Computer Networks

(iv) Both session layer and presentation layer

The OSI model has 7 layers while TCP/IP has 4 (or sometimes described as 5). The Application layer of TCP/IP absorbs the responsibilities of OSI's top three layers — Application, Presentation, and Session. So the Session and Presentation layers exist as distinct layers only in the OSI model; TCP/IP does not define them separately.

Answer: (ii) (1) (2) and (5)

A transmission medium is the physical path over which the signal travels from transmitter to receiver; media are either guided (wired) or unguided (wireless). Twisted pair (1) and fibre optics (5) are guided media, while microwaves and satellite signals (2) form the unguided category. All three carry the signal itself, so (1), (2) and (5) are correct, which is option (ii).

Repeaters (3) are active network devices that regenerate and retransmit a weakened signal - they sit on the medium but are not a medium themselves. Analog transmission (4) is a signalling technique (how bits are represented as a waveform), not a medium.

The remaining options fail for a single reason each: (i) omits the unguided medium (2), (iii) omits fibre (5), and (iv) wrongly includes the repeater (3), which is a device rather than a medium.

(iv) Both session layer and presentation layer

TCP/IP's Application layer merges the functionality of OSI's Session, Presentation, and Application layers into one. So Session and Presentation exist only as explicit, separate layers in the OSI model.

Answer: (i) denser than

In an optical fibre the core has a higher refractive index (denser) than the cladding. Light travelling in the core hits the core–cladding boundary at an angle greater than the critical angle and is reflected back by total internal reflection, so the signal stays inside the core.

Two functions of the OSI Session layer:

1. Session establishment, maintenance and termination — sets up, manages, and gracefully closes a logical connection (session) between two communicating applications, coordinating dialogue in simplex, half-duplex, or full-duplex mode.

2. Synchronization (checkpointing) — inserts checkpoints into a long stream of data so that, if a failure occurs, transmission can resume from the last checkpoint instead of restarting from scratch.

(ii) Data Link Layer and Bit synchronization

Bit synchronization (ensuring sender and receiver clocks agree on bit timing/boundaries) is a Physical layer function, not a Data Link layer function. The Data Link layer's actual jobs are framing, error control, flow control, and media access control — not raw bit-level timing.

The other pairs are correct: Network layer does routing; Transport layer does end-to-end communication; the MAC sub-layer handles channel sharing.

(ii) Host-to-host

The TCP/IP model's Host-to-Host (Transport) layer maps directly onto the OSI Transport layer — both provide end-to-end, process-to-process delivery via protocols like TCP and UDP. Application maps to OSI's top three layers combined; Internet maps to OSI's Network layer; Network access maps to OSI's Data Link + Physical layers.

A repeater is a Physical-layer device that receives an incoming signal, regenerates (reshapes and re-times) it to its original strength and quality, and retransmits it onward. It extends the maximum distance a signal can travel over a medium by compensating for attenuation, without altering the data content. Repeaters operate purely on the electrical/optical signal and have no awareness of frames, addresses, or higher-layer data.

(iii) Wide Area Network

A WAN spans large geographical areas — cities, countries, or continents — typically connecting multiple LANs, often using leased telecom lines or satellite links. The internet itself is the largest example of a WAN.

(i) Mesh

In a mesh topology every device has a dedicated point-to-point link to every other device.

• Security: data travels along a private link, not a shared medium, so other nodes can't eavesdrop.
• Robustness: failure of one link doesn't isolate any node — alternate paths always exist.
• No traffic/collision problem: since links aren't shared, there's no contention or traffic congestion between unrelated device pairs.

Star, Ring, and Bus all rely on a shared medium or central point, so they don't offer all three properties simultaneously.

Baud rate is the number of signal units (symbol changes) transmitted per second over a communication channel. It measures the rate of change of the physical signal, not the number of bits transferred.

Baud rate equals bit rate only when each signal unit carries exactly 1 bit (binary signaling). In general:

Bit rate = Baud rate × log₂(L)

where L is the number of distinct signal levels used per symbol. With more levels per symbol (e.g. QAM), the bit rate can exceed the baud rate.

(ii) Bridge

A bridge operates at the Data Link layer and forwards frames between two LAN segments based on MAC addresses, effectively joining them into one larger logical LAN while still filtering traffic by segment. A repeater only regenerates signals within the same LAN (no frame-level forwarding), a router works at the Network layer forwarding packets between different networks (not LAN frames), and a modem converts digital signals to analog for transmission over telephone lines.

Twisting the two wires of a pair causes each wire to be, on average, the same distance from any external noise source over the length of the cable. Any electromagnetic interference (EMI) or crosstalk from a nearby wire pair therefore induces nearly equal noise voltage on both wires.

Since the receiver measures the difference in voltage between the two wires (differential signaling), noise that affects both wires equally gets cancelled out. The tighter the twist (more twists per unit length), the better the noise cancellation — this is why Cat 5e/Cat 6 cables have progressively tighter twist rates than older cabling.

(i) Host-to-host

The Host-to-Host layer in TCP/IP provides the same end-to-end, process-to-process delivery service as the OSI Transport layer, using TCP (reliable) or UDP (unreliable).

