3 Network Fundamentals - Summary of Slides 1 - 45

Overview

Topic	Slide Range	Notes
1. History & Fundamentals of the Internet	Slides 1–16	ARPANET, TCP/IP, WWW, client-server, HTTP statelessness
2. Circuit vs. Packet Switching	Slides 17–23	FDM, TDM, statistical multiplexing, delay tradeoffs
3. Delays, Loss, Throughput	Slides 24–35	d_proc, d_queue, La/R, traceroute, throughput bottlenecks
4. Protocol Layers and Encapsulation	Slides 36–44	Layer model, encapsulation, host/router/switch roles

Historical Background and Internet Foundations (Slides 1–16)

Early developments: ARPANET, Cyclades, and ALOHANet pioneered packet switching.
The Internet emerged as a global interconnection of autonomous systems using the TCP/IP protocol suite.
Tim Berners-Lee’s World Wide Web (1989–1991) introduced:
- A unified model of hyperlinked documents (HTML)
- The HTTP protocol (stateless)
- URLs for addressing
- The browser-server interaction model
The stateless nature of HTTP means each request is handled independently — servers do not retain memory of previous interactions.

Transmission Media and Infrastructure (Slides 16–20)

Data transmission can occur over:
- Copper (UTP): electrical signals
- Fiber optics: light pulses
- Radio: electromagnetic waves (Wi-Fi, LTE)
Fiber-optic links offer high bandwidth and low latency — widely used in backbone and undersea cables.
Satellite communication has higher propagation delay (≥500 ms round-trip) due to distance (~36,000 km geostationary orbit).
Real-world systems combine many media and technologies in layered infrastructure.

Circuit Switching vs. Packet Switching (Slides 18–23)

Circuit switching: fixed, reserved paths (e.g. telephony)
- Uses TDM (Time Division Multiplexing) or FDM (Frequency Division Multiplexing)
Packet switching: data is broken into packets routed independently
- Uses statistical multiplexing
- No reservation of bandwidth; packets share the link dynamically
Trade-offs of packet switching:
- More efficient use of bandwidth under bursty traffic
- Potential for packet delay, loss, and reordering

Network Performance Metrics (Slides 24–32)

Four types of delay:

Processing delay: time to examine packet header and perform checks
Queueing delay: time waiting in the router buffer
Transmission delay:

\[ d_{\text{trans}} = \frac{L}{R} \]

where:
- \(L\): packet size (bits)
- \(R\): link bandwidth (bps)
Propagation delay:

\[ d_{\text{prop}} = \frac{d}{s} \]

where:
- \(d\): physical distance (meters)
- \(s\): signal propagation speed (m/s)

Total node delay:

\[ d_{\text{nodal}} = d_{\text{proc}} + d_{\text{queue}} + d_{\text{trans}} + d_{\text{prop}} \]

Traffic intensity and queue behavior:

Let:

\(a\): average packet arrival rate (packets/sec)
\(L\): packet size (bits)
\(R\): link bandwidth (bps)

Then:

\[ \text{Traffic intensity} = \frac{aL}{R} \]

Interpretation:

If \(\frac{aL}{R} \geq 1\): the queue grows without bound
As \(\frac{aL}{R} \rightarrow 1\): delay increases sharply

End-to-end delay over multiple hops:

\[ d_{\text{end-to-end}} = \sum_{i=1}^N (d_{\text{proc},i} + d_{\text{queue},i} + d_{\text{trans},i} + d_{\text{prop},i}) \]

where \(N\) is the number of routers.

Traceroute:

Uses IP TTL (Time-To-Live) field to probe each hop
When TTL reaches zero, routers send an ICMP “Time Exceeded” message
Allows measurement of round-trip time (RTT) per hop

Throughput (Slides 33–35)

Throughput: the rate at which data is successfully delivered (bps)

Two cases (Slide 34):

If:

\(R_S\): server’s sending rate
\(R_C\): client-side link rate

Then:

If \(R_S < R_C\), then \(\text{Throughput} = R_S\)
If \(R_S > R_C\), then \(\text{Throughput} = R_C\)

\[ \text{Throughput} = \min(R_S, R_C) \]

Summary: Protokollschichten und ihre Dienstmodelle (Slides 36 - 45)

(Slides 36–44)

This final chapter introduces the layered architecture of the Internet. It explains how each protocol layer serves the one above and relies on the one below, and how encapsulation enables structured communication.

