Ferkans — Interactive Telecom Tutor

Why Baseband Matters

Wireless signals travel through the air at carrier frequencies of hundreds of MHz to tens of GHz, but all the information is encoded in a relatively narrow band around the carrier. The complex baseband (complex envelope) representation strips away the carrier, leaving only the information-bearing part. This simplification is so powerful that virtually all wireless communication analysis — from modulation design to channel estimation — is done at baseband.

Definition:
Hilbert Transform

The Hilbert transform of a real signal $x(t)$ is

$\hat{x}(t) = \frac{1}{\pi} \,\text{p.v.}\!\int_{-\infty}^{\infty} \frac{x(\tau)}{t - \tau}\,d\tau = x(t) * \frac{1}{\pi t}$

where p.v. denotes the Cauchy principal value.

In the frequency domain, the Hilbert transform is a $\pm 90°$ phase shift:

$\hat{X}(f) = -j\,\mathrm{sgn}(f)\,X(f) = \begin{cases} -jX(f) & f > 0 \\ jX(f) & f < 0 \end{cases}$

Key property: $x(t)$ and $\hat{x}(t)$ are orthogonal over $(-\infty, \infty)$ and have the same energy.

,

Definition:
Analytic Signal

The analytic signal (pre-envelope) associated with a real signal $x(t)$ is

$x_+(t) = x(t) + j\hat{x}(t)$

Its Fourier transform is one-sided:

$X_+(f) = \begin{cases} 2X(f) & f > 0 \\ X(0) & f = 0 \\ 0 & f < 0 \end{cases}$

The analytic signal eliminates negative frequencies, keeping only the positive-frequency content with doubled amplitude.

Definition:
Complex Envelope (Complex Baseband)

A real bandpass signal centred at carrier frequency $f_c$ can be written as

$x(t) = \mathrm{Re}\!\left[\tilde{x}(t)\,e^{j2\pi f_c t}\right]$

where $\tilde{x}(t)$ is the complex envelope (complex baseband signal). It is related to the analytic signal by

$\tilde{x}(t) = x_+(t)\,e^{-j2\pi f_c t}.$

Expanding into real and imaginary parts:

$\tilde{x}(t) = x_I(t) + j\,x_Q(t)$

where $x_I(t)$ is the in-phase component and $x_Q(t)$ is the quadrature component. The passband signal is then

$x(t) = x_I(t)\cos(2\pi f_c t) - x_Q(t)\sin(2\pi f_c t).$

The spectrum of $\tilde{x}(t)$ is centred at $f = 0$ (baseband) and has bandwidth $W \ll f_c$ . This is why digital processing can operate at rate $\sim 2W$ rather than $\sim 2f_c$ .

,

Example: AM Signal in Baseband

A conventional AM signal is $x(t) = [1 + m(t)]\cos(2\pi f_c t)$ where $m(t)$ is the message with $|m(t)| \le 1$ . Find the complex envelope $\tilde{x}(t)$ and the I/Q components.

Solution

Identify I/Q

Comparing $x(t) = x_I(t)\cos(2\pi f_c t) - x_Q(t)\sin(2\pi f_c t)$ with $x(t) = [1 + m(t)]\cos(2\pi f_c t)$ :

$x_I(t) = 1 + m(t), \qquad x_Q(t) = 0.$

Complex envelope

$\tilde{x}(t) = x_I(t) + jx_Q(t) = 1 + m(t).$ $The complex envelope is **real** — AM carries no quadrature component. This wastes half the available degrees of freedom, which is why QAM (using both I and Q) is far more spectrally efficient.$ \blacksquare$

Theorem: Passband–Baseband Equivalence

Let a bandpass signal $x(t) = \mathrm{Re}[\tilde{x}(t) e^{j2\pi f_c t}]$ pass through a bandpass LTI system with impulse response $h(t) = \mathrm{Re}[\tilde{h}(t) e^{j2\pi f_c t}]$ .

Then the output $y(t) = \mathrm{Re}[\tilde{y}(t) e^{j2\pi f_c t}]$ with

$\tilde{y}(t) = \frac{1}{2}\,\tilde{x}(t) * \tilde{h}(t).$

Convolution at passband is equivalent to (scaled) convolution at baseband. All filtering, channel effects, and detection can be analysed using the baseband equivalents.

The carrier oscillation $e^{j2\pi f_c t}$ factors out of the convolution, leaving only the slowly varying envelopes to be convolved. This is why communications engineers work exclusively at baseband.

Proof

Frequency-domain argument

In the frequency domain, $Y(f) = X(f) H(f)$ . For narrowband signals, $X(f)$ and $H(f)$ have support only near $\pm f_c$ . The positive-frequency part gives $\tilde{Y}(f) = \frac{1}{2}\tilde{X}(f)\tilde{H}(f)$ , which corresponds to $\tilde{y}(t) = \frac{1}{2}\tilde{x}(t) * \tilde{h}(t)$ in the time domain. $\blacksquare$

,

Passband vs. Baseband Representation

Visualise a bandpass signal and its complex baseband equivalent. Adjust the carrier frequency and modulation to see how the passband waveform relates to the I/Q components and the baseband spectrum.

Parameters

Carrier frequency

f_c

(Hz)10

Modulation type

Message frequency (Hz)1

Why This Matters: I/Q Modulator and Demodulator Architecture

The complex baseband representation directly maps to hardware. A quadrature modulator takes $x_I(t)$ and $x_Q(t)$ , multiplies them by $\cos(2\pi f_c t)$ and $-\sin(2\pi f_c t)$ respectively, and sums — producing the passband signal. The receiver reverses the process with a quadrature demodulator (multiplication by $\cos$ and $-\sin$ , followed by low-pass filtering).

