Telecommunications and Data Communication experimental

Telecommunications and Data Communication From the Chappe optical telegraph to modern links: signalling, modulation, multiplexing, error control and the layered stack. Signals and modulation. Signals and modulation. Explain how information rides a carrier.. Explain the difference between analog and digital signal transmission, and apply the Nyquist-Shannon sampling theorem to compute the minimum sampling rate for a stated signal bandwidth.. Analog and digital signals, Modulation, Spectra, Sampling Analog and digital signals A signal is a function of time that carries information. An analog signal, such as a voice waveform, varies continuously in amplitude over a continuous range of values. A digital signal takes on a finite set of discrete levels, typically two (a 1 and a 0), held for a symbol period. Analog signals are natural for sound and light, but they accumulate noise at every stage of transmission and amplification. Digital signals can be regenerated exactly at each hop, because a receiver only has to decide which of a small number of levels was sent, not reconstruct an exact continuous value. This is why long-haul and stored communication has moved to digital representations even when the source, like a voice, is fundamentally analog: the analog signal is sampled and quantized into digital form near the source. Modulation Modulation is the process of impressing information onto a carrier wave so it can travel efficiently over a given medium, sharing spectrum with other signals and matching the propagation properties of the channel. A sinusoidal carrier has three properties that can be varied: amplitude, frequency, and phase. Amplitude modulation (AM) varies the carrier's amplitude with the message; frequency modulation (FM) varies its frequency; phase modulation (PM) varies its phase. Digital versions of the same ideas are amplitude-shift keying (ASK), frequency-shift keying (FSK), and phase-shift keying (PSK), where the carrier parameter is shifted among a discrete set of values, each representing one or more bits. Combining amplitude and phase shifts, as in quadrature amplitude modulation (QAM), packs more bits into each symbol at the cost of requiring a cleaner channel to distinguish the closely spaced symbol points. Spectra Every signal has a spectrum: a description of how its energy is distributed across frequency, found via the Fourier transform. A pure sine wave has energy at a single frequency; a square wave or a sharp pulse has energy spread across many harmonics. Modulating a message onto a carrier shifts and shapes the message's baseband spectrum up around the carrier frequency, a process called frequency translation. The bandwidth of a signal is the width of the frequency band that carries most of its energy, and it is bandwidth, not raw frequency, that determines how much information a signal can carry and how much channel spectrum it consumes. Sampling To convert an analog signal into digital form, it must be sampled: measured at regular time instants. The Nyquist-Shannon sampling theorem states that a signal band-limited to a maximum frequency f_max can be perfectly reconstructed if sampled at a rate of at least 2·f_max, called the Nyquist rate. Sampling below this rate causes aliasing, where high-frequency content folds back and is misread as a lower frequency, permanently corrupting the signal. After sampling, each sample is quantized to a finite number of bits, introducing quantization noise, and the resulting bit stream can be transmitted, stored, or further coded. Digitizing voice: a worked history The Bell System's T1 carrier, standardized in 1962, is the canonical worked example of the sampling theorem in industrial practice: the human voice band was treated as band-limited to roughly 4 kHz, so the Nyquist rate of 8 kHz became the standard sampling rate for digital telephony, a choice still embedded in narrowband voice codecs today. The Compact Disc, introduced commercially in 1982, made the same calculation for music rather than speech: with human hearing extending to roughly 20 kHz, a 44.1 kHz sampling rate was chosen to clear the Nyquist rate with margin for the anti-aliasing filter's imperfect roll-off, a decision that still defines the standard sample rate most consumer audio production tools default to. Spectrum as a shared, regulated resource Radio spectrum is not an infinite resource to be modulated onto freely; it is allocated by national regulators and international treaty precisely because two transmitters using overlapping frequencies in the same area interfere destructively with each other. The 2012 US Federal Communications Commission decision blocking LightSquared's proposed terrestrial broadband network illustrates the stakes: engineering tests showed the planned signal, transmitted in spectrum adjacent to the Global Positioning System (GPS) band, would swamp the front end of ordinary GPS receivers even though LightSquared's own allocated band was nominally separate, a reminder that a modulation scheme's practical safety depends on real receiver behaviour at band edges, not just on the nominal spectrum allocation on paper. Related CCI capabilities Computer Architecture (Course): (https://www.cambridgecyberinternational.com/en/insights/academy/computer-architecture/). Optics Primer Series (Course): (https://www.cambridgecyberinternational.com/en/insights/academy/optics/). Maths Refresher Series, Finance (Course): (https://www.cambridgecyberinternational.com/en/insights/academy/maths-finance/). System Dynamics (Course): (https://www.cambridgecyberinternational.com/en/insights/academy/system-dynamics/). CCI Lab: Run it, build