COMP347 (Revision 10) | TrustOpen University
Upon completion of this expanded tutorial, students will be able to:
The Internet is the world’s most complex engineered system, interconnecting billions of devices across the globe. This tutorial provides a rigorous, foundational introduction to the principles and components of computer networking. We begin by defining the Internet from complementary viewpoints, then dissect its physical and logical structure. We explore the network edge (end systems and access technologies) and the network core (packet‑switched and circuit‑switched architectures). We also cover the physical media that carry data, the hierarchical organisation of ISPs, and the fundamental performance metrics—delay, loss, and throughput—that govern network behaviour. Finally, we examine protocol layering, the hourglass model, and the standards and governance frameworks that keep the Internet cohesive. This tutorial sets the stage for all subsequent study of networking protocols and applications.
From a hardware perspective, the Internet is a network of networks that interconnects billions of computing devices. These include end systems (hosts) such as PCs, servers, smartphones, and IoT devices; networking devices like routers and switches; and communication links (fibre, copper, wireless). Data is transmitted as bits over these links, using protocols to govern exchange.
From a services perspective, the Internet is an infrastructure that provides a platform for distributed applications: the Web, email, file sharing, streaming, VoIP, gaming, and cloud computing. The Internet exposes a programming interface (the socket API) that allows applications to send and receive data.
A protocol defines the format, order, and rules for communication between entities. It specifies syntax (message structure), semantics (interpretation of fields), and timing (when messages are sent). Examples: HTTP, TCP, IP, Ethernet.
End systems (hosts) run applications. They are classified as clients (requesting services) or servers (providing services). The client‑server model is the dominant architectural paradigm; servers are always‑on and have fixed IP addresses, while clients are intermittently connected.
| Technology | Medium | Typical Downstream | Upstream | Shared? |
|---|---|---|---|---|
| DSL (ADSL/VDSL) | Twisted‑pair copper | 1–200 Mbps | 1–20 Mbps | No |
| Cable (DOCSIS 3.1) | Coaxial cable | 1–10 Gbps | 200 Mbps–1 Gbps | Yes |
| FTTH (GPON) | Fibre optic | 2.5 Gbps | 1.25 Gbps | Yes (PON) |
| Ethernet (switched) | Copper/Fibre | 10 Mbps–100 Gbps | 10 Mbps–100 Gbps | No |
| Wi‑Fi (802.11ax) | Radio (unlicensed) | 1–10 Gbps | 1–10 Gbps | Yes |
| 5G (NR) | Radio (licensed) | 100 Mbps–10 Gbps | 100 Mbps–10 Gbps | Yes |
Guided media (fibre, copper, coax) confine signals to a physical path. Unguided media (radio, microwave, satellite) transmit through free space. Fibre offers the highest bandwidth and lowest attenuation, making it the backbone of modern networks.
Data is divided into packets, each transmitted independently. Key features:
Transmission delay: \( d_{trans} = L / R \) (L = packet length, R = link rate). For a path with N links, total transmission delay is \( N \cdot L / R \) (ignoring propagation and processing).
A dedicated path is established before communication. Resources are reserved via FDM (frequency division) or TDM (time division). Advantages: guaranteed performance, no queuing delays. Disadvantages: inefficiency during idle periods, connection setup overhead.
| Metric | Packet Switching | Circuit Switching |
|---|---|---|
| Resource allocation | On‑demand | Pre‑allocated |
| Delay | Variable | Fixed |
| Loss | Possible | None |
| Efficiency | High | Low (idle) |
| Setup phase | None | Required |
Modulation converts bits to analog signals (e.g., QAM, PSK). Multiplexing shares a medium: FDM (frequency), TDM (time), WDM (wavelength), and OFDM (orthogonal, used in Wi‑Fi and 4G/5G).
The Internet is composed of thousands of independent ISPs. They are organised into tiers:
Peering is direct traffic exchange, often free. Transit is paid access to the wider Internet. IXPs (Internet Exchange Points) are physical locations where ISPs interconnect to reduce cost and latency.
\( \rho = \frac{L \cdot a}{R} \). If \( \rho < 1 \), queuing delay is bounded. As \( \rho \to 1 \), delay grows without bound. When \( \rho > 1 \), queue length grows indefinitely and loss occurs.
The bottleneck link—the link with the smallest capacity on the path—determines the maximum achievable throughput. For a path with rates \( R_1, R_2, ..., R_n \), throughput = \( \min(R_i) \) (assuming no congestion).
Encapsulation: each layer adds its own header to the data from the layer above.
IP is the “waist” of the hourglass: many applications above, many link technologies below. This narrow interface enables interoperability and innovation at both ends.
Adds Presentation (formatting, encryption) and Session (dialog control). The Internet combines these into the Application layer (e.g., TLS provides encryption, cookies manage sessions).
Internet‑Draft → Proposed Standard → Draft Standard → Internet Standard. This ensures thorough review and interoperability.
The original design lacked built‑in security, leading to threats like IP spoofing, DDoS, and eavesdropping. Modern solutions (TLS, firewalls, DNSSEC) are applied at endpoints or in middleboxes.
Q1: Define the Internet from the “nuts and bolts” perspective.
A network of networks interconnecting billions of devices (hosts, routers, switches) via communication links and protocols.
Q2: What are the four components of nodal delay?
Processing delay, queuing delay, transmission delay, propagation delay.
Q3: How does traffic intensity (\( \rho \)) affect queuing delay?
As \( \rho \) approaches 1, queuing delay grows without bound; for \( \rho > 1 \), queue length increases indefinitely and loss occurs.
