Shannon's theorem, compared to Nyquist's theorem, differs chiefly in that:

It puts an absolute ceiling on capacity for a noisy channel, via the signal-to-noise ratio

It applies only to noiseless channels

It shows capacity can always be increased without limit

Bandwidth, throughput and latency are best distinguished as:

Theoretical maximum capacity, actual achieved rate, and one-way travel time respectively

Three names for exactly the same quantity

Properties that never vary independently of each other

Time Division Multiplexing (TDM), compared to Frequency Division Multiplexing (FDM), works by:

Giving each signal the whole bandwidth but only for a short, repeating time slice

Splitting the medium's bandwidth into separate simultaneous frequency bands

Never allowing more than one signal ever

Requiring an analog-only medium

A single parity bit fails to detect an error when:

An even number of bits have been flipped

Exactly one bit has been flipped

No bits have been flipped at all

Any three bits have been flipped

Which of the following are true of digital signals compared to analog signals? (More than one option may be correct.)

A digital signal is generally more resistant to noise

Digital and analog signals are exactly the same thing

Broadband transmission, compared to baseband transmission, works by:

Modulating the signal onto one of several carrier frequencies, letting multiple signals share the medium simultaneously

Sending a digital signal directly, using the medium's entire bandwidth for one signal at a time

Requiring no carrier signal at all

Working only over an already-digital medium

Data Communication Fundamentals

NET's Unit 9 opens with Data Communication as its own sub-item — simplex and duplex modes, analog and digital signals, noiseless and noisy channel capacity, bandwidth, throughput and latency, modulation, transmission media and error handling. The course's Computer Networks chapters start one layer up, at framing and switching; this chapter covers the physical and signalling layer underneath them.

Communication Modes and Analog Versus Digital Signals

The three communication modes
Mode	Direction of data flow
Simplex	One direction only, always — a keyboard-to-computer link (the computer never sends keystrokes back).
Half-duplex	Both directions possible, but only ONE at a time — a walkie-talkie, where both parties cannot transmit simultaneously.
Full-duplex	Both directions SIMULTANEOUSLY — a telephone call, where both parties can speak and listen at the same time.

An analog signal varies CONTINUOUSLY over time, taking any value within a range (a sine wave being the canonical example); a digital signal takes only a FINITE number of distinct discrete levels (typically two, representing 0 and 1). Digital signals are far more resistant to noise — a small amount of noise on a digital signal can usually still be correctly interpreted as the nearest valid discrete level, whereas the same noise on an analog signal directly corrupts the continuous value being carried, with no 'nearest valid level' to snap back to.

Noiseless and Noisy Channel Capacity, Bandwidth, Throughput and Latency

Nyquist's theorem (noiseless channel): the maximum data rate = 2 × B × log₂(L), where B is the channel's bandwidth (in Hz) and L is the number of distinct signal levels used — a NOISELESS channel's capacity can, in principle, be increased WITHOUT LIMIT simply by using more signal levels L, since nothing in the formula bounds L.
Shannon's theorem (noisy channel): the maximum data rate (channel CAPACITY) = B × log₂(1 + SNR), where SNR is the signal-to-noise ratio — UNLIKE Nyquist's formula, this places an ABSOLUTE, unavoidable ceiling on capacity for any REAL (noisy) channel no matter how many signal levels are attempted, because increasing signal levels without also increasing power runs straight into the noise floor and the extra levels become indistinguishable from each other.
Bandwidth, throughput and latency are three distinct quantities: bandwidth is the channel's theoretical maximum capacity (a property of the medium); throughput is the ACTUAL data rate achieved in practice (always ≤ bandwidth, reduced by congestion, protocol overhead, etc.); latency (delay) is the TIME for one bit/packet to travel from sender to receiver, entirely separate from how much data can flow per second — a high-bandwidth, high-latency link (a satellite connection) and a low-bandwidth, low-latency link (a short cable) are both real, common configurations that a bandwidth-alone view would conflate.

Data Encoding, Modulation and Transmission

Digital-to-analog conversion (modulation) is needed to send digital data over an analog-only medium: AMPLITUDE modulation varies the carrier wave's amplitude to represent bits, FREQUENCY modulation varies its frequency, and PHASE modulation varies its phase — real systems typically combine several of these (e.g. QAM combines amplitude and phase) to pack more bits per symbol.
Baseband versus broadband transmission: baseband sends a digital signal DIRECTLY over the medium, using the medium's ENTIRE bandwidth for one signal at a time (typical of a wired LAN); broadband instead MODULATES the signal onto one of several carrier frequencies, allowing MULTIPLE signals to share the same medium simultaneously, each on its own frequency band — the same underlying idea as frequency-division multiplexing.
Multiplexing combines several signals to share one physical medium efficiently: FDM (Frequency Division) splits the medium's bandwidth into separate frequency bands, one per signal, all transmitted simultaneously; TDM (Time Division) instead gives each signal the whole bandwidth but only for a short, repeating TIME SLICE — the two are complementary axes (splitting frequency versus splitting time) rather than competing solutions to the identical problem.

Transmission Errors and Error-Handling Mechanisms

Transmission errors are either a SINGLE-BIT error (one bit flipped, rare in serial transmission but common when errors are independent) or a BURST error (a contiguous run of bits flipped, the common case in real serial links, since noise typically affects a span of time rather than one isolated instant). Error DETECTION (parity bits, checksums, CRC — cyclic redundancy check) lets the receiver notice that corruption occurred, without necessarily saying which bit; error CORRECTION (Hamming codes and similar) goes further, encoding enough redundancy that the receiver can identify and fix a limited number of flipped bits without needing retransmission at all — the trade-off is always more redundancy (and hence lower effective throughput) for a stronger correction/detection guarantee.

