OSI vs TCP/IP Models & Packet Encapsulation
“Layer responsibilities, protocol data units (PDU), and header encapsulation.”
1.The Principle of Layering & Architectural Separation
§ 1.1.Computer networks rely on modular layering to decompose massive communication complexity into independent functional abstractions. Each layer n provides specific services to layer (n+1) via a Service Access Point (SAP), while relying on the services of layer (n-1).
Protocols vs Services: A protocol is a set of formal rules governing peer-to-peer communication across hosts at the same layer. A service is a set of operations that a layer provides to the layer immediately above it through an interface.
Encapsulation & Decapsulation: As user data travels down the sender's protocol stack, each layer prepends a protocol-specific header (and occasionally an error-detecting trailer at Layer 2). The resulting packet is termed a Protocol Data Unit (PDU): Application Message → Transport Segment/Datagram → Network Datagram/Packet → Data Link Frame → Physical Raw Bits.
GATE questions often test exact terminology: Application Layer produces 'Messages', Transport Layer produces 'Segments' (TCP) or 'User Datagrams' (UDP), Network Layer produces 'Packets/Datagrams', and Data Link Layer produces 'Frames'.
2. OSI 7-Layer vs TCP/IP 5-Layer Architectural Comparison
Standard comparison of responsibilities, protocol units, and design philosophies:
| Feature / Aspect | Layer | OSI Model (Theoretical) |
|---|---|---|
| Layer 7 (Application) | User network interface (HTTP, SMTP, DNS, FTP) | Merges OSI Session, Presentation & App into one application layer |
| Layer 6 (Presentation) | Data serialization, ASCII/EBCDIC, SSL/TLS encryption format conversion | Handled directly inside user application libraries |
| Layer 5 (Session) | Dialogue control, token management, synchronization checkpoints | Integrated into application or transport layer |
| Layer 4 (Transport) | End-to-end process-to-process delivery (TCP, UDP with port numbers) | End-to-end host communication with port addressing |
| Layer 3 (Network) | Host-to-host routing, logical IP addressing across networks | Internet Protocol (IPv4/IPv6), ICMP, routing |
| Layer 2 (Data Link) | Hop-to-hop node delivery, framing, CRC error detection, MAC access | Network Interface / Data Link (Ethernet, Wi-Fi) |
| Layer 1 (Physical) | Raw bit transmission, signal voltages, cable pins | Physical transmission medium (Copper, Fiber, Radio) |
3.End-to-End Delay Analysis: The Four Delay Components
§ 1.3.Whenever a packet travels from source to destination across a store-and-forward packet-switched network, total node delay d_nodal comprises exactly four components: d_nodal = d_proc + d_queue + d_trans + d_prop.
1. Nodal Processing Delay (d_proc): Time taken by the router to examine the packet header, verify the checksum, and determine the output link via routing table lookup (typically microseconds).
2. Queuing Delay (d_queue): Time the packet waits in the router output buffer queue before transmission. Depends on network congestion and traffic intensity I = (L * a) / R, where L = packet size (bits), a = average arrival rate (packets/sec), and R = transmission rate (bps). If I > 1, queue grows without bound and packets drop.
3. Transmission Delay (d_trans): Time required to push all packet bits onto the physical link: d_trans = L / R, where L = packet length in bits, R = link bandwidth/capacity in bits per second (bps).
4. Propagation Delay (d_prop): Time taken for a single bit to travel across the physical medium from sender to receiver: d_prop = d / s, where d = physical distance in meters, s = signal propagation speed in the medium (typically 2 10^8 m/s in copper/fiber, 3 10^8 m/s in vacuum).
Transmission delay depends strictly on packet size L and bandwidth R (L / R). Propagation delay depends strictly on distance d and signal velocity s (d / s). Bandwidth does NOT speed up signal travel; it only pushes bits onto the wire faster!
4. GATE Step-by-Step Numerical: Store-and-Forward Packet Switching Delay
In store-and-forward packet switching, intermediate routers must receive the ENTIRE packet before forwarding the first bit to the next link!
Problem: A source host sends a file of size F = 1.5 Megabytes over a path of N = 4 identical links. Each link has bandwidth R = 10 Mbps and propagation delay d_prop = 5 ms. The file is split into packets of size P = 1500 bytes (including 20 bytes header). Neglect processing and queuing delays.
Step 1: Compute packet transmission delay per link: d_trans = (1500 8 bits) / (10 10^6 bps) = 12,000 / 10,000,000 = 1.2 milliseconds.
Step 2: Number of packets M = Total Data / Packet Data. Here, packet payload = 1500 - 20 = 1480 bytes. M = ceil(1.5 * 10^6 / 1480) = ceil(1013.51) = 1014 packets.
Step 3: Pipelined store-and-forward transmission formula: Total Time T = (Time for Packet 1 to reach destination) + (Remaining M - 1 packets arriving back-to-back at destination).
Packet 1 traverses 4 links: 4 d_trans + 4 d_prop = 4 1.2 ms + 4 5 ms = 4.8 ms + 20 ms = 24.8 ms.
Remaining (M - 1) = 1013 packets arrive one by one spaced by d_trans: 1013 * 1.2 ms = 1215.6 ms.
Total End-to-End Delay = 24.8 ms + 1215.6 ms = 1240.4 ms = 1.2404 seconds.
Primary Academic References & Exam Sources
Every concept in Notes From a B.Tech Brain is synthesized from authoritative academic texts, seminal peer-reviewed research, verified exam archives, and renowned university lectures.
Curated for clarity, rigorous numerical solving, and semester excellence:
GATE-level numerical drilling and PYQ analysis.
Step-by-step conceptual walkthroughs with worked GATE numericals.
Clear whiteboard lectures for foundational and semester exam coverage.