(i) Topology

Network topology describes how nodes and links are physically laid out (physical topology) or how data logically flows between them (logical topology) — e.g. bus, star, ring, mesh.

Answer: (ii) Manchester encoding

In Manchester encoding every bit has a transition in the middle of the bit interval: one bit value is a low-to-high transition and the other is a high-to-low transition (which value maps to which depends on the convention used). This gives self-clocking and no DC component.

NRZ has no guaranteed transition per bit, and carrier/amplitude modulation are analog modulation techniques.

(i) 3.2

Transmission delay = Packet size / Bandwidth

Packet size = 32 bits
Bandwidth   = 10 Mbps = 10 × 10^6 bits/sec

Delay = 32 / (10 × 10^6) s
      = 3.2 × 10^-6 s
      = 3.2 microseconds

Note: the numeric answer works out to 3.2 microseconds, not milliseconds as the question states — this is a common typo in this question across question banks. The intended/keyed answer is still (i) 3.2, matching the computed value.

(iii) Optical fiber

Optical fiber transmits data as light pulses through glass/plastic strands, offering the highest bandwidth and speed (multi-Gbps to Tbps range) among common media — far exceeding coaxial cable and twisted pair, which are limited by electrical signal attenuation and interference.

(ii) distance between the routers

Transmission delay = Packet size / Bandwidth — it depends only on packet length and transmission rate. Distance between routers affects propagation delay (= distance / signal speed), which is a separate delay component. So transmission delay does not depend on distance.

(ii) 802.3

IEEE 802.3 is the standard for Ethernet (CSMA/CD-based LANs). For reference: 802.2 is Logical Link Control (LLC), 802.4 is Token Bus, and 802.5 is Token Ring.

(ii) 12,000 bytes

In asynchronous mode, each character frame = 1 start bit + 8 data bits + 1 parity bit + 2 stop bits = 12 bits per byte.

Baud rate        = 9600 bits/sec
Bits per byte     = 1 (start) + 8 (data) + 1 (parity) + 2 (stop) = 12

Bytes/sec  = 9600 / 12 = 800 bytes/sec
In 15 sec  = 800 × 15 = 12,000 bytes

Two protocols in the channelization category of multiple access protocols:

1. FDMA (Frequency Division Multiple Access) — divides the available bandwidth into fixed non-overlapping frequency bands, permanently allocating one band per station.

2. TDMA (Time Division Multiple Access) — divides channel access into fixed time slots, with each station allowed to transmit only during its assigned slot.

(CDMA, Code Division Multiple Access, is also acceptable — it separates stations using distinct orthogonal codes rather than frequency or time.)

(iv) ARPANET

ARPANET (Advanced Research Projects Agency Network), launched in 1969 by the U.S. Department of Defense, is widely recognized as the first operational packet-switched network and the direct precursor to the modern internet.

(iv) All of the above

The Physical layer covers everything needed to physically move raw bits across a medium: mechanical specs (connector/cable shape and pin layout), electrical specs (voltage levels, signal timing), and specifications for the transmission medium itself, including optical/IR links. So all three listed items fall under its scope.

The TCP/IP protocol suite is organized into four layers, each with distinct duties:

+-------------------------------+
|   Application Layer           |  HTTP, FTP, SMTP, DNS, Telnet
+-------------------------------+
|   Transport Layer (Host-Host) |  TCP, UDP
+-------------------------------+
|   Internet Layer              |  IP, ICMP, ARP, IGMP
+-------------------------------+
|   Network Access Layer        |  Ethernet, Wi-Fi, PPP
+-------------------------------+

1. Network Access (Link) Layer
Combines the OSI Physical + Data Link layers. Responsible for:

• Physical addressing (MAC addresses)
• Framing data into frames for the transmission medium
• Media access control and error detection at the link level
• Interfacing directly with the physical hardware (NIC, cabling, wireless medium)

2. Internet Layer
Equivalent to the OSI Network layer. Responsible for:

• Logical (IP) addressing of hosts
• Routing packets across interconnected networks, choosing best paths
• Fragmentation and reassembly of packets to fit different network MTUs
• Handled mainly by IP, with ICMP for error reporting and ARP for address resolution

3. Transport (Host-to-Host) Layer
Equivalent to the OSI Transport layer. Responsible for:

• End-to-end, process-to-process delivery between applications
• TCP: connection-oriented, reliable, ordered delivery with flow and congestion control
• UDP: connectionless, best-effort, low-overhead delivery
• Segmentation of application data and reassembly at the receiver

4. Application Layer
Combines OSI's Application, Presentation, and Session layers. Responsible for:

• Providing network services directly to user applications
• Data formatting, encryption/compression (presentation concerns)
• Session management between communicating processes
• Implemented by protocols like HTTP (web), FTP (file transfer), SMTP (email), DNS (name resolution)

Each layer only interacts with the layers directly above and below it, passing data downward by encapsulation (adding headers) on the sending side, and upward by decapsulation (removing headers) on the receiving side.

