Internet Protocol Edge Core Performance Layers OSI Homework

Chapter 1: Computer Networks and the Internet

Kurose & Ross, Computer Networking: A Top-Down Approach, 9th edition (2025) – Full lecture coverage

1. What is the Internet?

📡 "Nuts and Bolts" View

Billions of connected computing devices: hosts / end systems (PCs, servers, smartphones, IoT).
Packet switches: routers & switches forward data chunks (packets).
Communication links: fiber, copper, radio, satellite – transmission rate = bandwidth.
Network of networks: interconnected ISPs.
Protocols control sending/receiving: HTTP, TCP, IP, WiFi, 4G/5G, Ethernet.
Standards: RFC (Request for Comments), IETF (Internet Engineering Task Force).
"Fun" devices: web‑enabled toaster, smart watches, connected cars, weather forecaster.

🛠️ "Services" View

Infrastructure providing services to distributed apps: Web, streaming, email, games, e‑commerce, social media, video conferencing.
Programming interface (socket API): allows apps to "connect" and use Internet transport service – analogous to postal service.
Provides service options: reliable data delivery, real‑time delivery, etc.

The Internet is a network of networks – access ISPs, regional ISPs, tier-1 ISPs (Level3, AT&T, NTT), and content provider networks (Google, Meta, Microsoft).

2. What is a Protocol?

A protocol defines the format, order of messages exchanged between network entities, and the actions taken upon message transmission/reception. Every communication activity on the Internet is governed by protocols.

Human protocols: “What’s the time?” → reply; introductions.

Network protocols: HTTP request → HTTP response; TCP SYN → SYN-ACK.

3. Network Edge: Hosts, Access Networks & Physical Media

End systems (clients & servers) connect to the edge router via access networks. A host sends packets: it breaks an application message into chunks of L bits and transmits at rate R (transmission delay = L/R).

📡 Cable (HFC)
Hybrid fiber coax, up to 1.2 Gbps downstream, shared medium, FDM.

📞 DSL
Digital Subscriber Line, 24‑52 Mbps down / 3.5‑16 Mbps up, dedicated.

🏠 Home networks
Wireless router + modem (cable/DSL/fiber).

🏢 Enterprise (Ethernet/WiFi)
100 Mbps – 10 Gbps, switches, router.

☁️ Data center networks
10s–100s Gbps links, thousands of servers.

📶 Wireless access
WiFi (802.11) up to Gbps; 4G/5G (100's Mbps over ~10 km); Bluetooth; satellite (Starlink <100 Mbps down, 270 ms geostationary).

🔌 Physical Media

Twisted pair (copper): Category 5/6 up to 10 Gbps.
Coaxial cable: two concentric conductors, broadband, 100's Mbps per channel.
Fiber optic: glass fiber, high speed (10's–100's Gbps), low error, immune to electromagnetic noise.
Wireless radio: WiFi, 4G/5G, Bluetooth, microwave, satellite; half‑duplex; affected by reflection, obstruction, interference.

4. Network Core: Packet/Circuit Switching & Internet Structure

📦 Packet switching

Store‑and‑forward: entire packet must arrive before forwarding → transmission delay = L/R.
Queueing and loss can occur when buffers fill.
Efficient for bursty data, no call setup, resources shared.
Congestion possible → requires reliable data transfer and congestion control.

🔌 Circuit switching

End‑end resources reserved for a "call".
FDM (frequency division) or TDM (time slots).
Guaranteed performance, but idle resources cannot be used by others (no sharing).
Traditional telephone networks.

Example (1 Gb/s link, each user active 10% of time at 100 Mb/s):
• Circuit switching → max 10 users.
• Packet switching → with 35 users, probability that more than 10 are active simultaneously is < 0.0004 → supports many more users.

🌍 Internet structure: a "network of networks"

Access ISPs connect to regional / tier‑1 ISPs (Level3, AT&T, NTT).
Content provider networks (Google, Meta, Microsoft) build their own networks, bypassing tier‑1 ISPs to bring content closer to users.
Two key core functions: forwarding (local action: map arriving packet to output link) and routing (global action: determine source‑destination paths using routing algorithms).

5. Performance: Loss, Delay, Throughput

⏱️ Four sources of packet delay

d_nodal = d_proc + d_queue + d_trans + d_prop

• d_proc: nodal processing (bit error check, output link) – typically < microseconds.
• d_queue: queueing delay – depends on congestion level.
• d_trans = L/R (transmission delay).
• d_prop = d/s (propagation delay; s ≈ 2×10⁸ m/s).

📊 Queueing delay & traffic intensity

Traffic intensity = La/R (a = average packet arrival rate, L = packet length, R = link bandwidth).
• La/R → 0 : small average delay.
• La/R → 1 : delay becomes large.
• La/R > 1 : infinite average delay → buffer overflow and packet loss.

