Signum Signal

Introduction

The signum function (also written as sgn or sign function) is a deceptively simple signal that extracts the polarity of its argument — it tells you whether a value is positive, negative, or zero. While it might seem too basic to deserve a full discussion, the signum function plays a surprisingly important role in signal processing, particularly in understanding the Hilbert transform, analytic signals, hard limiting (clipping), zero-crossing detection, and the mathematical relationship between the unit step function and other basic signals.

For B.Tech students, the signum function serves as a bridge between the unit step function and more advanced concepts. Its Fourier transform introduces you to signals that exist only in a distributional sense, and its applications in communication systems (hard limiters, sign-based detectors) give practical relevance to this fundamental building block.

Mathematical Definition

Continuous-Time Signum Function

The signum function is defined as:

$$\text{sgn}(t) = \begin{cases} +1 & t > 0 \\ 0 & t = 0 \\ -1 & t < 0 \end{cases}$$

Alternatively, it can be expressed as:

$$\text{sgn}(t) = \frac{t}{|t|} \quad \text{for } t \neq 0$$

Or in terms of the unit step function:

$$\text{sgn}(t) = 2u(t) - 1$$

This last relationship is crucial — it directly connects the signum to the unit step. Conversely:

$$u(t) = \frac{1}{2}[1 + \text{sgn}(t)]$$

Visual Description

The signum function looks like a switch: it's $-1$ for all negative time, jumps instantaneously to $+1$ at $t = 0$, and stays at $+1$ for all positive time. It has a discontinuity of size 2 at the origin (jumping from $-1$ to $+1$).

Properties of the Signum Function

Odd Symmetry

The signum function is an odd function:

$$\text{sgn}(-t) = -\text{sgn}(t)$$

This makes intuitive sense — reversing time flips the sign. The step function $u(t)$ is neither even nor odd, but by subtracting its mean value (1/2), we get the odd sgn function (scaled by 1/2).

Derivative

The derivative of the signum function is:

$$\frac{d}{dt}\text{sgn}(t) = 2\delta(t)$$

This follows from $\text{sgn}(t) = 2u(t) - 1$ and $du/dt = \delta(t)$. The derivative is zero everywhere except at the origin where the jump of magnitude 2 creates an impulse of strength 2.

Multiplication Property

For any signal $x(t)$:

$$x(t) \cdot \text{sgn}(t) = |x(t)| \cdot \text{sgn}(x(t)) \cdot \text{sgn}(t)$$

More usefully:

$$|t| = t \cdot \text{sgn}(t)$$

This means the absolute value function can be expressed as the product of $t$ and $\text{sgn}(t)$.

Signum Squared

$$[\text{sgn}(t)]^2 = 1 \quad \text{for } t \neq 0$$

Squaring the signum removes the sign information and gives a constant (except at the origin).

Fourier Transform of the Signum Function

The Fourier transform of $\text{sgn}(t)$ is an important result:

$$\mathcal{F}\{\text{sgn}(t)\} = \frac{2}{j\omega} = \frac{2}{j2\pi f}$$

Derivation

Since $\text{sgn}(t)$ is not absolutely integrable (it doesn't decay), we use a limiting approach. Define:

$$\text{sgn}_\epsilon(t) = \begin{cases} e^{-\epsilon t} & t > 0 \\ -e^{\epsilon t} & t < 0 \end{cases}$$

This is a "damped" version that approaches $\text{sgn}(t)$ as $\epsilon \to 0$. Its Fourier transform is:

$$\mathcal{F}\{\text{sgn}_\epsilon(t)\} = \frac{-1}{\epsilon + j\omega} + \frac{1}{-\epsilon + j\omega} = \frac{-2j\omega}{\epsilon^2 + \omega^2}$$

Taking the limit as $\epsilon \to 0$:

$$\mathcal{F}\{\text{sgn}(t)\} = \frac{-2j\omega}{\omega^2} = \frac{2}{j\omega}$$

Significance

The result $\text{sgn}(t) \leftrightarrow 2/j\omega$ shows that:

• The spectrum is purely imaginary (as expected for an odd function)
• The magnitude decays as $1/|\omega|$ — the signum has more low-frequency content
• There is a phase of $-90°$ for positive frequencies and $+90°$ for negative frequencies