TCP/IP Protocol Stack Diagram

+-----------------------------------------------------+
|                Application Layer                     |
|   HTTP | FTP | SMTP | DNS | Telnet | SNMP             |
+-----------------------------------------------------+
|                Transport Layer                        |
|              TCP            |          UDP             |
+-----------------------------------------------------+
|                 Internet Layer                        |
|          IP  |  ICMP  |  ARP  |  IGMP                  |
+-----------------------------------------------------+
|              Network Access Layer                     |
|      Ethernet | Wi-Fi (802.11) | PPP | Frame Relay     |
+-----------------------------------------------------+
              |  physical medium (copper/fiber/air)  |

Data flows top-down at the sender (each layer adds its own header — encapsulation) and bottom-up at the receiver (each layer strips its header — decapsulation).

Why layered architecture is needed

1. Modularity — each layer solves one well-defined problem (e.g. routing, reliable delivery) independently of the others, making the system easier to design, understand, and reason about.

2. Independent evolution — a layer's internal implementation can change (e.g. Ethernet replaced by Wi-Fi) without requiring changes to the layers above it, as long as the interface between layers stays the same.

3. Interoperability — standardized interfaces between layers let hardware and software from different vendors work together, since each only needs to honor the layer boundary contract.

4. Simplified troubleshooting — problems can be isolated to a specific layer (e.g. a routing issue is a Network-layer problem, not an Application-layer one), speeding up diagnosis.

5. Reusability — lower layers (e.g. IP, Ethernet) can be reused by many different upper-layer applications, avoiding duplicated effort.

Without layering, every application would need to handle addressing, routing, framing, and physical transmission itself — an unmanageable, tightly-coupled design.

The first four layers of the OSI model (bottom-up) handle everything from raw bit transmission up to end-to-end delivery:

1. Physical Layer

• Transmits raw bits over a physical medium (copper, fiber, wireless).
• Defines mechanical/electrical specs: voltage levels, cable/connector types, pin layouts.
• Handles bit synchronization, line coding, and data rate (bits/sec).
• Defines transmission mode: simplex, half-duplex, full-duplex.

2. Data Link Layer

• Organizes raw bits into frames for transmission.
• Physical (MAC) addressing — attaches sender/receiver hardware addresses to frames.
• Flow control — prevents a fast sender from overwhelming a slow receiver.
• Error control — detects (and sometimes corrects) bit errors using checksums/CRC.
• Access control — determines which device gets to use the shared medium at a given time (via MAC sub-layer protocols like CSMA/CD).

3. Network Layer

• Logical addressing — assigns and uses IP addresses, independent of physical hardware.
• Routing — determines the best path for packets across multiple interconnected networks.
• Packet forwarding — moves packets from source to destination, hop by hop, through routers.
• Fragmentation — splits packets to fit within a network segment's Maximum Transfer Unit (MTU), reassembling at the destination.

4. Transport Layer

• Process-to-process delivery — uses port numbers to deliver data to the correct application/process on the destination host, not just the correct host.
• Segmentation and reassembly — splits data into segments at the sender, reassembles in order at the receiver.
• Connection control — connection-oriented (TCP) or connectionless (UDP) service.
• Flow and congestion control — manages the rate of data transfer to avoid overwhelming the receiver or the network.

Circuit switching vs packet switching

Circuit switching: a dedicated physical path (circuit) is set up between sender and receiver before data transfer, kept for the whole session, and released afterwards (e.g. the traditional telephone network).

Packet switching: the message is split into small packets, each carrying source and destination addresses. Packets travel independently, share links, and are reassembled at the destination (e.g. the Internet).

Basis	Circuit switching	Packet switching
Path	Dedicated path set up before transfer	No dedicated path; each packet is routed separately
Phases	Setup → data transfer → teardown	No setup needed (datagram) or a virtual circuit is set up
Resource use	Bandwidth reserved even when idle; poor utilisation	Links shared through statistical multiplexing; efficient
Delay	Setup delay, then constant, low delay	Variable delay due to queuing and routing
Order of data	Arrives in order on the same path	May arrive out of order (datagram); reordered at destination
Data unit	Continuous stream	Packets
Congestion	Blocked at setup time (busy signal)	Occurs during transfer (delay or packet loss)
Reliability	A failure breaks the call	Packets can be rerouted around failures
Header overhead	None per data unit	Every packet carries a header
Best for	Real-time voice with constant rate	Bursty data traffic (web, email, file transfer)
Cost	Higher	Lower
Example	PSTN telephone network	Internet (IP)

Circuit:  A ===== S1 ===== S2 ===== B     (one reserved path)
Packet:   A -> S1 -> S2 -> B,  A -> S1 -> S3 -> S2 -> B   (paths may differ)

Duties of the Physical Layer

• Converts data bits into signals (electrical, optical, or radio) suitable for the transmission medium, and vice versa on receipt.
• Defines mechanical and electrical specs of connectors and cables (pin layout, voltage levels).
• Specifies data rate (bits/sec) and bit duration.
• Handles synchronization between sender and receiver clocks.
• Defines line configuration (point-to-point vs multipoint) and transmission mode (simplex/half-duplex/full-duplex).
• Determines physical topology (bus, star, ring, mesh).