Every modern transceiver (Wi-Fi, LTE, 5G NR) uses this I/Q architecture. Impairments like I/Q imbalance (gain and phase mismatch between the two branches) are a major practical concern and are corrected by digital signal processing at baseband.

Common Mistake: The Factor of 1/2 in Baseband Convolution

Mistake:

Writing $\tilde{y}(t) = \tilde{x}(t) * \tilde{h}(t)$ without the factor of $1/2$ .

Correction:

The correct relation is $\tilde{y}(t) = \frac{1}{2}\tilde{x}(t) * \tilde{h}(t)$ . The factor arises because $\mathrm{Re}[\cdot]$ extracts half the analytic signal. Some textbooks absorb this factor into the definition of $\tilde{h}$ ; be consistent with whichever convention you use.

Quick Check

A passband signal is $x(t) = 3\cos(2\pi f_c t) + 4\sin(2\pi f_c t)$ . What is the complex envelope $\tilde{x}(t)$ ?

$3 + j4$

$3 - j4$

$4 + j3$

$5 e^{j\arctan(4/3)}$

Correction:

3 - j4

Correct. Writing $x(t) = 3\cos(2\pi f_c t) - (-4)\sin(2\pi f_c t)$ , we identify $x_I = 3$ and $x_Q = -4$ , so $\tilde{x} = x_I + jx_Q = 3 - j4$ .

Quick Check

The spectrum of a real signal $x(t)$ satisfies $X(-f) = X^*(f)$ . What is the spectrum of its analytic signal $x_+(t)$ ?

$X(f)$ for all $f$

$2X(f)$ for $f > 0$ , $0$ for $f < 0$

$X(f) + X(-f)$

$|X(f)|^2$

Correction:

2X(f)

for

f > 0

,

0

for

f < 0

Correct. The analytic signal doubles the positive-frequency content and zeros out negative frequencies: $X_+(f) = 2X(f)$ for $f > 0$ , $X(0)$ for $f = 0$ , and $0$ for $f < 0$ .

Key Takeaway

All wireless analysis is done at baseband. The complex envelope $\tilde{x}(t) = x_I(t) + jx_Q(t)$ strips the carrier, reducing the signal bandwidth from $\sim f_c$ (GHz) to $\sim W$ (MHz). Digital processing operates at rate $\sim 2W$ , not $\sim 2f_c$ — a reduction of three orders of magnitude at typical cellular frequencies.

⚠️Engineering Note

I/Q Imbalance in Practical Transceivers

Real quadrature modulators and demodulators suffer from I/Q imbalance: a gain mismatch $\Delta g$ and phase mismatch $\Delta\phi$ between the in-phase and quadrature branches. The received baseband signal becomes

$\tilde{r}(t) = \alpha\,\tilde{x}(t) + \beta\,\tilde{x}^*(t)$

where $\alpha \approx 1$ and $\beta \propto \Delta g + j\Delta\phi$ . The conjugate term $\tilde{x}^*(t)$ creates image interference: a signal at frequency $+f$ leaks into $-f$ .

Typical imbalance values in commercial RF front-ends: gain $\Delta g \approx 0.5$ – $2$ %; phase $\Delta\phi \approx 1$ – $5°$ . The resulting image rejection ratio (IRR) is 25–40 dB — insufficient for high-order QAM (64-QAM requires IRR $> 35$ dB). Digital compensation algorithms estimate $(\alpha, \beta)$ from pilot symbols and pre-correct the signal.

Practical Constraints

•
Typical IRR without calibration: 25–40 dB (insufficient for 64-QAM and above)
•
Digital compensation can achieve IRR > 60 dB with proper pilot design
•
Wideband signals experience frequency-dependent I/Q imbalance, requiring per-subcarrier correction

Why This Matters: Hardware Implementation of I/Q Processing

The complex baseband framework maps directly to transceiver hardware design. Chapter 23 (RF Front-End and Hardware Impairments) covers the practical architecture of quadrature modulators and demodulators, including amplifier nonlinearity, phase noise, and the digital calibration techniques that compensate for the I/Q imbalance discussed in this section.

Complex Envelope

The baseband equivalent $\tilde{x}(t) = x_I(t) + jx_Q(t)$ of a bandpass signal. Contains all information; the carrier is removed.

Hilbert Transform

A $\pm 90°$ phase shift: $\hat{X}(f) = -j\,\mathrm{sgn}(f)\,X(f)$ . Used to construct the analytic signal.

In-Phase and Quadrature (I/Q) Components

The real part $x_I(t)$ (in-phase) and imaginary part $x_Q(t)$ (quadrature) of the complex envelope. Together they carry all the information of the bandpass signal.

Bandpass Signals and Complex Baseband Representation

Why Baseband Matters

Definition: Hilbert Transform

Definition: Analytic Signal

Definition: Complex Envelope (Complex Baseband)

Example: AM Signal in Baseband

Identify I/Q

Complex envelope

Theorem: Passband–Baseband Equivalence

Frequency-domain argument

Passband vs. Baseband Representation

Parameters

Why This Matters: I/Q Modulator and Demodulator Architecture

Common Mistake: The Factor of 1/2 in Baseband Convolution

Quick Check

Quick Check

Key Takeaway

I/Q Imbalance in Practical Transceivers

Why This Matters: Hardware Implementation of I/Q Processing

Complex Envelope

Hilbert Transform

In-Phase and Quadrature (I/Q) Components

Definition:
Hilbert Transform

Definition:
Analytic Signal

Definition:
Complex Envelope (Complex Baseband)