BSc CSIT (TU) Science Computer Networks (BSc CSIT, CSC258) Question Paper 2080 Nepal

OSI Reference Model vs TCP/IP Protocol Suite

Both are layered network architectures, but they differ in number of layers, design origin and practical use.

OSI layers: Application, Presentation, Session, Transport, Network, Data Link, Physical. TCP/IP layers: Application, Transport, Internet, Network Access (Host-to-Network).

Layers of the TCP/IP Protocol Architecture

1. Application Layer Combines the functions of OSI Application, Presentation and Session layers. It provides services directly to user applications and handles high-level protocols, encoding and dialog control. Protocols: HTTP, FTP, SMTP, DNS, TELNET, SNMP, POP3/IMAP. Data unit: message.

2. Transport Layer Provides end-to-end (process-to-process) communication between hosts using port numbers. Two main protocols:

• TCP — connection-oriented, reliable, ordered delivery with flow control, error control and congestion control (uses the 3-way handshake).
• UDP — connectionless, unreliable but fast and low-overhead, used for DNS, streaming, VoIP. Data unit: segment (TCP) / datagram (UDP).

3. Internet (Network) Layer Responsible for logical addressing and routing of packets across interconnected networks. The core protocol is IP (IPv4/IPv6), which provides best-effort, connectionless delivery. Supporting protocols: ICMP (error/control messages), ARP (IP-to-MAC resolution), IGMP (multicast group management). Data unit: packet/datagram.

4. Network Access (Host-to-Network) Layer Corresponds to OSI Data Link + Physical layers. It defines how data is physically transmitted over a given medium, handling framing, physical (MAC) addressing, media access control, and the electrical/optical signalling. Examples: Ethernet, Wi-Fi (802.11), PPP, Token Ring. Data unit: frame / bits.

Conclusion: The OSI model is a comprehensive 7-layer reference framework, whereas the TCP/IP suite is the practical 4-layer architecture that actually runs the Internet, mapping the upper three OSI layers into a single Application layer and the lower two into the Network Access layer.

Routing

Routing is the process of selecting an optimal path for forwarding packets from a source host to a destination host across one or more interconnected networks. It is performed at the network layer by routers, which maintain routing tables built by routing algorithms. Routing algorithms are broadly classified as static (manually configured) and dynamic (e.g., Link State and Distance Vector).

Link State Routing (Dijkstra's Algorithm)

In link-state routing every router:

1. Discovers its neighbors and learns the cost to each.
2. Builds a Link State Packet (LSP) describing these costs.
3. Floods the LSP to all routers, so every router obtains the complete topology of the network.
4. Independently runs Dijkstra's shortest-path algorithm on this graph to compute the least-cost path (and hence next hop) to every other node.

Dijkstra's Algorithm Steps

Let s = source node
N' = {s}              // set of nodes whose least cost is finalized
For each node v:
    if v adjacent to s: D(v) = cost(s,v)
    else:               D(v) = infinity

Loop until N' contains all nodes:
    find node w not in N' with minimum D(w)
    add w to N'
    for each neighbor v of w not in N':
        D(v) = min( D(v), D(w) + cost(w,v) )   // relaxation

Worked Example

Consider this graph with source A (edge costs shown):

        2        3
    A ----- B ------- C
    |       |         |
   1|      3|        1|
    |       |         |
    D ----- E ------- F
        1        2

Edge costs: A-B=2, A-D=1, B-C=3, B-E=3, D-E=1, C-F=1, E-F=2.

Running Dijkstra from A:

Shortest distances from A: B=2, C=5, D=1, E=2, F=4 with paths e.g. A→D→E→F (cost 4) and A→B→C (cost 5).

Characteristics

• Fast convergence; no count-to-infinity problem.
• Each router needs full topology, so more memory/CPU.
• Used by OSPF and IS-IS routing protocols.

Distance Vector Routing Algorithm

In distance vector (DV) routing, each router maintains a table (vector) giving the best known distance (cost) and next hop to every destination. The algorithm is based on the Bellman-Ford equation:

where $D_x(y)$ is the least cost from $x$ to $y$, $c(x,v)$ is the link cost to neighbor $v$.

Operation:

1. Each router knows only the cost to its directly connected neighbors.
2. Periodically, every router sends its entire distance vector to its immediate neighbors.
3. On receiving a neighbor's vector, a router updates its own table using Bellman-Ford and keeps the minimum-cost route.
4. This repeats until the tables stabilize (converge).

Protocols: RIP, the original ARPANET routing, IGRP. It is simple and needs little knowledge of the topology, but converges slowly.

Count-to-Infinity Problem

DV routing reacts quickly to good news (a shorter path appears) but very slowly to bad news (a link goes down). When a destination becomes unreachable, routers keep advertising stale routes to each other, each incrementing the cost by one in turn, so the cost slowly counts upward toward infinity instead of immediately declaring the route dead.