Caravan analogy: 10 cars (packet), toll booth (link), service time = transmission, propagation speed = highway speed. Illustrates store‑and‑forward and the difference between transmission and propagation delay.

📈 Throughput

Throughput = rate (bits/sec) at which bits are transferred from sender to receiver. The bottleneck link determines end‑end throughput = min(R_s, R_c) in the simple two‑link case. In complex networks, the link with the smallest capacity limits overall throughput.

🔎 traceroute

Measures real delays: source sends packets with increasing TTL values; each router returns a message. Example output (UMass to Eurecom):

1  cs-gw (128.119.240.254)  1 ms  1 ms  2 ms
2  border1-rt-fa5-1-0.gw.umass.edu  1 ms  1 ms  2 ms
...
15 fantasia.eurecom.fr  132 ms 128 ms 136 ms

Packet loss: buffers (queues) have finite capacity; a packet arriving at a full buffer is dropped (lost). It may or may not be retransmitted.

6. Protocol Layers, Service Models & Encapsulation

Imagine you want to send a gift to a friend in another country. You would: (1) choose what goes inside, (2) wrap it in a box, (3) write the address on the box, (4) hand it to a courier, and (5) the courier network moves it across the country. Each step is independent: the courier does not care what is inside the box, and if the wrapping paper rips, the gift inside is still safe.

This is exactly the idea of layering in computer networks. Complex communication is broken into layers. Each layer provides a service to the layer above it, using the services of the layer below. Changes inside one layer do not affect other layers.

Why layering helps:

Modularity: You can replace one layer (e.g., swap Ethernet for WiFi) without rewriting the layers above it.
Reuse: Many applications share the same transport layer (TCP or UDP).
Simplicity: Each layer only worries about its own job — no need to know every detail of the whole system.

📚 The 5-Layer Internet Stack

From top (closest to the user) to bottom (closest to the wire):

Layer	What it does (in plain words)	Example protocols	Unit of data
5. Application	The network service your program uses: web pages, email, file transfer, video calls	HTTP, SMTP, DNS, FTP	Message
4. Transport	Breaks the message into pieces and delivers it reliably (or quickly) from program to program on different hosts	TCP, UDP	Segment
3. Network	Routes the pieces across the Internet from one host to another host (end-to-end)	IP, OSPF, BGP	Datagram (packet)
2. Link	Moves the packet across one physical link (one hop): from your laptop to your router, router to router, etc.	Ethernet, WiFi, PPP	Frame
1. Physical	Puts the bits on the actual wire or wireless medium	Copper, fiber, radio signals	Bit

📦 Encapsulation: How a Message Travels Down the Stack

When you send a message, each layer adds its own header to the data before passing it down. Think of it like a set of nested envelopes:

Application: Your message "Hello"

Transport (TCP): Add TCP header (ports, sequence #) [TCP-H | "Hello"]

Network (IP): Add IP header (source and dest IP addresses) [IP-H | TCP-H | "Hello"]

Link (Ethernet): Add Ethernet header + trailer (MAC addresses) [Ether-H | IP-H | TCP-H | "Hello" | Ether-T]

Physical: Send as bits over the wire 010110100101...

At the receiver, the process is reversed: each layer strips off its own header and passes the remaining data up. By the time the message reaches the application, the headers are all gone and the original message is recovered.

Postal Analogy: If you mail a gift, the gift goes inside a box (application data). The box is put into a shipping carton (transport header). The carton gets a shipping label (network header). The whole thing is loaded onto a truck with a manifest (link header). At each stage, someone only reads the layer they are responsible for.

7. ISO/OSI Reference Model vs. Internet Stack

The OSI (Open Systems Interconnection) model is a theoretical framework developed by ISO in the late 1970s. It defines 7 layers, while the practical Internet protocol stack uses only 5 layers. Both describe the same communication process – OSI simply splits the top into more pieces.

Which one is actually used? The Internet uses the 5-layer model. OSI is mainly a teaching and analysis tool. However, many OSI terms are still used in practice ("Layer 2 switch", "Layer 3 router", "Layer 7 firewall").

OSI 7 Layers ↔ Internet 5 Layers — Visual Comparison

The OSI column on the left maps directly to the Internet-stack column on the right. Solid arrows (↔) show a direct 1-to-1 correspondence; the dashed amber arrows (↘) show OSI layers whose functions are absorbed into the Internet application layer.

