Chapter 1: Computer Networks and the Internet

What are hosts/end systems?

• Billions of connected computing devices

• Run network apps at the Internet edge

How is data moved across the Internet?

• Packet switches: forward packets → routers, switches

• Communication links: fiber, copper, radio, satellite

• Bandwidth: transmission rate

What makes up a network?

• Devices + routers + links

• Managed by an organization

• Examples: home, mobile, enterprise, datacenter, content-provider networks

Internet = ______ of _______

network of networks

• Interconnected ISPs

• Home, mobile, enterprise, datacenter, content-provider networks

Internet protocols

• Control sending/receiving of messages

• Examples: HTTP, TCP, IP, WiFi, 4G, Ethernet

Internet standards

• RFC = Request for Comments

• IETF = Internet Engineering Task Force

What does the Internet provide as a service?

• Infrastructure: supports applications

  • Web, streaming, email, games, e-commerce, social media, etc.

• Programming interface: “hooks” for apps to send/receive data

• Service options: different Internet transport services → like a postal service

What do protocols define?

• Format of messages

• Order messages are sent/received

• Actions taken when messages are transmitted/received

What are the 3 main parts of Internet structure?

• Network edge: hosts → clients & servers

  • Servers often in datacenters

• Access networks: connect hosts to Internet

  • Wired / wireless links

• Network core: interconnected routers

  • “Network of networks”

How do end systems connect to the edge router?

Access networks

• Residential access networks

• Institutional access networks → school, company

• Mobile access networks → WiFi, 4G/5G

Key considerations:

• Transmission rate: bits/second

• Shared vs. dedicated access among users

Your device → access network → edge router → Internet core

How does cable-based Internet access work?

• Cable modem → connects home to cable network

• Splitter → separates channels/signals

• Cable headend → central provider facility

• Uses FDM (Frequency Division Multiplexing)

  • Different channels → different frequency bands

What is HFC?

• Hybrid Fiber-Coax: fiber + coaxial cable network

• Asymmetric: downstream > upstream

  • Downstream: 40 Mbps–1.2 Gbps

  • Upstream: 30–100 Mbps

• Fiber/coax connects homes → ISP router

• Homes share the access network to the cable headend

How does DSL provide Internet access?

• Uses existing telephone line → central office

• DSLAM: separates/routes Internet data from voice

  • Data → Internet

  • Voice → telephone network

• Different frequencies → voice + data simultaneously

• Dedicated line (not shared)

• Downstream: 24–52+ Mbps

• Upstream: 3.5–16 Mbps

What are the two types of wireless access networks?

• WLAN (WiFi):

  • ~100 ft range

  • 802.11b/g/n

  • 11, 54, 450 Mbps

  • Connects end system → access point/router

• Wide-area cellular:

  • 10s of km range

  • Provided by mobile/cellular operators

  • 10s of Mbps

  • 4G, 5G (6G coming)

• Both: shared wireless access networks

What are enterprise networks?

• Used by companies, universities, etc.

• Mix of wired + wireless technologies

• Connect switches + routers

• Ethernet (wired): 100 Mbps, 1 Gbps, 10 Gbps

• WiFi (wireless): 11, 54, 450 Mbps

How does a host send data?

• Application message → broken into packets

• Each packet = L bits

• Transmitted into access network at rate R

• R = transmission rate / link capacity / bandwidth

• Packet transmission delay: time to transmit packet onto link

  • Delay = L / R

What are physical links/media?

• Bit: propagates between transmitter ↔ receiver

• Physical link: medium between transmitter & receiver

• Guided media: signal travels through a solid medium

  • Copper, fiber, coax

• Unguided media: signal propagates freely

  • Radio

• Twisted Pair (TP):

  • 2 insulated copper wires

  • Cat 5: 100 Mbps, 1 Gbps Ethernet

  • Cat 6: 10 Gbps Ethernet

What are the key features of coaxial and fiber optic cables?

