Data Communication and Networking¶

5-mark answer and 10-mark extension

For 5 marks - write this¶

Switching selects a path through intermediate nodes so information can move from source to destination. In circuit switching, signaling first establishes an end-to-end path, resources remain reserved during transfer, and release clears the connection. After setup it gives ordered delivery, fixed bandwidth, and nearly constant delay, which suits traditional PSTN voice, but idle speech intervals still consume capacity.

In message switching, no circuit is reserved. Each node receives and stores the complete addressed message, then forwards it when the next link is available. Link use and routing are flexible, but storage and whole-message waiting produce high and message-length-dependent delay, so it is unsuitable for interactive voice.

In packet switching, the message is divided into bounded packets with headers. Nodes store and forward one packet at a time, statistically sharing links. A datagram network such as IP routes packets independently; a virtual-circuit network such as X.25 or Frame Relay first establishes a logical route and then carries packets/frames with a short VC identifier.

Add for a 10-mark variant¶

The circuit call flow is setup request -> address analysis and cross-connect at each switch -> connect/acknowledgment -> continuous transfer -> release and resource clearing. Setup can fail when no circuit is free. Once connected, payload needs little per-unit addressing overhead. A leased circuit is permanently set up; a dialed PSTN circuit is temporary.

For message switching, node \(i\) cannot forward until all \(L\) bits arrive; on a rate-\(R\) link even the serialization component is \(L/R\) per hop, in addition to queueing and propagation. Priority and alternate routing are possible, but one long message can occupy buffers and links. Packetization replaces this with packets of length \(P\), allowing pipeline operation across successive links. Each packet incurs header overhead, processing and queueing; congestion can cause loss, duplication, variable delay, or out-of-order delivery in a datagram service. Higher layers may add sequencing, retransmission and jitter buffering.

Example: a ten-minute circuit-switched call holds one channel even during silence; packet voice sends coded speech packets only as generated and shares capacity, but must control loss, jitter and delay. Virtual circuits provide route consistency and easier QoS/accounting, while datagrams are robust and flexible under route changes. Actual networks may combine methods, such as packet transport carrying an emulated circuit, so “packet” does not automatically mean best-effort or connectionless.

Practice target: 9 minutes for definitions and comparison or 18 minutes for flows, delay reasoning, both packet modes, example, and caveat.

Exam-ready answer

X.25 and Frame Relay are connection-oriented packet WAN technologies using switched or permanent virtual circuits, but they reflect different transmission conditions. X.25 was designed for noisy analog links. A DTE connects through DCE to packet-switching exchanges; X.21/V-series interfaces provide the physical layer, LAPB provides reliable link frames with acknowledgments, sequencing, flow/error control and retransmission, and the packet-layer protocol establishes SVCs or carries PVC traffic using logical-channel identifiers. Reliability at each hop gives high overhead and delay.

Frame Relay assumes reliable digital facilities. It mainly offers a lean data-link service: frames carry a locally significant DLCI that each switch maps to the next link. A service contract specifies CIR and burst parameters. FECN warns the forward receiver of congestion, BECN warns the reverse source, and DE marks traffic eligible for discard. Corrupt frames are discarded rather than recovered hop by hop; end systems provide recovery if the application needs it.

Frame Relay is therefore faster but transfers more responsibility to endpoints. DLCI field interpretation, CIR enforcement, and optional SVC support depend on the provider/profile; neither technology is a modern Internet routing protocol, and both have largely yielded to IP/MPLS and Ethernet services.

Practice target: 8 minutes; contrast design assumptions and stacks, then define DLCI, CIR, FECN, BECN, and DE from the diagram.

Exam-ready answer

The Open Systems Interconnection (OSI) model, standardized by ISO, is a seven-layer reference framework in which each layer offers services to the layer above and uses services below. It separates communication functions without requiring one particular protocol suite.

For transmission, an application creates data; transport divides it and adds a transport header, network adds logical source/destination information, data link adds a local-link header and trailer/FCS, and physical emits symbols or bits. Each intermediate device processes only the layers needed for its role. At the destination, decapsulation removes headers in reverse order and delivers the original data.

