Questions: Complex Baseband and In-Phase/Quadrature Representation

Question 1

A WiFi signal occupies a 20 MHz bandwidth centered at a 2.4 GHz carrier. What sampling rate is required to digitize it in passband, and approximately what rate suffices for complex baseband?

• A. Passband: ≥ 40 MHz; Complex baseband: ≥ 4.81 GHz — passband is always more efficient
• B. Passband: ≥ 4.81 GHz; Complex baseband: ≥ 20 MHz — a 240× reduction in required sampling rate
• C. Both require ≥ 2.4 GHz — the carrier frequency sets the minimum sampling rate regardless of bandwidth
• D. Passband: ≥ 20 MHz; Complex baseband: ≥ 40 MHz — complex sampling requires higher rate due to two channels

Answer: (B)

Nyquist requires sampling above twice the highest frequency present. For a passband signal centered at 2.4 GHz with 20 MHz bandwidth, the highest frequency is 2.41 GHz, requiring fs > 4.82 GHz. The complex baseband signal has bandwidth 20 MHz centered at zero, requiring only fs > 20 MHz. This ~240× reduction in sampling rate is why digital radios universally process signals in complex baseband — it makes ADC design feasible. The I and Q channels each run at 20 MHz, not 40 MHz, because together they represent a complex (two-dimensional) sample.

Question 2

In a QAM-64 constellation diagram, the dots representing received symbols are scattered around 64 ideal points. What do I and Q represent in this plot?

• A. I is the signal magnitude and Q is the signal phase at each symbol instant
• B. I and Q are the in-phase and quadrature Cartesian components of the complex baseband signal, sampled at each symbol time
• C. I is the real part of the carrier frequency and Q is the imaginary part of the carrier
• D. I is the signal amplitude after demodulation and Q is the noise floor measurement

Answer: (B)

A constellation diagram plots I(t) on the horizontal axis and Q(t) on the vertical axis, sampled at symbol times. I and Q are the Cartesian components of the complex baseband signal x̃(t) = I(t) + jQ(t) — orthogonal projections obtained by mixing the bandpass signal with cosine and sine at the carrier. They are NOT magnitude and phase. Magnitude is √(I² + Q²) and phase is arctan(Q/I) — derived quantities, not the axes themselves. The scatter of dots around ideal constellation points reveals noise, phase error, and other impairments directly in I-Q space.

Question 3

The I and Q components of a complex baseband signal are the magnitude and phase of the original bandpass signal, expressed in polar form.

• T. True
• F. False

Answer: False

This is the most common misconception about I-Q representation. I and Q are Cartesian components — orthogonal projections obtained by mixing with cosine and sine at the carrier frequency. Magnitude = √(I² + Q²) and phase = arctan(Q/I) are derived from I and Q, but I and Q themselves are not the polar coordinates. For example, a BPSK signal alternating between bits has I = ±1 and Q = 0; its magnitude is always 1. Confusing I/Q with magnitude/phase leads to errors in signal processing and modulation analysis.

Question 4

Processing a bandpass signal in complex baseband reduces the required sampling rate because the baseband representation contains only the information-bearing bandwidth, not the carrier frequency.

• T. True
• F. False

Answer: True

Exactly. The complex baseband signal x̃(t) = I(t) + jQ(t) has a bandwidth of B (the modulation bandwidth) centered at zero frequency. Nyquist sampling only needs to capture this bandwidth, requiring fs > B. The carrier frequency fc has been mathematically removed by the mixing and lowpass filtering operations. This is why every digital radio downconverts to baseband before analog-to-digital conversion — the ADC only needs to handle the information bandwidth, not the carrier + bandwidth.

Question 5

Why does complex baseband representation allow digital radios to use dramatically lower sampling rates than passband sampling, and what operations produce the I and Q components?

Model answer: Complex baseband strips the carrier away, leaving only the information-bearing modulation bandwidth B centered at zero. Passband sampling must satisfy Nyquist for the full carrier + bandwidth (fs > 2(fc + B/2)), while complex baseband only requires fs > B. The I and Q components are produced by multiplying the bandpass signal by cos(2πfct) and sin(2πfct) respectively, then lowpass filtering both products to remove the double-frequency mixing terms. The result is two real signals at baseband whose combination x̃(t) = I(t) + jQ(t) completely characterizes the original modulated signal.

This sampling efficiency is not abstract — it's the reason digital radios are physically realizable. A 5G signal at 28 GHz with 400 MHz bandwidth would require a 57+ GHz ADC for passband sampling; complex baseband reduces this to 400 MHz. The downconversion hardware (mixers, lowpass filters) does the mathematical work, and the ADC only needs to handle baseband. The 'complex' in complex baseband refers to the mathematical representation, not extra complexity — it actually simplifies the digital processing enormously.