Fourier Series: Exponential Form

While the sine-cosine form of the Fourier series is intuitive, the exponential Fourier series offers a far more compact and mathematically powerful representation. This form expresses a periodic signal as a sum of complex exponentials, unifying amplitude and phase into a single complex coefficient and dramatically simplifying operations like differentiation, integration, and modulation. It is the essential bridge between Fourier series for periodic signals and the Fourier transform for aperiodic ones, forming the analytical backbone of signal processing, communications, and vibration analysis.

From Sine-Cosine to Complex Exponential Form

Recall that the standard trigonometric Fourier series represents a periodic function $x(t)$ with period $T_0$ and fundamental frequency $f_0=1/T_0$ (${\omega }_0=2\pi f_0$) as:

$$x(t)=a_0+\sum _{n=1}^{\infty }[a_n\cos (n{\omega }_{0}t)+b_n\sin (n{\omega }_{0}t)]$$

The transition to exponential form leverages Euler's formula, which connects trigonometric and exponential functions: $e^{j\theta }=\cos \theta +j\sin \theta$, where $j$ is the imaginary unit. From this, we can express cosine and sine as sums of complex exponentials:

$$\cos (n{\omega }_{0}t)=\frac{e^{jn{\omega }_{0}t}+e^{-jn{\omega }_{0}t}}{2}\text{and}\sin (n{\omega }_{0}t)=\frac{e^{jn{\omega }_{0}t}-e^{-jn{\omega }_{0}t}}{2j}$$

Substituting these identities into the trigonometric series and carefully regrouping terms leads to a single, unified summation that runs over both positive and negative integers $n$:

$$x(t)=\sum _{n=-\infty }^{\infty }{c}_n{e}^{jn{\omega }_{0}t}$$

This is the exponential Fourier series. The summation index $n$ now represents all harmonic indices, including negative ones. The coefficients $c_n$ are, in general, complex numbers.

Derivation and Properties of the Complex Coefficients

The complex coefficients $c_n$ are calculated directly from the signal $x(t)$ using the following integral over one period:

$$c_n=\frac{1}{T_0}{\int }_{T_0}x(t)e^{-jn{\omega }_{0}t}dt$$

This formula is central. For $n=0$, the exponential becomes 1, and $c_0$ is simply the average value of the signal: $c_0=a_0$. For $n\ne 0$, each $c_n$ encapsulates the amplitude and phase information for the harmonic at frequency $n{\omega }_0$.

The complex coefficients have profound symmetry properties for real-valued signals $x(t)$. If $x(t)$ is real, then $c_{-n}$ is the complex conjugate of $c_n$: $c_{-n}=c_n^{\ast }$. This implies two key facts:

1. The magnitude spectrum is even: $|c_{-n}|=|c_n|$.
2. The phase spectrum is odd: $\angle c_{-n}=-\angle c_n$.

These properties confirm that the negative frequency components are not "new" information for a real signal; they are a mathematical consequence of using complex exponentials as the basis. The coefficient $c_n$ can be expressed in polar form as $c_n=|c_n|e^{j\angle c_n}$, where $|c_n|$ is the magnitude of the $n$-th harmonic and $\angle c_n$ is its phase shift.

Advantages and Mathematical Simplification

The exponential form's elegance becomes clear during mathematical manipulation. Consider differentiation. The derivative of a complex exponential is straightforward: $$\frac{d}{dt}{e}^{jn{\omega }_{0}t}=jn{\omega }_0{e}^{jn{\omega }_{0}t}$$ Therefore, if $x(t)=\sum c_n{e}^{jn{\omega }_{0}t}$, then its derivative is simply: $$\frac{dx(t)}{dt}=\sum _{n=-\infty }^{\infty }[jn{\omega }_0{c}_n]e^{jn{\omega }_{0}t}$$ The Fourier series of the derivative is obtained by just multiplying each coefficient $c_n$ by $jn{\omega }_0$. This is far simpler than differentiating a sum of sines and cosines. Similar simplicity applies to integration and convolution operations, which are fundamental to linear system analysis.

This compactness also streamlines the multiplication of series and the analysis of modulation schemes in communications. Multiplying two exponential series often reduces to a simple index shift (convolution in the frequency index), a task that is cumbersome in trigonometric form.

Connecting to the Fourier Transform

The exponential Fourier series provides a direct conceptual path to the Fourier Transform. Consider a periodic signal with period $T_0$. Its spectrum consists of discrete lines at frequencies $n{\omega }_0$, weighted by $c_n$.

Now, imagine letting the period $T_0$ approach infinity. A single period of the signal begins to look like a single, aperiodic pulse. In the limit, the fundamental frequency ${\omega }_0=2\pi /T_0$ becomes infinitesimally small ($\Delta \omega$), and the harmonic frequencies $n{\omega }_0$ become a continuous variable $\omega$. The discrete sum in the Fourier series morphs into an integral, and the formula for the coefficients $c_n$ transforms into the formula for the continuous Fourier Transform $X(\omega )$. This limiting argument shows that the Fourier transform is the natural extension of Fourier series for non-periodic signals, with both sharing the exponential kernel $e^{-j\omega t}$.