Relationship to the Hilbert Transform

The Hilbert transform of a signal $x(t)$ is defined as:

$$\hat{x}(t) = x(t) * \frac{1}{\pi t}$$

In the frequency domain:

$$\hat{X}(f) = X(f) \cdot (-j \cdot \text{sgn}(f))$$

The factor $-j\text{sgn}(f)$ shifts all positive frequency components by $-90°$ and all negative frequency components by $+90°$. This is directly related to our signum function — the Hilbert transformer is essentially a $-90°$ phase shifter, and its frequency response is $-j\text{sgn}(f)$.

The Hilbert transform is fundamental in:

• Creating analytic signals (single-sideband representation)
• Envelope detection
• SSB modulation in communication systems
• Instantaneous frequency estimation

Applications of the Signum Function

Hard Limiter / Clipper

A hard limiter clips the signal to $\pm 1$ regardless of amplitude:

$$y(t) = \text{sgn}(x(t))$$

This is used in FM receivers (removing amplitude variations before frequency detection) and in digital communication receivers (1-bit quantization for sign-based detection).

Zero-Crossing Detection

The zero crossings of a signal correspond to the transitions in $\text{sgn}(x(t))$. The zero-crossing rate is related to the frequency content of the signal — a useful feature in speech processing and audio analysis.

Sign-Based Correlation

In some communication systems, the sign of the received signal is used for detection:

$$\hat{d} = \text{sgn}(r)$$

where $r$ is the received sample. This 1-bit detection is suboptimal but extremely simple to implement and surprisingly robust — it suffers only about 2 dB penalty compared to optimal detection.

Absolute Value Computation

Since $|x| = x \cdot \text{sgn}(x)$, the signum function is implicitly present in any system that computes the magnitude or absolute value of a signal — including envelope detectors, RMS meters, and full-wave rectifiers.

Worked Examples

Example 1: Express $x(t) = e^{-|t|}$ using the signum function.

Solution: $|t| = t \cdot \text{sgn}(t)$, so: $$x(t) = e^{-t \cdot \text{sgn}(t)}$$

Alternatively: $x(t) = e^{-t}u(t) + e^{t}u(-t)$

Example 2: Find the Fourier transform of $x(t) = \text{sgn}(t) \cdot e^{-2|t|}$.

Solution: $x(t) = e^{-2t}u(t) - e^{2t}u(-t) = e^{-2t}u(t) + (-e^{2t})u(-t)$

$$X(\omega) = \frac{1}{2+j\omega} - \frac{1}{2-j\omega} = \frac{(2-j\omega) - (2+j\omega)}{4+\omega^2} = \frac{-2j\omega}{4+\omega^2}$$

Example 3: Verify that $\frac{d}{dt}\text{sgn}(t) = 2\delta(t)$.

Solution: Using $\text{sgn}(t) = 2u(t) - 1$: $$\frac{d}{dt}[2u(t) - 1] = 2\frac{d}{dt}u(t) - 0 = 2\delta(t) \quad \checkmark$$

Discrete-Time Signum

The discrete-time signum is:

$$\text{sgn}[n] = \begin{cases} +1 & n > 0 \\ 0 & n = 0 \\ -1 & n < 0 \end{cases}$$

And the relationship holds: $\text{sgn}[n] = 2u[n] - 1 - \delta[n]$ (note the extra $\delta[n]$ term because $u[0] = 1$ but $\text{sgn}[0] = 0$, not 1).

Key Takeaways

• The signum function extracts signal polarity: $+1$ for positive, $-1$ for negative values
• Related to the unit step: $\text{sgn}(t) = 2u(t) - 1$ and $u(t) = [1 + \text{sgn}(t)]/2$
• Its Fourier transform is $2/(j\omega)$ — purely imaginary spectrum (odd function)
• The derivative of sgn is $2\delta(t)$ — an impulse of strength 2 at the jump
• Connected to the Hilbert transform through the frequency response $-j\text{sgn}(f)$
• Practical applications: hard limiters, zero-crossing detectors, sign-based detection, envelope extraction