Fiber Optic vs Twisted Pair

Aspect	Fiber Optic Cable	Twisted Pair Cable
Signal type	Light pulses	Electrical signals
Bandwidth/Speed	Very high (Gbps–Tbps range)	Lower (up to ~10 Gbps for Cat6a)
Attenuation	Very low — supports long distances without repeaters	Higher — needs repeaters more often over distance
EMI immunity	Immune to electromagnetic interference	Susceptible to EMI and crosstalk
Cost	More expensive, harder to install/splice	Cheaper, easier to install
Security	Hard to tap without detection	Easier to tap/eavesdrop
Typical use	Backbone links, long-haul, high-speed networks	LANs, short-to-medium distance connections

Importance of the Physical Layer

The Physical layer is the foundation of the entire protocol stack — every bit that any higher layer wants to send must ultimately be converted into a physical signal and pushed onto the medium. Without a correctly functioning Physical layer, no communication (however well-designed at higher layers) can occur at all. It bridges the gap between the abstract, logical world of data (bits, frames, packets) and the physical world of electrical/optical/radio signals.

Main duties of the Physical Layer

1. Bit representation — encodes bits (0s and 1s) into signals using a chosen line-coding scheme (e.g. NRZ, Manchester encoding).
2. Data rate — defines the transmission rate (bits per second) supported by the medium.
3. Synchronization of bits — sender and receiver clocks must be synchronized so the receiver correctly samples each bit.
4. Line configuration — specifies whether the connection is point-to-point (two devices) or multipoint (shared medium among many devices).
5. Physical topology — defines how devices are physically connected (bus, star, ring, mesh).
6. Transmission mode — defines the direction of signal flow: simplex, half-duplex, or full-duplex.
7. Mechanical/electrical specification — defines connector shapes, pin assignments, voltage levels, and cable characteristics.

Architecture

In a ring topology, each device connects to exactly two neighboring devices, forming a closed loop:

        A --- B
       /       \
      E         C
       \       /
        D-----'

Working

• Data travels around the ring in one direction (unidirectional) or both (bidirectional/dual-ring).
• Each node acts as a repeater: it receives a frame, checks the destination address, and either passes it along unchanged (if not the destination) or copies it (if it is the destination) before also forwarding it onward to remove it from the ring.
• Access to the ring is commonly controlled with a token (Token Ring): a special frame circulates; a node may transmit only when it holds the token, preventing collisions.

Advantages

• No collisions — orderly, predictable access via token passing.
• Equal access opportunity for all nodes (fair, no central bottleneck).
• Relatively simple to install and reconfigure by adding nodes at any point in the loop.
• Performance degrades more gracefully under heavy load than contention-based systems (like plain Ethernet/CSMA).

Disadvantages

• A single link or node failure can break the entire ring (in a simple, non-redundant ring), disrupting the whole network — mitigated by dual-ring designs.
• Adding or removing a node typically requires briefly bringing the ring down.
• Total network delay increases with the number of nodes, since each node adds processing/repeat delay.
• Troubleshooting can be harder than in a star topology, since there's no central point to monitor from.

Aspect	OSI Model	TCP/IP Model
Number of layers	7 (Physical, Data Link, Network, Transport, Session, Presentation, Application)	4 (Network Access, Internet, Transport, Application)
Development	Theoretical reference model, developed by ISO before the protocols existed	Developed based on protocols (TCP, IP) that already existed and were in use
Session/Presentation	Explicit, separate layers	Merged into the Application layer
Usage	Used as a conceptual/teaching reference; rarely implemented directly	Actually implemented and used in real-world networks, including the internet
Protocol dependence	Generic — not tied to specific protocols	Tightly tied to its own suite (TCP, UDP, IP)
Approach	Layer-first: layers defined, then protocols fitted in	Protocol-first: protocols existed, model described them afterward
Reliability guarantee	Both connection-oriented and connectionless service supported at Network layer	Only Internet layer is connectionless (IP); reliability is optional and handled at Transport (TCP vs UDP)

Similarities

• Both are layered architectures based on the concept of independent, stacked protocol layers.
• Both provide comparable Transport and Network layer functionality (delivery and routing).
• Both use encapsulation/decapsulation as data passes between layers.

Key contrast: OSI is a conceptual model used mainly for teaching and standardization discussions, while TCP/IP is the practical model that the actual internet runs on.

Aspect	TCP/IP Model	OSI Model
Layers	4 layers	7 layers
Origin	Built from real, existing protocols (TCP, IP)	Built as a theoretical standard before matching protocols existed
Session/Presentation layers	Not separate — merged into Application layer	Exist as distinct layers
Practical use	Actually used to run the internet	Mainly used as a teaching/reference framework
Layer boundaries	Looser, layers sometimes overlap in function	Strict, well-defined boundaries between layers

  OSI (7 layers)              TCP/IP (4 layers)
  ---------------             ------------------
  Application     \
  Presentation      >---------  Application
  Session          /
  Transport        ----------  Transport
  Network          ----------  Internet
  Data Link         \
  Physical            >-------  Network Access

In short, TCP/IP is a condensed, practically-driven model, while OSI is a more granular, standards-first model that's mostly used today as a conceptual reference for discussing network layers.