with it, read the thinking, reuse the data. (https://www.cambridgecyberinternational.com/en/insights/lab/) Capacity and noise. Capacity and noise. Reason about capacity limits.. Compute channel capacity from bandwidth and signal-to-noise ratio using the Shannon-Hartley theorem, and explain why SNR degrades with distance.. Bandwidth, Noise, Shannon capacity, SNR Bandwidth Bandwidth is the range of frequencies a channel passes without excessive attenuation, measured in hertz. It is a property of the medium and the equipment attached to it: a telephone twisted pair might pass roughly 3 kHz of voice band, a coaxial cable can pass hundreds of megahertz, and an optical fibre can pass terahertz. Bandwidth sets a hard ceiling on how fast a signal can change, which in turn limits how many distinguishable symbols per second (the baud rate) a channel can support. A wider band allows either faster symbol rates or more room for multiple simultaneous signals, but wider bands also let in more noise power, which is why bandwidth alone does not determine how much information can actually get through. Noise Every real channel adds unwanted disturbances to the signal: thermal noise from the random motion of electrons, interference from other transmitters, crosstalk from adjacent wires, and attenuation distortion. Noise is typically modelled as additive white Gaussian noise (AWGN), meaning it has equal power across all frequencies and its instantaneous amplitude follows a Gaussian distribution. Noise power tends to scale with bandwidth, since a wider passband admits noise from a wider frequency range, while signal power is set by the transmitter. The receiver's job is to distinguish the intended signal from this noise floor, and its ability to do so is what ultimately bounds reliable communication. Shannon capacity Claude Shannon proved that a channel of bandwidth B (Hz) with signal-to-noise ratio SNR has a maximum error-free information rate, the channel capacity C, given by C = B·log2(1 + SNR) bits per second. This is a hard theoretical ceiling: no coding scheme, however clever, can sustainably exceed it, but sufficiently good error-correcting codes can approach it arbitrarily closely. The formula shows that capacity grows only linearly with bandwidth but logarithmically with SNR, so doubling bandwidth is a much more effective way to add capacity than doubling transmit power. This single formula underlies the design of every modern modem, Wi-Fi radio, and fibre-optic link. SNR Signal-to-noise ratio (SNR) is the ratio of received signal power to noise power in the channel, usually expressed in decibels as SNR(dB) = 10·log10(S/N). A higher SNR means the receiver can reliably distinguish more distinct signal levels, which supports denser modulation schemes carrying more bits per symbol. SNR degrades with distance because signal power attenuates while noise power stays roughly constant, which is why long links need repeaters, amplifiers, or lower data rates. Engineers budget SNR margin above the minimum needed for a target bit error rate, because real channels fade and vary over time and a design with no margin fails intermittently. Approaching the Shannon limit Claude Shannon's 1948 paper establishing the channel capacity formula this module states did not merely describe an abstract bound; it set a concrete target that decades of engineering then chased. Telephone-line modems illustrate this directly: the theoretical capacity of an ordinary analog voice line under standard noise conditions caps out close to 56 kbit/s in one direction, and the International Telecommunication Union (ITU)'s 1998 V.90 standard was explicitly engineered to approach that ceiling as closely as the phone network's actual noise and quantization characteristics would allow, after which further gains required abandoning the analog local loop model entirely in favour of digital subscriber line and, eventually, fibre technologies with a fundamentally larger bandwidth-SNR product to work with. Link budgets in practice A link budget is the practical accounting exercise that turns this module's capacity formula into an engineering design: starting from transmit power, subtracting free-space or cable attenuation over the link distance, subtracting losses at connectors and passive components, and adding receiver antenna gain, the result is the received signal power, which combined with the receiver's noise floor yields the achievable SNR and, via the Shannon formula, the maximum reliable data rate the link can sustain. Long-haul fibre-optic systems deployed since the early 2000s space amplifiers roughly every 80 to 100 kilometers specifically because the link budget's attenuation term would otherwise drive SNR, and therefore capacity, toward zero well before the signal reached its destination. Related CCI capabilities Computer Architecture (Course): (https://www.cambridgecyberinternational.com/en/insights/academy/computer-architecture/). Optics Primer Series (Course): (https://www.cambridgecyberinternational.com/en/insights/academy/optics/). Maths Refresher Series, Finance (Course): (https://www.cambridgecyberinternational.com/en/insights/academy/maths-finance/). System Dynamics (Course): (https://www.cambridgecyberinternational.com/en/insights/academy/system-dynamics/). CCI Lab: Run it, build with it, read the thinking, reuse the data. (https://www.cambridgecyberinternational.com/en/insights/lab/) Multiplexing and error control. Multiplexing and error control. Share a medium and correct errors.. Distinguish frequency-division from time-division multiplexing, and evaluate which error-detection or error-correction code fits a stated channel error-rate and retransmission-cost