Example: A—B—C with costs 1. If link B–C fails, B may learn a (false) route to C from A (which still advertises C via B). B updates to cost 3, A to 4, B to 5… looping until the cost reaches the value defined as infinity.

Methods to Solve It

1. Define Infinity: Set a small maximum metric (e.g., 16 in RIP) so the count terminates quickly and the route is declared unreachable.
2. Split Horizon: A router never advertises a route back over the interface from which it learned that route.
3. Split Horizon with Poisoned Reverse: The router does advertise such routes back, but with cost = infinity, explicitly telling the neighbor "do not use me to reach this".
4. Route Poisoning: When a route fails, immediately advertise it with infinite cost.
5. Hold-Down Timers: After a route is marked unreachable, ignore updates about it for a set period to let the bad news propagate fully.
6. Triggered Updates: Send an update immediately when a change occurs instead of waiting for the periodic timer, speeding convergence.

Purpose of ICMP

ICMP (Internet Control Message Protocol) is a network-layer protocol used by hosts and routers to send error-reporting and diagnostic/control messages about IP packet processing. Since IP itself is unreliable and connectionless, ICMP provides feedback (e.g., when a datagram cannot be delivered) and supports network troubleshooting (used by ping and traceroute). ICMP messages are encapsulated inside IP datagrams.

Common ICMP Message Types

• Echo Request / Echo Reply (Type 8 / 0) — used by ping to test reachability.
• Destination Unreachable (Type 3) — network/host/port unreachable.
• Time Exceeded (Type 11) — TTL expired in transit (used by traceroute).
• Redirect (Type 5) — informs a host of a better next-hop router.
• Source Quench (Type 4) — congestion control (now deprecated).
• Parameter Problem (Type 12) — bad IP header field.
• Timestamp Request / Reply (Type 13 / 14) — clock/round-trip estimation.

Symmetric vs Asymmetric Key Cryptography

Summary: Symmetric cryptography is fast and ideal for bulk data encryption but suffers from the key-distribution problem. Asymmetric cryptography solves key distribution and enables digital signatures, but is slower; in practice both are combined (asymmetric to exchange a symmetric session key, symmetric for the data).

Port Numbers

A port number is a 16-bit identifier (0–65535) used at the transport layer to identify a specific process or application running on a host. While the IP address identifies the host, the port number identifies which application on that host the data belongs to — enabling multiplexing/demultiplexing of many connections. Port ranges:

• Well-known ports (0–1023): standard services, e.g., HTTP=80, HTTPS=443, FTP=21, SSH=22, DNS=53, SMTP=25.
• Registered ports (1024–49151): assigned to user/registered applications.
• Dynamic/Ephemeral ports (49152–65535): temporarily allocated to client connections.

Socket Address

A socket address is the combination of an IP address and a port number, written as IP:Port (e.g., 192.168.1.5:80). It uniquely identifies one endpoint of a communication. A connection is uniquely defined by a 4-tuple:

plus the protocol. For example, a browser request might use the socket pair 192.168.1.5:51000 (client) ↔ 142.250.0.1:443 (server), allowing the OS to deliver data to the correct application.

Multiplexing

Multiplexing is the technique of combining multiple signals/data streams from several sources so they can be transmitted simultaneously over a single shared communication channel. At the sender a MUX combines the signals; at the receiver a DEMUX separates them. It improves the utilization (efficiency) of expensive transmission links.

Frequency Division Multiplexing (FDM)

• An analog technique used when the channel bandwidth is greater than the combined bandwidth of the signals.
• The available bandwidth is divided into several non-overlapping frequency bands, and each signal is modulated onto a different carrier frequency.
• Guard bands separate the channels to prevent interference.
• All signals are sent at the same time but on different frequencies.
• Examples: radio/TV broadcasting, traditional telephone trunks, cable TV.

Time Division Multiplexing (TDM)

• A digital technique that shares the link in time rather than frequency.
• Each source is given the entire bandwidth but only for a short time slot, in a round-robin fashion.
• Synchronous TDM: fixed time slots pre-allocated to each source (slots wasted if a source has no data).
• Statistical/Asynchronous TDM: slots allocated dynamically only to active sources, improving efficiency.
• Example: T1 carrier, digital telephony.

Key difference: FDM divides the frequency axis (signals coexist in time), while TDM divides the time axis (signals take turns using full bandwidth).

Guided vs Unguided Transmission Media

Guided (wired) media provide a physical conductor/path that guides signals from one point to another. Unguided (wireless) media transmit signals through free space (air, vacuum) without a physical conductor, using electromagnetic waves.