⚠️ A single parity bit detects but does not correct, and misses even-count errors

A single parity bit can detect an ODD number of bit-flips (it changes the overall parity, which the check catches) but is BLIND to an EVEN number of flips (two bit-flips restore the original parity, so the check passes despite two bits being wrong) — and even when it does detect an error, it has no way to say WHICH bit is wrong, so it cannot correct anything by itself. CRC catches a far larger and more structured class of errors (including all burst errors up to the length of the check itself) but is still fundamentally a DETECTION, not correction, mechanism — retransmission (or a separate correcting code) is still needed once CRC flags a problem.

Key takeaways

Simplex (one way always), half-duplex (both ways, one at a time) and full-duplex (both ways simultaneously) are the three communication modes.
Nyquist bounds a noiseless channel's capacity with no absolute ceiling as signal levels increase; Shannon's theorem puts an absolute ceiling on any real (noisy) channel via SNR.
Bandwidth (theoretical maximum), throughput (actual achieved rate) and latency (one-way travel time) are three distinct, independently-varying quantities.
FDM splits bandwidth across simultaneous signals; TDM splits time across signals sharing the whole bandwidth in turn — complementary, not competing.
A single parity bit detects odd-count errors but misses even-count ones and cannot correct anything; CRC detects far more but is still detection, not correction.

Practice questions (8)

Attempt each one before opening the answer. Every explanation names the tempting wrong option as well as the right one, because that is where marks are lost.

A telephone call, where both parties can speak and listen at the same time, is an example of:
1. Full-duplex communication
2. Simplex communication
3. Half-duplex communication
4. No communication at all
Show answer
Answer: A — Full-duplex communication
Simultaneous two-way flow is exactly the defining feature of full-duplex, unlike half-duplex, where only one party can transmit at any given moment.
Shannon's theorem, compared to Nyquist's theorem, differs chiefly in that:
1. It puts an absolute ceiling on capacity for a noisy channel, via the signal-to-noise ratio
2. It applies only to noiseless channels
3. It ignores bandwidth entirely
4. It shows capacity can always be increased without limit
Show answer
Answer: A — It puts an absolute ceiling on capacity for a noisy channel, via the signal-to-noise ratio
Nyquist's noiseless-channel formula has no upper bound on signal levels and hence no absolute ceiling; Shannon's formula for a real, noisy channel does impose one via the SNR term, since more signal levels eventually become indistinguishable in the presence of noise.
Bandwidth, throughput and latency are best distinguished as:
1. Theoretical maximum capacity, actual achieved rate, and one-way travel time respectively
2. Three names for exactly the same quantity
3. All measured in seconds only
4. Properties that never vary independently of each other
Show answer
Answer: A — Theoretical maximum capacity, actual achieved rate, and one-way travel time respectively
A high-bandwidth, high-latency link (satellite) and a low-bandwidth, low-latency link (a short cable) both being real, common configurations is exactly why the three must be kept conceptually separate.
Time Division Multiplexing (TDM), compared to Frequency Division Multiplexing (FDM), works by:
1. Giving each signal the whole bandwidth but only for a short, repeating time slice
2. Splitting the medium's bandwidth into separate simultaneous frequency bands
3. Never allowing more than one signal ever
4. Requiring an analog-only medium
Show answer
Answer: A — Giving each signal the whole bandwidth but only for a short, repeating time slice
TDM splits along the time axis (full bandwidth, brief turns) while FDM splits along the frequency axis (partial bandwidth, all the time) — complementary strategies for sharing one medium among several signals.
A single parity bit fails to detect an error when:
1. An even number of bits have been flipped
2. Exactly one bit has been flipped
3. No bits have been flipped at all
4. Any three bits have been flipped
Show answer
Answer: A — An even number of bits have been flipped
An even number of flips restores the original overall parity, so the parity check passes even though the data is corrupted — an odd number of flips, by contrast, is reliably caught.
Which of the following are true of digital signals compared to analog signals? (More than one option may be correct.)
1. A digital signal takes only a finite number of distinct discrete levels
2. A digital signal is generally more resistant to noise
3. An analog signal varies continuously over time
4. Digital and analog signals are exactly the same thing
Show answer
Answer: A — A digital signal takes only a finite number of distinct discrete levels; B — A digital signal is generally more resistant to noise; C — An analog signal varies continuously over time
Digital signals' finite discrete levels give them a 'nearest valid level' to snap back to after small noise, which is exactly what makes them more noise-resistant than a continuously-varying analog signal.
Name the error-detection technique (an acronym) that detects a far larger and more structured class of errors, including all burst errors up to the length of the check itself, but still only detects rather than corrects.
Numerical answer — type the value.
Show answer
Answer: CRC
CRC (Cyclic Redundancy Check) is the standard detection mechanism used in most link-layer protocols; despite its strength, it remains a detection rather than correction technique, so retransmission or a separate correcting code is still needed once it flags a problem.
Broadband transmission, compared to baseband transmission, works by:
1. Modulating the signal onto one of several carrier frequencies, letting multiple signals share the medium simultaneously
2. Sending a digital signal directly, using the medium's entire bandwidth for one signal at a time
3. Requiring no carrier signal at all
4. Working only over an already-digital medium
Show answer
Answer: A — Modulating the signal onto one of several carrier frequencies, letting multiple signals share the medium simultaneously
Broadband's defining feature is modulating onto a carrier frequency so several signals can coexist on separate bands of the same medium — baseband instead sends the raw digital signal directly, using the whole bandwidth for one signal at a time.