CS 536 Park Introduction — Computer Networks

Page 1 — Overview of a computer network

• What is a computer network? A system that enables communication between distributed computing devices.
• Components of a computer network:
  • Host devices: PCs, servers, laptops, handhelds
  • Routers & switches: IP routers, Ethernet switches, WiFi routers
  • Links: wired, wireless, quantum
  • Protocols: IP, TCP, UDP, CSMA/CA, OSPF, BGP
  • Applications: DNS, HTTP, SMTP, SNMP, SSL
  • Humans and bots: spam, DoS, worm
• Distinction:
  • Hardware side: hosts, routers, links
  • Software side: protocols & applications
• Protocols as glue: they bind all components together across layers and domains

Page 2 — Protocol stack and protocol examples (low- to high-layer view)

• NIC (Network Interface Card) layer: firmware → e.g., Ethernet card, WLAN card; cellular air interface (CDMA or TDMA) → mainly ROM code
• Device driver: part of the Operating System → handles fast and slow interrupt processing
• ARP (Address Resolution Protocol) and RARP (Reverse ARP): OS → NICs; role is translation between address forms
• IP: OS → software glue of the global Internet
• Concept: protocols provide the handshakes and encodings that allow different hardware/software pieces to interoperate

Page 3 — Protocols and application examples (layer relationships)

• Routing protocols above IP: OSPF, RIP, BGP
  • Intra-domain routing: within organizations (intra-domain) using router OS (e.g., IOS)
  • Inter-domain routing: across the global Internet using BGP
• TCP and UDP: Transport layer
  • TCP: reliable file transfer for text, images, video (connection-oriented, error-checked)
  • UDP: best-effort transport suited for multimedia streaming (low latency)
• Application-layer protocols: DNS, HTTP, SMTP, SNMP, SSL
• Application-layer software: ssh, web browsers, PHP, P2P (BitTorrent), YouTube, Netflix, Facebook, Twitter, CDNs, bots

Page 4 — What layers are relevant (historical and current perspective)

• 1970s: focus on lower layers and hardware
• 1980s: both lower and higher layers become relevant
• 1990s: higher layers gain prominence
• Today: both lower and higher layers are important, plus hardware remains relevant
• Driving forces:
  • Ubiquitous wireless networks
  • Data centers and streaming services
• Boundary between telephony and data networks has disappeared
• Emergence of myriad devices: Internet of Things (IoT)

Page 5 — Illustrative examples of networks and spectra

• Digital TV and freed-up UHF spectra enabling sub-GHz spectra reuse: 300 ext{–}700~ ext{MHz} for data networking
• Short-distance services: RFID for electronic payments, tolls, inventory control, home security
• Wires being replaced or augmented by wireless
• CAN (Controller Area Network) bus and vehicle networks for specialized control
• Relevance for the future: self-driving vehicles connected to GPS, cellular networks, and IP networks

Page 6 — Computer networks enable communication (basic model)

• Simplest communication instance: two hosts A and B connected by a network N
• Information type: analog or digital
• Simplest case: a single bit
• Modern networks: content is digital (bits) but transmission is analog (electromagnetic waves)
• Concept: use analog signals to convey digital information

Page 7 — Forms of the network N (building blocks)

• Point-to-point link: dedicated, direct connection between A and B
  • Examples: single wire, line-of-sight antennae
• Broadcast link: A’s transmission can be heard by all devices in range
  • Multicast: logical addressing (not necessarily physical)
• Internetwork (network of networks): e.g., Purdue campus network, tier-1 AT&T intranet, global IP Internet

Page 8 — Capabilities required of A, B, and N

• Information abstraction:
  • Digital content representation: encode/decode information
  • Message structure: header, payload, trailer; app payload could be a file, streaming media, or protocol interaction
  • Endianness considerations (little/big endian) and data formatting
• Analog representation and transmission:
  • Transmitting digital content via analog signals over physical media (copper, fiber, wireless)
  • Digital transmission using square waves has specific roles

Page 9 — Information protection and reliability

• Information corruption: bit flips → Bit Error Rate (BER)
  • Example magnitudes: BER \,\approx\,10^{-9} for fiber optic cables; BER \,\approx\,10^{-6} for wireless
• Information loss: packet drops at routers/hosts due to congestion and buffering limits
  • Buffer overflow as a culprit; relates to resource provisioning and scheduling
• Security objectives:
  • Confidentiality: protect against eavesdropping
  • Authentication: protect against ID theft
  • Integrity: protect against tampering
  • Intrusion detection/prevention and DoS mitigation to protect infrastructure