profile.. TDM/FDM, Coding, Error detection and correction, ARQ TDM/FDM Multiplexing lets many signals share one physical medium. Frequency-division multiplexing (FDM) splits the available bandwidth into non-overlapping sub-bands, each carrying one signal continuously, separated by small guard bands to prevent adjacent-channel interference; classic radio and cable television use FDM. Time-division multiplexing (TDM) instead gives each signal the full bandwidth but only for a short, recurring time slot, cycling through all signals in a fixed frame; digital telephone trunks (T1/E1) use TDM. FDM suits analog channels and continuous signals, while TDM suits digital signals that are naturally organized into discrete bits and frames. Statistical TDM improves on fixed TDM by allocating slots only to sources that currently have data, improving efficiency when traffic is bursty. Coding Line coding maps bits to physical signal levels for transmission, balancing goals such as clock recovery, direct-current (DC) balance, and bandwidth efficiency; examples include non-return-to-zero (NRZ), Manchester, and 4B/5B encoding. Separately, channel coding (error control coding) deliberately adds redundant bits to a message so that errors introduced by the channel can be detected or corrected at the receiver without retransmission. The trade-off is bandwidth or time spent on redundancy versus resilience to noise; the right amount of coding depends on the channel's error rate and the cost of retransmission or failure. Error detection and correction Parity bits detect single-bit errors cheaply but cannot correct them. The checksum, used in the Internet Protocol (IP) and Transmission Control Protocol (TCP), sums data words to catch common errors economically. The cyclic redundancy check (CRC) treats the message as a polynomial and divides it by a fixed generator polynomial, catching all single-bit and burst errors up to the length of the generator with very high probability, and is used in Ethernet frames. Hamming codes go further, adding enough parity bits at calculated positions to not only detect but locate and correct single-bit errors, based on the Hamming distance between valid codewords: a minimum distance of 3 allows single-error correction, and a minimum distance of 4 allows single-error correction with double-error detection (SECDED). ARQ Automatic repeat request (ARQ) is a class of protocols that recovers from detected errors by retransmission rather than correction. In stop-and-wait ARQ, the sender transmits one frame and waits for an acknowledgment before sending the next, which is simple but wastes time on long links. Go-Back-N ARQ lets the sender transmit several frames before an acknowledgment, but on detecting an error it discards and retransmits everything from the failed frame onward. Selective-repeat ARQ improves on this by retransmitting only the specific frames that were lost or corrupted, at the cost of more complex buffering at the receiver. All ARQ schemes rely on an underlying error-detecting code, such as a CRC, to know when a retransmission is needed. Line coding trade-offs in practice Manchester encoding, used in classic 10 Mbit/s Ethernet, guarantees a transition in the middle of every bit period, which makes clock recovery trivial for the receiver but doubles the bandwidth needed compared to the raw bit rate, since each bit requires a full transition cycle to encode unambiguously. NRZ encoding avoids this bandwidth penalty by using the simple presence or absence of a transition, but a long run of identical bits produces no transitions at all, starving the receiver's clock-recovery circuit; 4B/5B encoding, used in Fast Ethernet, splits the difference by mapping every 4 data bits to a carefully chosen 5-bit pattern guaranteed to contain enough transitions, trading a fixed 25% bandwidth overhead for reliable clock recovery without Manchester's full doubling. Forward error correction versus ARQ Forward error correction, exemplified by Hamming codes, and automatic repeat request both recover from channel errors, but they make opposite bets about where to spend resources. Forward error correction spends bandwidth continuously, on every transmission, whether or not an error actually occurs, but never needs a round trip back to the sender, which matters enormously on links with long propagation delay, such as a satellite hop, where an ARQ round trip can cost hundreds of milliseconds. ARQ instead spends nothing extra when the channel is clean and only pays the retransmission cost when an error is actually detected, which is more efficient on low-latency, low-error links but becomes punishing exactly where forward error correction is most valuable: high-latency or high-error-rate channels where waiting for a retransmission request is expensive or where the error rate is high enough that repeated retransmissions themselves start to fail. Related CCI capabilities Computer Architecture (Course): (https://www.cambridgecyberinternational.com/en/insights/academy/computer-architecture/). Optics Primer Series (Course): (https://www.cambridgecyberinternational.com/en/insights/academy/optics/). Maths Refresher Series, Finance (Course): (https://www.cambridgecyberinternational.com/en/insights/academy/maths-finance/). System Dynamics (Course): (https://www.cambridgecyberinternational.com/en/insights/academy/system-dynamics/). CCI Lab: Run it, build with it, read the thinking, reuse the data. (https://www.cambridgecyberinternational.com/en/insights/lab/) The layered stack. The layered stack. Place functions in layers.. Explain encapsulation across the protocol stack, and construct a correct layer mapping for a given real-world protocol against the