Ethernet is a family of wired LAN technologies, standardized as IEEE 802.3, that defines how devices on a local network frame, address, and access a shared or switched medium. Originally based on CSMA/CD (Carrier Sense Multiple Access with Collision Detection) over a shared bus, modern Ethernet mostly runs over switched, full-duplex links where collisions don't occur.

An Ethernet frame includes fields for destination/source MAC address, a type field, the data payload, and a trailing CRC for error detection.

Common cable types used for Ethernet

Cable type	Standard	Typical speed	Notes
Coaxial (Thicknet/Thinnet)	10BASE5 / 10BASE2	10 Mbps	Legacy, bus-topology Ethernet, largely obsolete
Twisted pair (UTP), Cat 5e/6/6a	100BASE-TX, 1000BASE-T, 10GBASE-T	100 Mbps – 10 Gbps	Most common in modern LANs; uses RJ-45 connectors
Fiber optic (single-mode/multi-mode)	1000BASE-SX/LX, 10GBASE-SR/LR	1 Gbps – 100+ Gbps	Used for backbones, long-distance, or high-speed links; immune to EMI

Twisted-pair copper (Cat 5e/6) is by far the most common choice for desktop/office connections today, while fiber is preferred for data-center backbones and long-haul links.

Twisted-pair cable consists of two insulated copper wires twisted together in a helical pattern. Its key transmission characteristics are:

1. Signal type — carries analog or digital electrical signals.

2. Bandwidth — limited compared to coaxial or fiber; UTP Cat 5e/6 supports up to a few hundred MHz to ~500 MHz of usable bandwidth, enough for 1–10 Gbps over short distances.

3. Attenuation — signal strength decreases with distance and increases with frequency; requires repeaters/amplifiers over long runs (typically limited to ~100 m for Ethernet UTP before needing a repeater).

4. Noise susceptibility — vulnerable to electromagnetic interference (EMI) and crosstalk from adjacent pairs/cables, since the copper conductors are exposed to external fields. Twisting reduces (but doesn't eliminate) this by promoting equal noise pickup on both wires, which cancels out under differential signaling.

5. Two variants:

   • UTP (Unshielded Twisted Pair) — no extra shielding, cheaper, more prone to interference.
   • STP (Shielded Twisted Pair) — has metallic foil/braid shielding around the pairs, better noise immunity, but more expensive and harder to install.

6. Cost and ease of installation — the cheapest and easiest guided medium to install and terminate, which is why it dominates LAN cabling despite its bandwidth/distance limitations relative to fiber.

Structure of fiber-optic cable

A fiber-optic cable consists of a glass or plastic core, surrounded by cladding with a lower refractive index. Light injected into the core at an angle greater than the critical angle undergoes total internal reflection, repeatedly bouncing off the core-cladding boundary and traveling down the fiber with minimal loss. An outer jacket provides physical protection.

  +---------------------------------------+
  |              Jacket                    |
  |  +-----------------------------------+  |
  |  |            Cladding               |  |
  |  |   +---------------------------+   |  |
  |  |   |          Core             |   |  |  <- light travels here
  |  |   +---------------------------+   |  |
  |  |                                    |  |
  |  +-----------------------------------+  |
  +---------------------------------------+

Propagation modes

1. Multimode, step-index — core has uniform refractive index; light rays take multiple different paths ("modes") bouncing at different angles, causing modal dispersion (rays arrive at slightly different times), limiting distance/bandwidth. Larger core diameter (~50–62.5 µm), cheaper LEDs as light source.

2. Multimode, graded-index — core's refractive index decreases gradually from center to edge, causing light to travel in curved paths. Rays taking longer geometric paths travel faster (lower refractive index near the edge), which reduces modal dispersion compared to step-index multimode.

3. Single-mode — very narrow core (~8–10 µm), allowing only one straight-line mode of light to propagate. Eliminates modal dispersion almost entirely, supporting the highest bandwidth and longest distances, but requires a precise, more expensive laser light source.

Single-mode is used for long-haul/backbone links; multimode is used for shorter, cheaper LAN/building-scale runs.

OSI vs TCP/IP model

Both are layered reference models for network communication.

   OSI (7 layers)          TCP/IP (4 layers)
  Application  ─┐
  Presentation  ├────────▶  Application
  Session      ─┘
  Transport    ────────▶  Transport
  Network      ────────▶  Internet
  Data Link    ─┐
  Physical     ─┘──────▶  Network Access (Link)

Similarities

1. Both use a layered architecture, where each layer provides services to the layer above and uses the one below.
2. Both have application, transport, and network layers with similar functions.
3. Both use packet switching (not circuit switching).
4. Both rely on encapsulation: each layer adds its own header as data moves down.
5. Both are used to describe networking and to develop compatible, interoperable protocols.
6. Layer functions are independent, so one layer can change without affecting the others.