Page 10 — Performance considerations

• Focus on software overhead and limitations
• Throughput: ext{bps}; high-speed hardware does not guarantee equivalent end-to-end throughput due to overhead and algorithms
  • Data compression via source coding can improve effective throughput
• Latency (delay): ext{msec}; physical distance and speed of light (SOL) impose fundamental limits
• Queuing and buffering at routers and OSs contribute to latency; important for video/audio streaming, voice, and interactive applications

Page 11 — Characteristics and capabilities of a network N

• Connectivity types:
  • Point-to-point link
  • Multi-access link
  • Internetwork
• Physical medium:
  • Wired
  • Wireless
• Location type:
  • Stationary
  • Mobile

Page 12 — Point-to-point link details

• Involves NICs at A and B
• Wired types: copper and fiber with various grades
• Wireless types: line-of-sight (LOS) antennas; examples include roof-top building-to-building links, infrared remotes, 60 GHz networks
• Concept: A and B do not require persistent names (in principle)

Page 13 — Multi-access link characteristics

• Topology examples:
  • Classical bus (old Ethernet): broadcast to every reachable device
  • Wireless media with omni-directional antennas (e.g., WiFi)
  • Wireless with semi-directional antennas (e.g., GPS satellites, cellular towers): signal casts a cone
• Addressing needs: local addressing to identify “From” and “To” within the LAN

Page 14 — Access control and LAN technologies

• Key issue on multi-access links: access control because the link is a shared resource
• How to share: simultaneous transmissions possible? Various LAN technologies/protocols address this
• Examples of LAN technologies and protocols:
  • WiFi, Bluetooth, RFID, Ethernet, OFDMA, CAN
• Note: much of LAN technology revolves around solving the access-sharing problem

Page 15 — Internetwork concept

• Internetwork (A B C D E) is a recursive definition:
  • Point-to-point and multi-access links are networks
  • A network of networks is an internetwork
• Ultimately, networks reduce to point-to-point and multi-access links; everything else is composition

Page 16 — Complications introduced by internetworks

• New names beyond LAN addresses:
  • IP addresses (IPv4 and IPv6) become dominant
• Protocol translation:
  • LANs may speak different link-layer technologies (Ethernet, WLAN) requiring translation at internetwork edges
• Path selection and routing:
  • How to forward packets from sender to receiver within and across organizations (e.g., Purdue internal routing, Purdue to its service providers)

Page 17 — Long-path performance and mobility management

• How fast to send on a long path:
  • Links have different speeds and traffic conditions influencing feedback control
• Congestion control: mechanisms to coordinate sender/receiver for efficient end-to-end performance
• Location management for mobility:
  • Handoffs as hosts move between networks (LAN handoff, IP handoff, Mobile IP)

Page 18 — LAN vs WAN distinctions and boundaries

• Technical distinction:
  • LAN: point-to-point and multi-access networks
  • WAN: internetworks that span large geographic areas
• Geographic proximity is not strictly defining WAN; there are counter-examples where WAN-like networks are local and not geographically close

Page 19 — Naming and addressing limitations; processes in hosts

• LAN addresses and IP addresses are often insufficient for end-to-end communication
• Entities are usually apps (processes) running inside host/router operating systems (e.g., Linux, Windows, IOS)
• IP addresses specify the NIC of a host/server/router, not the application process
• Multi-homed hosts may have multiple IP addresses (one per NIC)

Page 20 — Process identification and addressing mechanisms

• To identify a destination process, use 16-bit port numbers (supported by OSs)
• Typical addressing scheme: (IP address, port number) pair
• IP address translation: IP address must eventually be translated to a LAN address within the network
• Port numbers can be used to extend IP addressing in practice
• Sometimes port number is not required for certain communications

Page 21 — Organization-level identifiers and inter-domain routing

• Autonomous System Number (ASN) as organization-level identifier
  • Examples: Purdue ASN 17; Netflix AS40027; AT&T AS7018
• Role of ASNs in global routing policy and inter-domain routing
• IP addresses are associated with autonomous systems to support policy-based routing across domains
• Policy issues arise from separating administrative control and routing decisions across AS boundaries