Layering Motivation (Slide 36–37)

Networks are complex (hosts, routers, media, apps).
Solution: Schichtenarchitektur for modular design.
Each layer performs actions and uses only the services of the layer below.

Protocol Layering: Foundations (Slides 36–38)

Due to network complexity, functionality is divided into layers, each with clear responsibilities.
A layer \(k\) uses only the services of layer \(k-1\):

\[ \text{Layer } k \longrightarrow \text{uses services of Layer } (k-1) \]
Each layer communicates vertically (service interface) and defines horizontal protocols (with its counterpart on the remote host).
Layering enables:
- Modularity
- Replaceability
- Interoperability
- Abstraction from hardware details

The Internet Stack (Slide 38–39)

Layer	Function	Examples
Application	Application protocols, user data	HTTP, FTP, SMTP
Transport	Process-to-process delivery	TCP, UDP
Network	Host-to-host delivery, routing	IP, ICMP
Data Link	Frame-level delivery on local links	Ethernet, Wi-Fi, PPP
Physical	Transmission of bits over the medium	Fiber, DSL, 5G

Layers are identified by who communicates (e.g. processes, hosts, links).
Data is encapsulated step by step as it moves downward.

Protocol Scope by Device (Slide 40)

Device	Implements Up To
Host	All 5 layers
Router	Network layer (IP)
Switch	Data Link layer (MAC)

Encapsulation (Slides 41–43)

Each layer adds its own header (and possibly trailer). The result:

Frame = [Data Link hdr] + [IP hdr] + [TCP hdr] + Message + [Trailer]

At the receiver, each layer removes its own header.

Only hosts process all layers.
Routers read only IP headers.
Switches forward based on MAC addresses.

OSI Model (Slide 44)

A 7-layer reference model defined by ISO, used mostly for conceptual clarity.

OSI Layer	Added vs. Internet Model
7: Application	Matches Internet’s application layer
6: Presentation	Data format, compression, encryption
5: Session	Dialog management
4–1: Transport → Physical	Same as in Internet stack

Internet model simplifies OSI: layers 5–7 are often merged into the application.

This version keeps all major points but trims redundant explanation and tightens the wording for effective study reference.

Would you like it exported to markdown or added to your ongoing Quarto notes?

Unified Protocoll Stack Overview

Layer	Communication Endpoint	Data Unit Name	What It Contains	Adds Header/Footer?	Can Split Data?	Typical Protocols
Application	Applications or processes (e.g., browser ↔︎ web server)	Message	App-level data (e.g. HTTP, SMTP)	No	Yes — application logic (e.g. file chunks)	HTTP, FTP, SMTP, DNS, TLS, SSH, POP, IMAP
Transport (TCP/UDP)	Sockets on end hosts (process ↔︎ process)	Segment	Message + TCP/UDP header	Yes — transport header	Yes — TCP segments long messages	TCP, UDP
Network (IP)	Hosts or end systems (host ↔︎ host, abstracting from processes)	Packet (or Datagram)	Segment + IP header	Yes — network header	Yes — IP may fragment large packets	IP (v4/v6), ICMP, IGMP
Data Link	Directly connected devices (e.g. Host ↔︎ Router)	Frame	Packet + MAC header + trailer (e.g. CRC)	Yes — frame header and trailer	No — one packet per frame	Ethernet, Wi-Fi (802.11), PPP, ARP
Physical	Physical interfaces (e.g., NICs, cables, radio) exchanging raw bits	Bits	Encoded electrical/optical/radio signals	N/A (not in software)	No — transmits one bit at a time	DSL, Optical Fiber, Ethernet Cable, 5G