Differences

Basis	OSI model	TCP/IP model
Full form	Open Systems Interconnection	Transmission Control Protocol / Internet Protocol
Developed by	ISO (1984)	US Department of Defense (ARPANET)
Layers	7	4 (sometimes shown as 5)
Nature	Generic, protocol-independent reference model	Protocol-specific model built around TCP/IP
Approach	Vertical: model first, protocols later	Horizontal: protocols first, model later
Session and presentation	Separate layers	Merged into the application layer
Transport layer	Connection-oriented service only, for reliability	Both TCP (reliable) and UDP (unreliable)
Network layer	Supports both connection-oriented and connectionless	Connectionless (IP) only
Service, interface, protocol	Clearly separated concepts	Not clearly separated
Reliability of design	Well defined, but complex and rarely used in practice	Simple, practical, and the standard of the Internet
Protocols	Fit in after the model was made	Built first, e.g. HTTP, FTP, TCP, UDP, IP
Usage	Mainly for teaching and reference	Actually used in networks

Ethernet (IEEE 802.3) has evolved through several generations, each increasing speed while largely keeping the same frame format:

Type	Standard	Speed	Typical Medium
Standard Ethernet	10BASE5, 10BASE2, 10BASE-T	10 Mbps	Coax (bus) or UTP (star, via hub)
Fast Ethernet	100BASE-TX, 100BASE-FX	100 Mbps	Cat 5 UTP or fiber
Gigabit Ethernet	1000BASE-T, 1000BASE-SX/LX	1 Gbps	Cat 5e/6 UTP or fiber
10-Gigabit Ethernet	10GBASE-T, 10GBASE-SR/LR	10 Gbps	Cat 6a UTP or fiber
Switched Ethernet	—	Varies	Uses switches instead of hubs; each port is its own collision domain, enabling full-duplex, collision-free operation

Key distinctions:

• Standard/legacy Ethernet used a shared bus/hub and CSMA/CD for collision handling.
• Fast and Gigabit Ethernet kept backward-compatible frame formats but increased clock rates and improved encoding schemes to push more bits/sec over the same or better cabling.
• Switched Ethernet replaced the shared-medium model with point-to-point, full-duplex links to a switch, eliminating collisions altogether and letting all the newer speed standards run without CSMA/CD overhead.

Guided vs Unguided Media

Aspect	Guided Media	Unguided Media
Physical path	Signal is confined to a physical conductor (cable) — twisted pair, coaxial, fiber	Signal propagates freely through air/space — radio waves, microwaves, infrared
Direction control	Signal direction is fixed by the cable path	Signal can spread in multiple directions (omnidirectional) or be focused (directional, e.g. satellite dish)
Security	Harder to intercept without physical access	Easier to intercept since signals radiate openly
Mobility	Requires physical connection — no device mobility	Supports mobile/untethered devices (Wi-Fi, cellular)
Typical examples	Twisted pair, coaxial cable, fiber optic	Radio waves, microwaves, infrared, satellite links

Advantages of optical fiber over twisted-pair and coaxial cable

1. Higher bandwidth — supports much higher data rates (multi-Gbps to Tbps) than copper media.
2. Lower attenuation — signal travels much farther before needing regeneration, reducing the number of repeaters needed.
3. Immunity to EMI — since it carries light, not electrical current, it's unaffected by electromagnetic interference and doesn't cause crosstalk.
4. Better security — much harder to tap into without physically breaking the fiber (which is detectable).
5. Smaller size and lighter weight — easier to route in dense cable runs compared to bulky coaxial cable.
6. No electrical hazard — since it doesn't carry electrical current, it's safe in electrically noisy or hazardous environments (e.g. near heavy machinery).

The transfer rate (useful information rate) is the fraction of the total bit rate that actually carries information bits, excluding framing/overhead bits.

T1 (Asynchronous transmission)

Each character frame = 2 start bits + 8 information bits + 1 parity bit + 1 stop bit = 12 bits total, of which 8 are useful information bits.

Useful fraction = 8 / 12

Transfer rate (T1) = Bit rate x (useful bits / total bits per frame)
                    = 1200 x (8/12)
                    = 1200 x 0.667
                    = 800 bits/sec

T2 (Synchronous transmission)

Each frame = 3 sync characters (8 bits each) + 30 information characters (8 bits each)

Overhead bits = 3 x 8 = 24 bits
Information bits = 30 x 8 = 240 bits
Total frame size = 24 + 240 = 264 bits

Useful fraction = 240 / 264 = 0.9091

Transfer rate (T2) = 1200 x (240/264)
                    = 1200 x 0.9091
                    ≈ 1090.9 bits/sec

Result: T1 ≈ 800 bps, T2 ≈ 1090.9 bps. Synchronous transmission (T2) achieves a higher effective transfer rate because its fixed per-frame overhead (3 sync bytes) is amortized over more information bytes (30), unlike asynchronous transmission where every single character carries its own start/stop/parity overhead.

Fast Ethernet (IEEE 802.3u) raised standard Ethernet's speed from 10 Mbps to 100 Mbps while preserving the original Ethernet frame format, MAC addressing, and (originally) CSMA/CD access method, making it backward-compatible and easy to migrate to.