OSI Model (7 layers)

Internet Stack (5 layers)

Application OSI Layer 7

↔

Application HTTP · DNS · SMTP covers OSI L5–L7

Presentation not a separate Internet layer

↘ folded

Session not a separate Internet layer

↘ folded

Transport OSI Layer 4

↔

Transport TCP · UDP

Network OSI Layer 3

↔

Network IP · Routing (OSPF, BGP)

Link OSI Layer 2

↔

Link Ethernet · WiFi · PPP

Physical OSI Layer 1

↔

Physical Bits on the wire

Why does the Internet stack stop at 5 layers?OSI's Presentation layer (encryption, compression, encoding) and Session layer (synchronization, checkpoints, recovery) are useful abstractions, but in practice the Internet fits these services inside the application layer itself — for example TLS in HTTPS handles encryption, and session cookies handle state. This keeps the network core simple and pushes complexity to the edges (the "end-to-end" principle).

What do Presentation and Session do, in practice?

Presentation (OSI Layer 6): Handles encoding, encryption, and compression. In the modern Internet, this is done by the application or by protocols like TLS (which encrypts HTTPS) and gzip (which compresses HTTP responses). These live in or just above the Application layer.
Session (OSI Layer 5): Manages "conversations" – opening, keeping alive, and closing sessions. In the modern Internet, this is done by cookies in HTTP (keeping user sessions), TCP connections (keeping the link open), and application logic.

Summary of the Mapping

Application (OSI L7) ↔ Application layer in the Internet stack – the top of the stack that interacts directly with user programs.
Presentation (OSI L6) and Session (OSI L5) — not present as separate Internet layers; their functions are absorbed into the application layer (e.g., TLS for encryption, cookies for session state).
Transport, Network, Link, Physical (OSI L4–L1) map 1-to-1 with the corresponding layers of the Internet stack.

Why does the Internet stack "collapse" the top layers? Because the Internet was designed with the end-to-end principle: keep the network core simple, and let the endpoints (the applications) handle complexity. So encryption, compression, and session management all live in the application layer, not in the network itself.

Homework Assignment (Theoretical, Graded)

Instructions: Answer all questions in a separate document (PDF or Word). Submit via the learning platform. This assignment counts towards your final grade (5%). Provide clear, well‑structured answers using terminology from the lecture.

Part I: Short Essays (40 points)

Explain the two perspectives of the Internet: the "nuts‑and‑bolts" view and the "services" view. Give one example for each.
Compare packet switching and circuit switching. In which scenario is packet switching clearly superior? In which scenario would circuit switching be preferred? Justify.
List the four sources of packet delay. Write the formula for each. Then describe how traffic intensity (La/R) affects queueing delay.
Define encapsulation in the context of the Internet protocol stack. Describe step by step how a message becomes a frame at the sender, and what happens at the receiver.
What are IP spoofing and Denial‑of‑Service (DoS) attacks? Propose one defense mechanism for each.

Part II: True/False + Correction (30 points)

In packet switching, the entire path’s resources are reserved before transmission begins. (True / False)
HTTP is an example of a transport-layer protocol. (True / False)
DSL access technology shares the link among several neighboring homes. (True / False)
Propagation delay depends only on the link’s transmission rate. (True / False)
The OSI model has 5 layers, while the Internet stack has 7 layers. (True / False)

For each false statement, write the correct version.

Part III: Simple Calculations (30 points)

A link has transmission rate R = 10 Mbps. You need to send a file of size 5 Mbits. Compute the transmission delay (ignore all other delays). Show the formula and the result.
Average packet arrival rate a = 400 packets/sec, packet length L = 1500 bits, link bandwidth R = 1 Mbps. Calculate the traffic intensity (La/R). Is the queueing delay expected to be small or large? Explain.
In a packet‑switched network, 25 users share a 5 Mbps link. Each user is active 20% of the time and requires 500 kbps when active. Would circuit switching or packet switching be more efficient? Justify briefly using the concept of statistical multiplexing (you may refer to the probability argument shown in class).

Grading criteria: correctness, completeness, proper use of terminology, and clarity of explanation.

Deadline: Two weeks from the lecture date.

Chapter 1 Summary

Internet: physical infrastructure (devices, links, switches, ISPs) and service provider for distributed apps (socket API).
Protocols define format, order, and actions; examples (HTTP, TCP, IP).
Network edge: access technologies (cable, DSL, Ethernet, WiFi, 4G/5G, satellite) and physical media (copper, fiber, radio).
Network core: packet switching (store‑and‑forward, queueing, loss) vs. circuit switching (FDM/TDM, reserved resources). Internet structure as a hierarchy of ISPs and content provider networks.
Performance: four delays (proc, queue, trans, prop); traffic intensity La/R; throughput bottleneck; traceroute tool.
Layering: 5‑layer Internet stack, encapsulation, OSI 7‑layer model.

Self‑review questions (exam preparation):
• Differentiate routing from forwarding.
• How does traceroute work? What information does it provide?
• What is the difference between transmission delay and propagation delay? Give a numerical example.
• Why are application‑layer protocols such as HTTP critical for end‑users?