Coaxial cable

• 2 concentric copper conductors

• Bidirectional

• Broadband: multiple frequency channels

• 100s Mbps/channel

Fiber optic cable

• Glass fiber → light pulses = bits

• 10s–100s Gbps

• Point-to-point, high-speed transmission

• Low error rate

  • Repeaters spaced far apart

  • Immune to electromagnetic noise

What are the key features/types of wireless radio links?

Wireless radio

• Signal carried in electromagnetic spectrum

• No physical wire

• Broadcast + half-duplex → sender ↔ receiver, not simultaneous

  • Example: walkie-talkie

  • Broadcast = multiple receivers can hear
    Half-duplex = communication goes both directions, one at a tim

• Propagation affected by:

  • Reflection, Obstructions, Interference

Types:

• Terrestrial microwave: up to 45 Mbps/channel

• Wireless LAN / WiFi: up to 100s Mbps

• Wide-Area / Cellular (4G): ~10s Mbps

• Satellite: up to 45 Mbps/channel

  • ~270 ms end-to-end delay

  • Geosynchronous vs. low-earth-orbit

What is the network core & how does packet switching work?

• Network core: mesh of interconnected routers

• Packet switching:

  • Hosts break messages → packets

  • Routers forward packets hop-by-hop

  • Packets travel across links from source → destination

  • Each packet transmitted at full link capacity

What is store-and-forward packet switching?

• Transmission delay: time to push L-bit packet onto link

  • L/R seconds

• Store-and-forward: router must receive entire packet before forwarding to next link

• End-to-end delay (2 links):

  • 2L/R (assuming zero propagation delay)

• Example: L = 10 Kbits, R = 100 Mbps

  • L/R = 0.1 ms transmission delay per hop

What happens when packet arrival rate > link transmission rate?

• Queueing: packets wait in a buffer to be transmitted

• Packet loss: if router’s buffer fills up, arriving packets are dropped/lost

Key idea:
Arrival rate > transmission rate → queue → buffer fills → packet loss

What are the 2 key network-core functions?

• Forwarding = local action

  • Move arriving packet input link → appropriate output link

  • Uses destination address in packet header

• Routing = global action

  • Determines source → destination paths

  • Uses routing algorithms

Recall: Forwarding = “where next?” | Routing = “which path?”

What is circuit switching?

• End-to-end resources reserved for a call

• Dedicated resources: no sharing

  • → Guaranteed/circuit-like performance

• Each link divided into circuits

• Idle circuit = wasted capacity (cannot be shared)

• Common in traditional telephone networks

• No available circuit → call rejected/blocked

Recall: Circuit switching = reserved path + dedicated resources

How does circuit switching allocate resources?

FDM — Frequency Division Multiplexing

• Frequency divided into narrow bands

• Each call gets its own frequency band

• Transmits at max rate of that band

TDM — Time Division Multiplexing

• Time divided into slots

• Each call gets periodic time slot(s)

• Transmits at max rate of the wider frequency band during its slot(s)

Recall: FDM = divide frequency | TDM = divide time

Why can packet switching support more users than circuit switching?

Given:

• Link = 1 Gbps

• Each user = 100 Mbps when active

• User active 10% of the time

Circuit switching:

• 1 Gbps ÷ 100 Mbps = 10 users max

• Resources reserved even when users are inactive

Packet switching:

• Users share link dynamically

• 35 users can use the network

• Probability of >10 users active simultaneously < 0.0004 - P(k) = C(35,k)(0.1)^k(0.9)^(35−k)

Recall: Packet switching → statistical sharing → more users

What are the advantages/disadvantages of packet switching?

Advantages:

• Great for bursty data → data sent intermittently

• Resource sharing

• Simpler → no call setup

Disadvantages:

• Congestion possible → packet delay + loss from buffer overflow

• Requires protocols for reliable transfer + congestion control

• Circuit-like behavior requires bandwidth guarantees (traditionally for audio/video)

Recall: Packet = efficient + flexible, but congestion can occur

How can access ISPs connect to the Internet?

• Access ISP → global transit ISP

• Customer ISP pays/has an economic agreement with provider ISP

• Global transit ISP provides connectivity to the rest of the Internet

What is at the center of the Internet’s “network of networks”?

