2.2 Complex Exponential Fourier Series

Complex Exponential Fourier Series

The complex exponential Fourier series represents any periodic signal as a weighted sum of complex exponentials at harmonically related frequencies. This representation is central to signal processing because it moves analysis from the time domain into the frequency domain, where operations like filtering, modulation, and system response become far more tractable.

Representation of Periodic Signals

A periodic signal $x(t)$ with period $T$ can be expressed as:

$x(t) = \sum_{k=-\infty}^{\infty} X[k] \, e^{j 2\pi k t / T}$

Each term in this sum is a complex exponential spinning at a frequency that's an integer multiple of the fundamental frequency $\omega_0 = 2\pi / T$. The integer $k$ is called the harmonic index.

The complex Fourier coefficients $X[k]$ carry two pieces of information about each frequency component:

• Magnitude $|X[k]|$: the strength (amplitude) of the $k$-th harmonic
• Phase $\angle X[k]$: the relative phase shift of that harmonic

The fundamental frequency $\omega_0$ is the lowest nonzero frequency in the series. All other components sit at harmonic frequencies $k\omega_0$, so the second harmonic is at $2\omega_0$, the third at $3\omega_0$, and so on. The $k = 0$ term is simply the DC (average) value of the signal.

Together, the full set of coefficients $\{X[k]\}$ completely characterizes the periodic signal in the frequency domain. If you know all the coefficients, you can perfectly reconstruct $x(t)$.

Properties and Advantages

The complex exponential form is preferred over other representations for several practical reasons:

• Compact notation. A single summation handles all frequency components, including both positive and negative frequencies.
• Mathematical convenience. Differentiating or integrating $e^{j\omega t}$ just multiplies by a constant, which makes system analysis much cleaner.
• Frequency-domain manipulation. Modifying individual coefficients lets you selectively filter or enhance specific harmonics.

These properties make the complex exponential Fourier series a workhorse across electrical engineering (power systems, communications), mechanical engineering (vibration analysis), and digital signal processing (filtering, spectrum analysis).

Trigonometric vs. Complex Exponential Series

Relationship Between the Two Forms

The trigonometric Fourier series writes a periodic signal as a sum of cosines and sines with real-valued coefficients $a_k$ and $b_k$. The complex exponential form uses complex-valued coefficients $X[k]$ instead. The bridge between them is Euler's formula:

$e^{j\theta} = \cos(\theta) + j\sin(\theta)$

This lets you convert directly between the two representations. The coefficient relationships are:

• For $k > 0$: $X[k] = \frac{a_k - j\,b_k}{2}$
• For $k > 0$: $X[-k] = \frac{a_k + j\,b_k}{2}$
• For $k = 0$: $X[0] = a_0$

Notice that positive and negative frequency coefficients are conjugates of each other (for real-valued signals). This is a direct consequence of the fact that cosines and sines can each be split into two complex exponentials spinning in opposite directions.

Why Use the Complex Exponential Form?

• It treats positive and negative frequencies in a unified way, rather than requiring separate cosine and sine terms.
• The coefficients $X[k]$ directly encode both amplitude and phase, whereas the trigonometric form splits that information across $a_k$ and $b_k$.
• Mathematical operations (convolution, differentiation, system transfer functions) are simpler with complex exponentials.

The trigonometric form can still be more intuitive when you're working with real-valued signals and want to visualize individual cosine/sine components. But for most analytical and computational work, the complex exponential form is standard.

Fourier Coefficients Calculation

Analysis Equation: Continuous-Time

To find the coefficients from a known signal, you use the analysis equation:

$X[k] = \frac{1}{T} \int_{T} x(t) \, e^{-j 2\pi k t / T} \, dt$

The integral is taken over any single complete period of length $T$.

Step-by-step process:

1. Choose a convenient integration interval (e.g., $0$ to $T$, or $-T/2$ to $T/2$).
2. Multiply $x(t)$ by the complex exponential $e^{-j 2\pi k t / T}$ for the specific harmonic index $k$ you want.
3. Integrate the product over one full period.
4. Divide by $T$ to normalize.
5. Repeat for each value of $k$ you need.

For simple waveforms (square waves, sawtooth waves, etc.), you can often evaluate this integral in closed form. For more complex signals, numerical integration is necessary.

Analysis Equation: Discrete-Time

For discrete-time periodic sequences with period $N$, the integral becomes a finite sum:

$X[k] = \frac{1}{N} \sum_{n=0}^{N-1} x[n] \, e^{-j 2\pi k n / N}$

This is the form used in digital signal processing. The summation can be computed efficiently using the Fast Fourier Transform (FFT), which reduces the computational cost from $O(N^2)$ to $O(N \log N)$.

Properties of the Coefficients

The coefficients $X[k]$ satisfy two important properties:

• Conjugate symmetry (for real-valued signals): $X[-k] = X^*[k]$, where $X^*[k]$ is the complex conjugate. This means the magnitude spectrum is even and the phase spectrum is odd, so you only need to compute coefficients for $k \geq 0$.
• Periodicity (discrete-time): $X[k + N] = X[k]$. The coefficients repeat every $N$ indices, so there are only $N$ unique coefficients.

Both properties can be exploited to cut computation time roughly in half.

Signal Spectrum Interpretation

Magnitude and Phase Spectrum

The spectrum of a periodic signal shows how its energy is distributed across frequencies. From the Fourier coefficients, you get two complementary plots:

• Magnitude spectrum $|X[k]|$ vs. $k$: Shows the amplitude of each harmonic. Peaks indicate the dominant frequency components. This is often plotted on a logarithmic (decibel) scale to handle the wide dynamic range typical of real signals.
• Phase spectrum $\angle X[k]$ vs. $k$: Shows the phase shift of each harmonic. This captures timing relationships between components. Two signals can have identical magnitude spectra but sound or behave very differently due to phase differences.

Reading and Interpreting Spectra

The horizontal axis represents harmonic index $k$ (or equivalently, frequency $k\omega_0$). Here's what to look for:

• Harmonic spacing reflects the signal's periodicity. All components sit at integer multiples of $\omega_0$.
• Even/odd symmetry in the signal shows up in which harmonics are present. A signal with half-wave symmetry, for example, contains only odd harmonics.
• Bandwidth tells you about the signal's complexity. A signal with energy concentrated in just a few harmonics has narrow bandwidth (smooth waveform), while one with many significant harmonics has wide bandwidth (sharp features like edges or discontinuities).
• Spectral rolloff rate (how quickly $|X[k]|$ decreases with $k$) reveals smoothness. A square wave's coefficients decay as $1/k$, while a smoother triangle wave's decay as $1/k^2$.

Applications of Spectral Analysis

Spectral analysis using the Fourier series appears across many domains:

• Audio/speech processing: Identifying pitch (fundamental frequency) and timbre (harmonic structure)
• Vibration analysis: Detecting mechanical faults by spotting unexpected harmonics in machine vibration data
• Telecommunications: Designing filters, multiplexing signals, and characterizing channel bandwidth
• Radar/sonar: Classifying targets based on their Doppler frequency signatures

In each case, the spectrum transforms a complicated time-domain waveform into a clear picture of its frequency content, making patterns visible that would be difficult to spot otherwise.