Worked Example: Shifted Square Wave

Let's solidify these concepts with an example. Consider a shifted square wave with period $T_0=2$, defined over one period as $x(t)=1$ for $0<t<1$ and $x(t)=-1$ for $1<t<2$. We want its exponential Fourier series coefficients $c_n$.

Step 1: Apply the coefficient formula. $$c_n=\frac{1}{2}{\int }_0^{2}x(t)e^{-jn\pi t}dt$$ (since ${\omega }_0=2\pi /2=\pi$).

Step 2: Split the integral over the two intervals. $$c_n=\frac{1}{2}\left[{\int }_0^1(1)e^{-jn\pi t}dt+{\int }_1^2(-1)e^{-jn\pi t}dt\right]$$

Step 3: Evaluate the integrals. $$c_n=\frac{1}{2}\left[\left(\frac{e^{-jn\pi t}}{-jn\pi }\right){|}_0^1-\left(\frac{e^{-jn\pi t}}{-jn\pi }\right){|}_1^2\right]=\frac{1}{-j2n\pi }\left[(e^{-jn\pi }-1)-(e^{-j2n\pi }-e^{-jn\pi })\right]$$

Step 4: Simplify using $e^{-j2n\pi }=1$. $$c_n=\frac{1}{-j2n\pi }(2e^{-jn\pi }-2)=\frac{j}{n\pi }(1-e^{-jn\pi })$$

Step 5: Handle the $n=0$ case separately via limit or direct integration. For $n=0$, $c_0=\frac{1}{2}{\int }_0^{2}x(t)dt=0$.

Step 6: Further simplify using $e^{-jn\pi }=(-1{)}^n$. For $n\ne 0$: $$c_n=\frac{j}{n\pi }(1-(-1{)}^n)$$ This results in $c_n=0$ for even $n$, and $c_n=\frac{j2}{n\pi }$ for odd $n$ (i.e., $n=\pm 1,\pm 3,\pm 5,...$).

The series is: $$x(t)=\sum _{\begin{array}{c}n=-\infty \\ n\text{ odd}\end{array}}^{\infty }\frac{j2}{n\pi }{e}^{jn\pi t}$$ Notice $c_{-n}=\frac{j2}{(-n)\pi }=-\frac{j2}{n\pi }=c_n^{\ast }$, confirming the conjugate symmetry for this real signal.

Common Pitfalls

1. Ignoring Negative Frequencies: A common mistake is to think the summation should only run from $n=0$ to $\infty$. The negative frequencies are essential for the math to work and represent real, rotating phasors. For a real signal, they always pair with their positive counterparts to yield a real-valued sinusoidal function.

1. Misinterpreting Phase from $c_n$: The phase of the $n$-th harmonic is $\angle c_n$. For example, in our square wave, for $n=1$, $c_1=j2/\pi =(2/\pi )e^{j\pi /2}$. Thus, the phase for the fundamental is $+\pi /2$ radians or +90°. This is correct, but remember this is the phase of the complex exponential component. The corresponding real sinusoid's phase depends on the combination of $c_n$ and $c_{-n}$.

1. Incorrect Coefficient Calculation for $n=0$: The formula $c_n=\frac{1}{T_0}\int x(t)e^{-jn{\omega }_{0}t}dt$ is perfectly valid for $n=0$ and gives the DC average. Always compute it directly or take the limit as $n\to 0$ if your general expression is indeterminate (like 0/0).

1. Forgetting Conjugate Symmetry Check: When deriving coefficients for a real-valued signal, a quick check that $c_{-n}=c_n^{\ast }$ can catch algebraic errors. If this symmetry doesn't hold, there's likely a mistake in your integration or simplification.

Summary

• The exponential Fourier series $x(t)={\sum }_{n=-\infty }^{\infty }{c}_n{e}^{jn{\omega }_{0}t}$ represents a periodic signal using complex exponential basis functions with both positive and negative harmonic indices $n$.
• The complex coefficients $c_n=\frac{1}{T_0}{\int }_{T_0}x(t)e^{-jn{\omega }_{0}t}dt$ encode both the magnitude and phase of each frequency component. For real signals, they exhibit conjugate symmetry: $c_{-n}=c_n^{\ast }$.
• This form is mathematically superior for operations like differentiation and integration, as these operations become simple coefficient multiplications in the frequency domain.
• It provides the direct conceptual link to the Fourier transform via a limiting argument where the period $T_0\to \infty$, transforming the discrete sum into a continuous integral.
• When working with this form, always account for the full range of $n$, interpret complex coefficients correctly, and use symmetry properties to verify your results.