• Small number of highly connected large networks

• Tier-1 ISPs:

  • Large commercial ISPs

  • National/international coverage

  • Examples: Level 3, Sprint, AT&T, NTT

• Content provider networks:

  • Private networks connecting data centers → Internet

  • Often bypass tier-1 & regional ISPs

  • Examples: Google, Facebook

• IXP: Internet Exchange Point → connects networks for exchanging traffic

How do packet loss and delay occur?

• Packets queue in router buffers → wait for transmission

• Transmission delay: packet is being transmitted

• Queueing delay: packet waits in buffer

• If arrival rate > output link capacity:

  • Queue builds up

  • No free buffer → packet dropped (loss)

Recall: Queue → delay | Full buffer → loss

What are the 4 sources of packet delay?

• Processing (dproc): check errors + determine output link

  • Typically < 1 ms

• Queueing (dqueue): waiting for transmission

  • Depends on router congestion

• Transmission (dtrans): time to push packet onto link

• Propagation (dprop): time for signal to travel across link

Total:
dnodal = dproc + dqueue + dtrans + dprop

What are transmission and propagation delays?

• Transmission delay (dtrans): time to push packet onto link

  • L = packet length (bits)

  • R = transmission rate (bps)

  • dtrans = L/R

• Propagation delay (dprop): time for signal to travel across link

  • d = physical link length

  • s = propagation speed ≈ 2 × 10⁸ m/s

  • dprop = d/s

Key difference:
Transmission = getting packet onto the link
Propagation = packet traveling across the link

What does the car/toll booth analogy show?

• Car = bit

• Caravan = packet (10 cars)

• Toll booth service = transmission delay

  • 12 sec/car × 10 = 120 sec = 2 min

• Driving to next booth = propagation delay

  • 100 km ÷ 100 km/hr = 1 hr

• Total time until caravan is lined up at 2nd booth = 62 min

Key idea: Transmission = push onto highway | Propagation = travel across highway

How does La/R affect average queueing delay?

• R = link bandwidth (bps)

• L = packet length (bits)

• a = average packet arrival rate

La/R:

• ≈ 0 → queueing delay small

• → 1 → queueing delay large

• > 1 → arrivals exceed capacity → average delay infinite

Recall: La/R measures how heavily the link is being used.

What does Traceroute measure?

• Measures delay from source → each router along path → destination

• Sends 3 probe packets to each router

  • Uses TTL = i to reach router i

• Router i sends a reply back

• Sender measures round-trip time (RTT) = time between sending probe and receiving reply

Recall: Traceroute = route + delay at each hop

How does packet loss occur?

• Router buffer/queue has finite capacity

• If a packet arrives when the buffer is full → packet is dropped/lost

• Lost packet may be:

  • Retransmitted by previous node

  • Retransmitted by source

  • Not retransmitted

What is throughput?

• Throughput: rate at which bits are sent sender → receiver

  • Instantaneous: rate at a specific point in time

  • Average: rate over a longer period

• Rs: server → link capacity

• Rc: receiver-side link capacity

• Bottleneck: slower link limits the throughput

Recall: Throughput = bits/time actually delivered

What determines average end-to-end throughput?

• Bottleneck link: link with the lowest transmission rate

• If Rs < Rc → throughput = Rs

• If Rs > Rc → throughput = Rc

Formula:
Throughput = min(Rs, Rc)

Recall: Slowest link = bottleneck = end-to-end throughput

What happens when 10 connections share a bottleneck link?

• R: backbone bottleneck capacity

• 10 connections fairly share R

  • Each gets R/10

• Per-connection throughput:

min(Rc, Rs, R/10)

• Bottleneck = smallest capacity

• In practice, Rc or Rs is often the bottleneck

What is network security?

• Goal: protect networks from attacks

  • How attackers attack networks

  • How to defend against attacks

  • How to design attack-resistant architectures

• Internet wasn't originally designed for security

  • Original vision: mutually trusting users

  • Security designers are “playing catch-up”

• Security considerations exist at all layers

What are the main types/effects of malware?