Examples explained:

• Guided: Twisted pair (UTP/STP) for LANs; Coaxial cable for cable TV; Optical fiber for high-speed, long-distance, EMI-free links.
• Unguided: Radio waves (Wi-Fi, AM/FM), Microwave (line-of-sight, terrestrial links), Infrared (short-range, e.g., remotes), Satellite communication.

LAN vs MAN vs WAN

Networks are classified by their geographical coverage:

Summary: A LAN connects devices in a limited area at high speed and low cost; a MAN spans a city and often interconnects several LANs; a WAN spans large geographic distances by interconnecting many LANs/MANs, with the Internet being the largest WAN.

Bandwidth

Bandwidth has two meanings:

• Bandwidth in hertz (Hz): the range of frequencies a channel can carry, i.e., $f_{high} - f_{low}$.
• Bandwidth in bits per second (bps): the maximum data rate (capacity) of a link. The two theorems below relate the frequency bandwidth to the achievable data rate.

Nyquist Theorem (noiseless channel)

For a noiseless channel, the maximum data rate is:

where $B$ = bandwidth in Hz and $L$ = number of discrete signal levels.

Example: For $B = 3000\,\text{Hz}$ and $L = 2$ levels:

It shows that increasing the number of signal levels can increase the data rate (but practically more levels are harder to distinguish under noise).

Shannon's Theorem (noisy channel)

For a noisy channel, the maximum (theoretical) capacity is:

where $S/N$ is the signal-to-noise ratio (linear, not in dB).

Example: For $B = 3000\,\text{Hz}$ and $\text{SNR} = 30\,\text{dB}$ (i.e. $S/N = 1000$):

Note: Nyquist gives the upper bound on signal levels for a clean channel, while Shannon gives the absolute upper bound on data rate for any noisy channel regardless of levels. The actual usable rate is the lower of the two.

Network Topologies

Topology is the physical or logical arrangement of nodes and links in a network.

1. Bus Topology

All devices connect to a single backbone cable.

• Advantages: Cheap, easy to install, less cabling.
• Disadvantages: A backbone fault brings down the whole network; difficult fault isolation; performance drops with more nodes (collisions).

2. Star Topology

All devices connect to a central hub/switch.

• Advantages: Easy to install and manage; failure of one link/node does not affect others; easy fault detection.
• Disadvantages: Central device is a single point of failure; more cable needed; cost of the hub/switch.

3. Ring Topology

Each device connects to exactly two neighbors forming a closed loop; data travels in one direction (or both in dual ring).

• Advantages: Orderly access (no collisions, token passing); performs well under heavy load.
• Disadvantages: A single node/link failure can break the ring (unless dual ring); adding/removing nodes disrupts the network.

4. Mesh Topology

Every node is connected to every other node (full mesh needs $n(n-1)/2$ links).

• Advantages: Highly reliable/robust (redundant paths); no traffic congestion; secure and easy fault diagnosis.
• Disadvantages: Very high cabling and I/O cost; complex installation and maintenance.

5. Tree (Hierarchical) Topology

Star networks connected to a central backbone in a hierarchy.

• Advantages: Scalable; easy to extend; fault isolation per branch.
• Disadvantages: Failure of the backbone/root affects large segments; more cabling; harder to configure.

Hybrid Topology

A combination of two or more topologies (e.g., star-bus). Flexible and scalable but complex and costly.

Framing

Framing is a data-link-layer function in which the stream of bits received from the network layer is divided into manageable, discrete units called frames, each with a header and trailer. Framing lets the receiver know where each frame begins and ends so that fields (addresses, control, data, checksum) can be correctly identified. A common method marks frame boundaries with special flag bytes/patterns; the techniques below prevent the data from being mistaken for these flags.

Byte (Character) Stuffing

Used in byte-oriented framing where a special flag byte (e.g., 01111110 / a FLAG character) marks the start/end of a frame. If the same byte pattern appears inside the data, an escape byte (ESC) is inserted before it so the receiver does not misread it as a flag.

• Rule: before any FLAG or ESC byte occurring in the payload, the sender inserts an ESC byte.
• The receiver removes (de-stuffs) the ESC bytes to recover the original data.
• Example: ... ESC FLAG ... in data is sent as ... ESC ESC FLAG ....

Bit Stuffing

Used in bit-oriented framing where the flag is the bit pattern 01111110 (six consecutive 1s).

• Rule: whenever the sender sees five consecutive 1s in the data, it automatically stuffs a 0 after them.
• This guarantees the flag pattern never appears accidentally inside the data.
• The receiver, on finding five 1s followed by a 0, removes that stuffed 0.
• Example: data 0111111 (five 1s then more) is transmitted as 011111**0**1.

Difference: Byte stuffing inserts whole escape bytes (8 bits) and is character-oriented; bit stuffing inserts a single bit and is byte-independent, making it more efficient.