Structure of the Physical layer in Fast Ethernet

The Fast Ethernet physical layer is split into sub-layers to support multiple media types under one MAC layer:

+---------------------------+
|      MAC Sublayer          |   (common across all implementations)
+---------------------------+
|   Reconciliation Sublayer  |
+---------------------------+
|   MII (Media Independent   |   <- allows the MAC to work with
|       Interface)           |      different physical media
+---------------------------+
|   PCS | PMA | PMD           |   (Physical Coding, Medium
|   (media-specific)          |    Attachment, Medium Dependent)
+---------------------------+
            |
     Physical Medium

The MII (Media Independent Interface) decouples the MAC sublayer from the specific physical medium, letting the same MAC logic run over copper or fiber implementations.

Implementation types

Standard	Medium	Max segment length
100BASE-TX	2-pair Cat 5 UTP/STP	100 m
100BASE-FX	2-strand multimode fiber	~400 m (half-duplex) / 2 km (full-duplex)
100BASE-T4	4-pair Cat 3/4/5 UTP (legacy, rarely used)	100 m

Each implementation adapts the electrical/optical signaling for its medium while presenting the same 100 Mbps MAC-layer service upward.

TCP/IP Protocol Suite

+-----------------------------------------------------+
|                Application Layer                     |
|   HTTP | FTP | SMTP | DNS | Telnet                    |
+-----------------------------------------------------+
|                Transport Layer                        |
|              TCP            |          UDP             |
+-----------------------------------------------------+
|                 Internet Layer                        |
|          IP  |  ICMP  |  ARP  |  IGMP                  |
+-----------------------------------------------------+
|              Network Access Layer                     |
|      Ethernet | Wi-Fi (802.11) | PPP                   |
+-----------------------------------------------------+

Network Access Layer — combines OSI's Physical and Data Link layers. Handles physical transmission of bits, framing, MAC addressing, and media access control (e.g. CSMA/CD in Ethernet).

Internet Layer — handles logical (IP) addressing and routing. IP is connectionless and best-effort; it doesn't guarantee delivery, order, or error correction on its own. ICMP reports errors (e.g. "destination unreachable"), ARP resolves IP addresses to MAC addresses, and IGMP manages multicast group membership.

Transport Layer — provides process-to-process delivery via port numbers.

• TCP: connection-oriented (3-way handshake), reliable (ACKs + retransmission), ordered, with flow control (sliding window) and congestion control.
• UDP: connectionless, no delivery guarantee, minimal overhead — used where speed matters more than reliability (e.g. video streaming, DNS queries).

Application Layer — the layer applications interact with directly. Absorbs OSI's Session and Presentation layer duties (session management, data formatting/encryption). Protocols include HTTP (web), SMTP (email), FTP (file transfer), and DNS (name resolution).

Each layer encapsulates the data from the layer above by prepending its own header, forming: [App data] -> [+TCP hdr] -> [+IP hdr] -> [+Frame hdr] -> bits on wire.

Advantages of fiber optics over copper

1. Much higher bandwidth — supports data rates from Gbps into the Tbps range, far beyond copper's practical limits.
2. Lower attenuation — signals travel tens of kilometers before needing regeneration, versus ~100 m for copper Ethernet.
3. Immunity to electromagnetic interference and crosstalk — since it carries light, not current, it's unaffected by nearby electrical noise or other cables.
4. Better security — tapping a fiber without detection (via measurable signal loss) is far harder than tapping copper.
5. Lighter and thinner — easier to route through conduits, especially for high-density cabling.
6. No electrical/spark hazard — safe for use in flammable or electrically noisy industrial environments.

Downsides of fiber optics over copper

1. Higher cost — both cable and the optical transceivers (transmitters/receivers) are more expensive than copper equivalents.
2. Installation complexity — splicing and terminating fiber requires precision tools and trained technicians; copper (especially UTP with RJ-45) is far simpler to terminate.
3. Fragility — glass fiber can crack or break if bent too sharply, unlike more flexible copper wire.
4. No power delivery — fiber can't carry electrical power (unlike copper, which supports Power over Ethernet for devices like IP cameras or APs).
5. Harder/costlier to repair — a break requires specialized fusion splicing equipment, whereas copper can often be re-terminated with basic tools.

Overall, fiber wins decisively on performance and distance, but copper remains preferable for cost-sensitive, short-distance, or power-over-cable scenarios.

Guided Media vs Unguided Media, and Advantages of Optical Fiber

Guided vs unguided media

Guided media (wired / bounded media) are physical conductors that confine the electromagnetic signal to a fixed path: twisted-pair wire (UTP/STP), coaxial cable and optical fibre. The signal travels only along the medium, so it can be measured, tapped only physically, and it does not radiate.

Unguided media (wireless / unbounded media) have no physical conductor; the signal is radiated through air or vacuum as electromagnetic waves and reaches any receiver within range by broadcasting: radio waves (AM/FM, Bluetooth, Wi-Fi), terrestrial microwave, satellite, infrared and laser.