The model has several advantages. Standard interfaces promote multivendor interoperability; modularity permits Ethernet to be replaced by WiFi without redesigning an application; protocol development and upgrades can be localized to a layer; encapsulation separates addressing, routing, reliability and media concerns; and layered observation makes troubleshooting systematic, from physical signal and link framing through IP reachability and application service. It also provides common terminology for education, design and conformance.

Example: when a browser sends HTTPS traffic, application data is represented and protected, TCP supplies ports, ordered delivery and retransmission, IP routes packets across networks, Ethernet or WiFi frames each local hop, and the physical layer sends electrical, optical, or radio symbols. A router primarily removes the incoming Layer-2 frame, examines the Layer-3 destination, and creates a new Layer-2 frame for the next hop; it does not forward the original Ethernet frame unchanged end to end.

OSI is a conceptual decomposition, not proof that every real protocol fits exactly one layer. TCP/IP combines application/presentation/session and combines physical/data-link as network access; security, tunneling, MPLS and middleboxes can cross boundaries. The benefit is disciplined functional separation, not mandatory implementation as seven independent software modules.

Practice target: 18 minutes; draw the seven-layer stack, give function/example/PDU for each, explain encapsulation, and state four advantages plus one real-world caveat.

Exam-ready answer

The OSI model is an ISO seven-layer reference framework; the TCP/IP model is the practical architecture associated with the DARPA/Internet protocol suite. Their functional mapping is:

OSI was specified as a protocol-independent model with clearer service, interface, and protocol separation. It explicitly isolates presentation and session functions and allows conceptual connection-oriented or connectionless network services. TCP/IP evolved with deployed protocols: IP is a connectionless best-effort internetwork layer, while reliability may be provided end to end by TCP; UDP offers connectionless low-overhead transport. TCP/IP is often shown as four layers, although some texts split link into data-link and physical and call it a five-layer Internet model.

In both models, application data is encapsulated into a transport segment/datagram, an IP packet where TCP/IP is used, a link frame, and finally bits/symbols; the receiver reverses the process. OSI is valuable for standard vocabulary, modular design, and fault isolation, whereas TCP/IP dominates real Internet implementation and interoperability. For example, HTTP/TCP/IP/Ethernet spans TCP/IP's four layers but maps to OSI Layers 7, 4, 3, 2 and 1, with presentation/session duties handled inside libraries or applications. The mapping is functional rather than one-to-one, and protocols such as TLS, MPLS and tunnels can straddle nominal boundaries.

Practice target: 8 minutes; reproduce the mapping table, then compare origin, nature, layer count, service treatment, deployment, and one caveat.

Exam-ready answer

The physical layer (OSI Layer 1) activates, maintains and deactivates the physical connection and transfers an unstructured stream of bits across one link. It defines the medium and connector/pinout; electrical, optical or radio signal levels; line coding/modulation; bit timing and clock recovery; bit rate and symbol rate; transmission direction; topology; maximum reach; and physical interface procedures. It does not interpret MAC addresses, frames, IP packets, ports, or application data.

The transmit path accepts a Layer-2 frame as bits, maps bits to symbols or line-code states, shapes/modulates the waveform, and launches it into twisted pair, coaxial cable, optical fiber, or radio. The receiver detects the waveform, recovers timing, decides symbols and presents bits to Layer 2, where framing and FCS checking occur. If each symbol has \(M\) distinguishable states, its nominal information mapping is

before coding overhead, where \(R_b\) is bit rate and \(R_s\) is baud/symbol rate.

Examples include RS-232 voltage, connector and timing specifications; Ethernet copper/fiber PHYs; DSL modems; optical transceivers; and IEEE 802.11 radio PHYs. A repeater regenerates/amplifies Layer-1 signaling and a hub repeats received bits to other ports; neither learns frames like a Layer-2 switch nor routes IP like a Layer-3 router. For example, a fiber Ethernet PHY converts coded electrical data to optical pulses and back while the Ethernet MAC above supplies addresses and frame check sequence. Practical limits arise from attenuation, noise, dispersion, bandwidth, BER and synchronization, and exact rates, wavelengths, coding and connector details are standard-dependent.

Practice target: 7 minutes; define Layer 1, trace transmit/receive operation, list six specifications, give four technologies/devices, and state what it does not do.