Q4: What is the difference between packet switching and circuit switching?
Packet switching uses store‑and‑forward and statistical multiplexing; circuit switching reserves dedicated resources before communication.
Q5: What is the role of an ISP?
An Internet Service Provider provides access to the Internet for end users and organisations; ISPs interconnect to form the global network.
Q6: What is an RFC and which organisation produces them?
RFCs (Requests for Comments) are documents that define Internet protocols; they are produced by the IETF.
Q7: Name two unguided media and one guided medium.
Unguided: radio, satellite; guided: fibre optic (also copper, coax).
Q8: What is the hourglass model of the Internet architecture?
IP is at the waist; many applications above and many link technologies below, enabling interoperability and innovation.
Q9: What is the purpose of the OSI presentation layer?
Data formatting, translation, encryption, and compression.
Q10: Why is security a challenge in the original Internet design?
It prioritised openness and interoperability over security, lacking built‑in authentication, encryption, or source validation.
Q11: What is the difference between a Tier‑1 and a Tier‑3 ISP?
Tier‑1 ISPs form the global backbone and peer with each other; Tier‑3 ISPs provide local access to end users.
Q12: Define statistical multiplexing.
Dynamic allocation of bandwidth among users based on their actual traffic demand; leads to high utilisation.
Exercise 1 – Transmission and propagation delay
A 1,500‑byte packet is sent over a 1,000‑km fibre link at 10 Gbps. Calculate the transmission delay and the propagation delay (fibre speed ≈ 2×10⁸ m/s).
Packet size = 1500 × 8 = 12,000 bits. \( d_{trans} = 12000 / 10^{10} = 1.2 \times 10^{-6} \) s = 1.2 μs.
\( d_{prop} = 1,000,000 / (2 \times 10^8) = 0.005 \) s = 5 ms.
Exercise 2 – Traffic intensity
A router has a 100 Mbps link. Packets arrive at an average rate of 8000 packets/s, each 1,500 bytes. Calculate the traffic intensity \( \rho \).
\( L = 1500 \times 8 = 12,000 \) bits. \( \lambda = 8000 \) packets/s.
\( \rho = \frac{\lambda \cdot L}{R} = \frac{8000 \times 12000}{100 \times 10^6} = \frac{96,000,000}{100,000,000} = 0.96 \).
Exercise 3 – Packet switching vs circuit switching
Consider a link of 1 Mbps. Each user generates 100 kbps when active and is active 10% of the time. How many users can circuit switching support? How many users can packet switching support (assuming a very small probability of overload)?
Circuit switching: each user needs 100 kbps; max users = 1 Mbps / 100 kbps = 10.
Packet switching: with 100 users, expected active = 100 × 0.1 = 10. Probability of more than 20 active is tiny; so 100 users (or more) can be supported.
Exercise 4 – Internet structure
Explain how the hierarchical ISP structure reduces routing table sizes.
Tier‑1 ISPs only need to know routes to other Tier‑1s and their customers; Tier‑2s need routes to their upstream providers and peers. This aggregation reduces the total number of routes compared to a flat mesh where every ISP would need a route to every other.
Exercise 5 – Encapsulation
Describe the encapsulation process for an HTTP request from a browser to a web server.
HTTP message → TCP header added (segment) → IP header added (datagram) → Ethernet header and trailer added (frame) → bits transmitted. At the server, headers are stripped in reverse order.
Exercise 6 – Queuing theory (M/M/1)
Given an M/M/1 queue with arrival rate \( \lambda = 500 \) packets/s and service rate \( \mu = 1000 \) packets/s, compute the average queuing delay.
\( \rho = 500 / 1000 = 0.5 \).
Average queuing delay \( W = \frac{\rho}{\mu - \lambda} = \frac{0.5}{1000 - 500} = \frac{0.5}{500} = 0.001 \) s = 1 ms.
Homework 1 – End‑to‑end delay calculation
A packet travels through 5 links, each with rate 100 Mbps and length 100 km (fibre). Processing delay per router is 0.1 ms. There are 4 routers. Calculate the total delay (excluding queuing).
Transmission delay per link = L/R; propagation = 100,000 / (2×10⁸) = 0.5 ms. Processing = 0.1 ms per router. Sum over 5 links and 4 routers.
Homework 2 – P2P scalability
Why is P2P distribution time asymptotically constant while client‑server grows with N? Derive the expression for \( T_{p2p} \).
The total upload capacity increases with N, so the term \( N F / (U_s + \sum u_i) \) does not grow with N.
Homework 3 – IETF process
Research the IETF RFC process. What are the stages from Internet‑Draft to Internet Standard? Why is “rough consensus” important?
Stages: Internet‑Draft → Proposed Standard → Draft Standard → Internet Standard. Rough consensus ensures practical, working solutions without formal voting.
Homework 4 – Security analysis
Explain how IP spoofing can be used in a DDoS attack. How does ingress filtering (RFC 2827) help mitigate it?
IP spoofing falsifies source addresses to hide the attack origin. Ingress filtering drops packets with source addresses not belonging to the network from which they originated.
Homework 5 – CDN and caching
Describe how a Content Delivery Network (CDN) reduces latency and offloads origin servers. Include DNS‑based routing and anycast.
CDNs replicate content at edge servers. DNS‑based routing directs clients to the nearest edge server. Anycast allows multiple servers to share an IP, routing to the closest topological one.
This tutorial has laid a rigorous foundation for understanding computer networks and the Internet. Key takeaways:
These concepts are essential for understanding all subsequent networking topics. In Tutorial 2, we will apply these principles to network application architectures and process communication.