Property	Guided media	Unguided media
Physical path	Fixed cable required	Open air / free space
Attenuation	Higher but predictable	Governed by distance, fading, obstacles
Interference	EMI on copper; none on fibre	Heavy — fading, multipath, weather, adjacent channels
Security	Hard to tap (fibre hardest)	Easily intercepted by anyone in range
Installation & cost	Cabling, ducts, connectors	No cable, but antennas and licences
Mobility	None — host is tied to the wire	High — roaming supported
Broadcast capability	Point-to-point (needs a bus/switch)	Naturally point-to-multipoint

Advantages of optical fibre over twisted-pair and coaxial cable

1. Very high bandwidth — a light carrier in the hundreds of THz range gives data rates of Gbps (10/40/100 Gbps per wavelength) over a single fibre; copper is limited to a few hundred MHz and much lower rates over distance.
2. Very low attenuation — typically about 0.2 dB/km at 1550 nm, against roughly 0.2 dB per 100 m for UTP. Repeaters/amplifiers are needed tens to hundreds of km apart instead of every kilometre or so.
3. Complete immunity to electromagnetic interference (EMI) and no electrical crosstalk — glass is a dielectric and carries no current, so it works safely beside motors, power lines and lift shafts where copper picks up noise.
4. Higher security and privacy — a fibre cannot be tapped by induction; attempting to bend it out causes a measurable power loss, so intrusion is detectable. Copper radiates signals that can be picked up remotely.
5. Much lower bit-error rate (of the order of $10^{-9}$ or better) and no ground-loop / lightning-induced currents.
6. Light weight and small diameter — far more channels per duct, important where conduit space is scarce.
7. Electrical isolation and safety — non-conducting, no sparking hazard in explosive atmospheres, and no risk of short circuits between networks.

Disadvantages (for completeness in the exam): higher cost of cable, transmitters (laser/LED) and detectors, and the fibre is fragile and difficult to splice or connect — it needs skilled installation.

Gigabit Ethernet (IEEE 802.3z / 802.3ab) delivers 1000 Mbps (1 Gbps) and comes in several physical-layer implementations, differing by medium:

Implementation	Medium	Max distance	Notes
1000BASE-T	Cat 5e/Cat 6 UTP copper (4 pairs)	100 m	Most common for desktop/office LAN cabling
1000BASE-SX	Multimode fiber (short wavelength, ~850 nm)	~550 m	Cost-effective fiber option for shorter runs
1000BASE-LX	Single-mode or multimode fiber (long wavelength, ~1300 nm)	Up to ~5 km (single-mode)	Used for longer backbone links
1000BASE-CX	Shielded copper (twinax), short-haul	~25 m	Used for very short connections, e.g. within a server rack

All variants present the same 1 Gbps MAC-layer interface upward, differing only in the Physical layer signaling and medium used — allowing network designers to pick the right implementation based on distance, cost, and existing cabling infrastructure.

Types of delay in computer networks

When a packet travels from source to destination through a network, it experiences four types of delay at each hop (router/link):

1. Processing delay — time a router takes to examine the packet header, check for bit errors, and determine the outgoing link (routing table lookup).

2. Queuing delay — time a packet waits in the router's output buffer before it can be transmitted, depending on current congestion at that router.

3. Transmission delay — time to push all of the packet's bits onto the link, equal to Packet size / Bandwidth.

4. Propagation delay — time for a bit to physically travel from one end of the link to the other, equal to Distance / Propagation speed.

Total delay per hop = Processing + Queuing + Transmission + Propagation delay.

Numerical problem

Data to send    = 200 Mb = 200 x 10^6 bits
Bandwidth        = 1 Gbps = 1 x 10^9 bits/sec
Routers in between = 3  ->  number of links (hops) = 3 + 1 = 4
                     (A -> R1 -> R2 -> R3 -> B)
Other delays (processing, queuing, propagation) = 0

Since only transmission delay is non-zero, and (assuming store-and-forward switching at each router, the standard assumption) each router must fully receive the data before forwarding it, the data must be transmitted in full once per link:

Transmission delay per link = Data size / Bandwidth
                             = (200 x 10^6) / (1 x 10^9)
                             = 0.2 sec  (200 ms)

Total delay = Transmission delay per link x Number of links
            = 0.2 x 4
            = 0.8 sec  (800 ms)

Total delay involved = 0.8 seconds (800 ms)

Aspect	Amplifier	Repeater
Signal type	Boosts analog signals	Regenerates digital signals
Method	Increases the amplitude/strength of the entire incoming waveform, including any noise picked up along the way	Detects the original bit pattern and reconstructs a clean, fresh digital signal, discarding accumulated noise
Noise handling	Noise is amplified along with the signal — degrades over multiple hops	Noise does not accumulate — each hop starts fresh
Layer	Physical layer, used in analog transmission (e.g. old telephone lines)	Physical layer, used in digital transmission (e.g. Ethernet, digital telecom)
Typical use	Long-distance analog communication links	LANs and digital transmission lines

In short: an amplifier blindly strengthens whatever signal (and noise) arrives, while a repeater reads the digital data and retransmits a clean copy — making repeaters more reliable for digital networks over long distances.