since bits are not transmitted during call set-up.
Packet Switching:
Number of packets = 4 (1024-24 = 1000 bits of message data for first three packets, fourth packet has 200 bits of message data and 800 dummy bits) Packet duration = 1024/9600 = 0.107 sec.
Entire 1024-bit packet received by each node from preceding node in 1024/9600 + 0.001 = 0.108 s.
Total message time is therefore 4 x 0.108 + 3 x 0.107 = 0.753 sec. (because there are 4 hops, and 3 packets in succession after the first complete packet is received at terminal node)
Transmission delay is 4(0.001) + 3(0.107) = 0.325 sec.
Note: Short packets reduce transmission delay due to store/forward nodes. There may also be processing delays at nodes. Shorter packets inefficient due to overhead bits. For circuit switching, set-up time can be a major contribution to message delay.
Hop 4
time
Node 2
Node 3
Node 4
Terminal Node
Hop 2
Hop 1
Hop 3
Terminal Node
Packet Duration
Propagation 0.108 s Time per Hop
The Internet
• World-wide interconnection of individual networks.
- Each host computer name (e.g. seas.upenn.edu) has unique corresponding 4-byte or 32-bit internet address (e.g. 130. 91. 5. 147).
- Logical organization for Internet names and addresses. First one to three bytes of internet address is network identifier, rest form the specific host address on this network.
- Domain Name System (DNS) allows name servers to provide addresses corresponding to names, to requesting hosts.
Internet Protocol or IP refers to network protocol that allows end-to-end
delivery of packets between terminal nodes or hosts on the Internet.
IP packets may be up to 64 kbytes long.
Transmission Control Protocol (TCP) supervises packet transport.
Allows error-free packet delivery in sequence for a given connection, by using error detection with re-transmission requests.
Provides for congestion control by implementing flow-control procedures at the end-nodes.
Shared Ethernet - A Local Area Network (LAN)
Concept: A packet communication scheme with the following features:
- Shared communication medium to which a number of computers (nodes) are connected.
- Medium may be a passive cable to which nodes are connected, or electronic "hub" to which nodes are connected by wire pairs. (e.g. 10Base-2 or 10Base-T ethernets).
- The network extends over a small geographical area (e.g. within a building).
- The medium allows a high bit transmission rate , say 10 Mbps or 100 Mbps.
- All nodes have equal access to the medium. Node can send packet successfully across medium to another node if medium is not carrying any other packet at the same time (i.e. no packet collision )
Medium Access Control: Carrier Sense Multiple Access/Collision Detection (CSMA/CD)
- If packet sent by node "collides" with another, result is garbled electrical signal.
- Nodes detect "collision" by monitoring nature of electrical activity on cable; or, electronic hub sees packet arrivals on more than one port, and sends out a "collision" signal that all nodes detect.
- Thus node knows when its packet has collided with another. It then backs off for a random time duration, and attempts to re-transmit.
- Before any packet transmission, node also senses for ongoing packet transmission to avoid colliding with it
- Collision detection is still necessary because a node may start its transmission just after another node has started, without knowing about that node's transmission because of propagation delay.
If maximum transmission rate is R bps on medium, there are many nodes, and all nodes have messages to transmit (heavily loaded network), under optimum conditions on average a fraction η of R is the actual throughput in bps on the medium because of collision losses. The fraction η is called the efficiency, its value depends on the maximum propagation delay on the medium and on the packet length.
Other LANs:
Token Ring, Fiber Distributed Data Interface (FDDI), etc.
Ethernet Address:
A unique 48-bit sequence for each ethernet device. Ethernet packets contain source and destination ethernet adresses. Ethernet address of a node is not related to its network address.
Address Resolution Protocol:
An ethernet node A can discover the ethernet address of another node B with network address "N" on the same network by sending a "broadcast" ethernet packet that is read by all other nodes. The packet message is "Node of network address N, please respond". The response is an ethernet packet and carries the ethernet address "Y" of the responding node B. Node A adds the information "Network Address N = Ethernet Address Y" to its own directory.
Routers:
More sophisticated device with multiple ports each connected to a network. Assume the local networks are ethernets.
- Each router port has an ethernet address.
- Router forwards incoming ethernet packets based on network address of destination.
- Router modifies ethernet addresses in forwarded packet.
- Router makes decision on best outgoing path for packet when destination is not on an adjacent network connected to a port.
Router S
LAN
P sends ethernet packet [a,r1; X,Y; data]
P
LAN
[X,Y; data]
[s2,b; X,Y; data]
s
[X,Y; data]
r
Router R
Network Address X Ethernet address a
r1 r
Q
Network Address Y Ethernet address b
Protocol Architecture
TCP/IP Protocol Architecture
Application Layer (e.g. FTP, SMTP)
Transport Layer (TCP)
Internet Layer (IP)
LAN/Link
Physical Layer
OSI (Open System Interconnection) Reference Model
Application Layer
Presentation
Session
Transport Layer
Network Layer
Data Link Layer
Physical Layer