• Virus: self-replicates by receiving + executing an infected object

  • Example: email attachment

• Worm: self-replicates by passively receiving an object that executes itself

• Spyware: records keystrokes/websites → uploads information

• Botnet: infected hosts controlled together

  • Used for spam or DDoS attacks

Recall: Virus = execute | Worm = self-execute | Spyware = collect | Botnet = controlled hosts

What is a Denial of Service (DoS) attack?

• Goal: make resources unavailable to legitimate users

• Overwhelm target with bogus traffic

• Steps:

  1. Select target

  2. Compromise hosts → create botnet

  3. Send packets to target from compromised hosts

Recall: Compromise hosts → flood target → deny legitimate access

What is packet sniffing/interception?

• Packet sniffing: reading/recording packets as they pass through a network

• Possible on broadcast/shared media

  • Shared Ethernet

  • Wireless

• Promiscuous network interface: captures all passing packets

  • May capture sensitive data, e.g. passwords

• Wireshark: packet-sniffing software

Recall: Shared media + promiscuous interface → packets can be captured

What is IP spoofing?

• IP spoofing: sending a packet with a false/faked source IP address

• Makes packet appear to come from another device

Recall: IP spoofing = fake source IP

Why are protocol layers/reference models used?

• Networks have many complex components:

  • Hosts, routers, links

  • Applications, protocols

  • Hardware + software

• Purpose: organize the complexity of networks

• Makes it easier to understand, design, and discuss network systems

• Layers: each layer implements a service with own internal actions

Recall: Layering = organize network complexity

What are the different benefits of layering?

• Structure: identifies components + their relationships

• Modularization: easier maintenance + updates

• Layer changes: implementation can change without affecting other layers

• Example: change in one layer’s procedure → rest of system unaffected

• Potential downside: layering can sometimes be harmful/restrictive

What are the 5 layers of the Internet protocol stack?

• Application: network applications → HTTP, SMTP, IMAP

• Transport: process-to-process data transfer → TCP, UDP

• Network: routes datagrams source → destination → IP, routing protocols

  • Forwarding + routing

• Link: data transfer between neighboring network elements → Ethernet, WiFi, PPP

• Physical: transmits bits over the medium

What are the key differences between OSI/ISO and the Internet stack?

• OSI/ISO = 7 layers

  • Application

  • Presentation

  • Session

  • Transport

  • Network

  • Link

  • Physical

Internet stack does NOT have:

• Presentation: data interpretation → encryption, compression, machine-specific conventions

• Session: synchronization, checkpointing, recovery

• These services, if needed, are handled by the application layer in the Internet stack.

Recall: OSI adds Presentation + Session

How does data move through the protocol stack?

Down = encapsulate/add headers | Up = decapsulate/remove headers

What information is contained in a TCP segment?

• Source + destination port # → identify applications

• Sequence number → counts bytes in data stream

• Acknowledgment number (ACK) → next expected byte

• Receive window → bytes receiver is willing to accept (flow control)

• Checksum → error detection

• Flags: connection/congestion management

  • SYN, FIN, RST

  • C, E → congestion notification

• Options → additional TCP features

• Application data → variable-length payload

• Header length → length of TCP header

What are the major milestones in Internet history?

• 1961–1969: Packet switching → ARPAnet

  • 1961: Kleinrock → packet-switching theory

  • 1967: ARPAnet conceived

  • 1969: first ARPAnet node

• 1970s: Internetworking

  • ALOHAnet, Ethernet

  • 1974: Cerf & Kahn → Internet architecture

  • 1979: ARPAnet → 200 nodes

• 1980s: Core protocols

  • TCP/IP (1983)

  • DNS (1983), SMTP (1982), FTP (1985)

  • TCP congestion control (1988)

  • ~100K hosts

• 1990s: Web + commercialization

  • ARPAnet decommissioned

  • Web, HTML, HTTP

  • Mosaic → Netscape

  • Web commercialization

• Late 1990s–2000s: Killer apps + security

  • IM, P2P

  • Network security becomes major concern

  • Gbps backbone links

• 2000s–2017: Modern Internet

  • ~18B devices

  • Smartphones, broadband, 4G/5G, WiFi

  • Social networks

  • Private provider networks

  • Cloud computing (AWS, Azure)