OSI or TCP/IP model.. Physical to application, Encapsulation, Why layers, Mapping to today Physical to application Network communication is organized into layers, each responsible for a narrow slice of the overall job and each relying on the layer below. The physical layer moves raw bits over a medium, dealing with voltages, light pulses, radio waves, connectors, and the modulation and sampling covered earlier in this course. The data link layer organizes bits into frames, handles addressing on a local link, and performs error detection, often using the coding techniques already introduced. The network layer routes packets across multiple links from source to destination, most visibly through Internet Protocol (IP) addressing. The transport layer provides end-to-end delivery guarantees between processes, exemplified by Transmission Control Protocol (TCP)'s reliable, ordered byte stream and User Datagram Protocol (UDP)'s minimal, unreliable datagram service. Above that, session, presentation, and application layers (often collapsed together in practice) handle dialogue management, data formatting such as encryption and compression, and the application-specific protocols like HTTP, Simple Mail Transfer Protocol (SMTP), and Domain Name System (DNS) that users and programs directly invoke. Encapsulation As data descends the stack for transmission, each layer wraps the data handed to it from above with its own header (and sometimes trailer), a process called encapsulation. An HTTP request becomes the payload of a TCP segment, which becomes the payload of an IP packet, which becomes the payload of an Ethernet frame, which becomes a stream of bits on the wire. At the receiving end, decapsulation reverses this: each layer strips its own header, reads the information it needs, and passes the remaining payload up to the next layer. This nesting means each layer only needs to understand its own header format, not the internals of the layers above or below it. Why layers Layering is an engineering discipline for managing complexity. It allows a layer to change its internal implementation, such as swapping fibre for radio at the physical layer, without any change to the layers above, provided the interface between layers stays fixed. It also allows independent evolution: new application protocols can be invented without touching how IP routes packets, and new routing algorithms can be deployed without changing applications. This separation of concerns comes at a small efficiency cost, since each header adds overhead and each layer boundary adds processing, but the resulting modularity has proven essential for building and scaling a network as heterogeneous as the Internet. Mapping to today Two layered models are commonly referenced: the seven-layer Open Systems Interconnection (OSI) model (physical, data link, network, transport, session, presentation, application), designed as a comprehensive reference, and the four-layer TCP/IP model (link, internet, transport, application) that actually describes the protocols the Internet runs on. In practice, a Wi-Fi radio and its 802.11 framing sit at the physical and link layers, IP sits at the internet/network layer, TCP or UDP sits at the transport layer, and protocols like HTTPS or DNS sit at the application layer, with session and presentation concerns folded into libraries such as Transport Layer Security (TLS). Middleboxes and layer violations Real networks are not purely faithful to the clean layering this module describes: a network address translation device inspects and rewrites transport-layer port numbers while nominally sitting at the network layer, and a deep-packet-inspection firewall reads application-layer content to make network-layer forwarding decisions, both examples of what network engineers call a layer violation, where a device at one nominal layer depends on information from another. These violations are not simply errors; they solve real operational problems, such as IPv4 address scarcity or malicious-traffic filtering, but they come at the cost of the very modularity this module's layering discussion identifies as the point of layering in the first place: an application protocol change can silently break a middlebox that was quietly depending on the old format. Real protocol stacks in practice A concrete browsing session makes the abstraction tangible: a laptop's Wi-Fi radio and its 802.11 framing occupy the physical and data link layers, IP occupies the network layer and handles routing across the path to the server, TCP occupies the transport layer and guarantees the HTTP request and response arrive complete and in order, and TLS, layered above TCP but conceptually covering the session and presentation functions, encrypts the HTTP payload before HTTPS, the application-layer protocol, ever sees it. Tracing a single web request through this stack, header by header, is the fastest way to internalize why encapsulation and layering are not abstract diagrams but the literal sequence of operations a packet undergoes between a browser's address bar and a server's response. Related CCI capabilities Computer Architecture (Course): (https://www.cambridgecyberinternational.com/en/insights/academy/computer-architecture/). Optics Primer Series (Course): (https://www.cambridgecyberinternational.com/en/insights/academy/optics/). Maths Refresher Series, Finance (Course): (https://www.cambridgecyberinternational.com/en/insights/academy/maths-finance/). System Dynamics (Course): (https://www.cambridgecyberinternational.com/en/insights/academy/system-dynamics/). CCI Lab: Run it, build with it, read the thinking, reuse the data. (https://www.cambridgecyberinternational.com/en/insights/lab/)