Inverse Z-Transform Methods

The z-transform is a powerful tool for analyzing and designing discrete-time systems, but its true utility is realized when you can move back from the z-domain to the time domain. Inverse z-transform methods are the mathematical bridge that allows you to recover the original discrete-time sequence, $x[n]$, from its transformed representation, $X(z)$. Mastering these techniques is essential for solving difference equations, determining system impulse responses, and predicting system behavior directly from transfer functions.

Defining the Inverse Process

Formally, the inverse z-transform is defined by a contour integral in the complex plane: $$x[n]=\frac{1}{2\pi j}{\oint }_{C}X(z)z^{n-1}dz$$ where $C$ is a counterclockwise closed contour within the Region of Convergence (ROC) of $X(z)$. While this integral definition is fundamental, it is often not the most practical for computation. Instead, you rely on three primary methods: partial fraction expansion, power series expansion, and the direct evaluation of the contour integral via residues. Your choice depends on the form of $X(z)$ and what you need to know about $x[n]$.

The Power of Partial Fraction Expansion

Partial fraction expansion is the most common and systematic hand-calculation method, especially when $X(z)$ is a rational function (a ratio of polynomials in $z$ or $z^{-1}$). The goal is to decompose a complex $X(z)$ into a sum of simpler terms whose inverse transforms are known from a table of z-transform pairs.

The standard approach works best when $X(z)$ is expressed as a rational function in $z^{-1}$ to align with common transform pairs like $a^{n}u[n]\leftrightarrow \frac{1}{1-a{z}^{-1}}$. You must first ensure the fraction is proper (the numerator's degree is less than or equal to the denominator's). If it's not, polynomial long division yields a finite-duration term plus a proper fraction.

For example, consider finding the inverse of $X(z)=\frac{1}{(1-0.5z^{-1})(1-0.25z^{-1})}$ for $|z|>0.5$.

1. Expand into partial fractions:

$$X(z)=\frac{A}{1-0.5z^{-1}}+\frac{B}{1-0.25z^{-1}}$$

1. Solve for coefficients: $A=2,B=-1$.
2. Apply the known pair ${\alpha }^{n}u[n]\leftrightarrow \frac{1}{1-\alpha z^{-1}}$, assuming a causal ROC:

$$x[n]=2(0.5{)}^{n}u[n]-(0.25{)}^{n}u[n]$$

This method's strength lies in its direct connection to tabulated pairs and its clear structure.

Generating Sequences with Power Series Expansion

The power series expansion method, also known as long division, directly produces the sequence $x[n]$ term-by-term by expanding $X(z)$ into a series in either $z$ or $z^{-1}$. The chosen direction of expansion is dictated by the ROC, which tells you if the sequence is causal, anti-causal, or two-sided.

• For a causal sequence (ROC: $|z|>\alpha$), you express $X(z)$ as a power series in $z^{-1}$: $X(z)={\sum }_{n=0}^{\infty }x[n]z^{-n}$. Performing long division to expand in ascending powers of $z^{-1}$ gives you $x[n]$ starting from $n=0$.
• For an anti-causal sequence (ROC: $|z|<\alpha$), you express $X(z)$ as a power series in $z$: $X(z)={\sum }_{n=-\infty }^{-1}x[n]z^{-n}$. Performing long division to expand in descending powers of $z$ gives you $x[n]$ for negative $n$.

This method is particularly useful when you need only the first few terms of $x[n]$ or when $X(z)$ is non-rational. However, it rarely yields a closed-form expression for $x[n]$.

Contour Integration and the Residue Theorem

The direct method of contour integration implements the defining integral using complex analysis. By applying the Cauchy's Residue Theorem, the integral simplifies to calculating the sum of residues of the integrand, $X(z)z^{n-1}$, at the poles enclosed by the contour $C$. $$x[n]=\sum [\text{Residues of }X(z)z^{n-1}\text{ at poles inside }C]$$

The contour $C$ must lie within the ROC. For a rational $X(z)$, you choose $C$ to enclose poles that correspond to the desired sequence component (e.g., interior poles for a causal part if $C$ is a large circle). This method is theoretically rigorous and can handle cases where other methods are cumbersome, but it is computationally more intensive and is less frequently used for basic rational functions.

The Critical Role of the Region of Convergence

The Region of Convergence (ROC) is not an afterthought; it is the final piece of information that uniquely determines $x[n]$. A given $X(z)$ expression can correspond to multiple different sequences, differentiated solely by its ROC. When performing the inverse transform, you must always pair your algebraic manipulation with the specified ROC.

• Causal Sequence: If the ROC is of the form $|z|>|\alpha |$ (the exterior of a circle), the corresponding time-domain sequence is right-sided and causal ($x[n]=0$ for $n<0$).
• Anti-causal Sequence: If the ROC is of the form $|z|<|\beta |$ (the interior of a circle), the sequence is left-sided and anti-causal ($x[n]=0$ for $n>0$).
• Two-sided Sequence: If the ROC is a ring $|\alpha |<|z|<|\beta |$, the sequence is two-sided, formed by a causal part from poles at smaller radii and an anti-causal part from poles at larger radii.

For example, $X(z)=\frac{1}{1-a{z}^{-1}}$ can correspond to $a^{n}u[n]$ if $|z|>|a|$, or to $-a^{n}u[-n-1]$ if $|z|<|a|$. The algebraic expression is identical; the ROC provides the essential context for inversion.

Common Pitfalls

1. Ignoring the ROC: The most critical error is performing partial fraction expansion correctly but then applying transform pairs without regard to the ROC. Always check whether the ROC is exterior or interior to the pole magnitude to decide if you use the causal or anti-causal form of the transform pair.
2. Incorrect Form for Expansion: When using partial fractions, ensure $X(z)$ is expressed in the correct form (typically $z^{-1}$) to match your table of transform pairs. Using a table for transforms in $z$ while your expansion is in $z^{-1}$ will lead to an incorrect answer.
3. Misapplying Long Division Direction: When using the power series method, performing long division in the wrong direction (e.g., expanding in $z^{-1}$ for an anti-causal ROC) will generate a series representing the wrong sequence. Let the ROC dictate whether you divide to get ascending powers of $z^{-1}$ or $z$.
4. Overlooking Pole Order: When applying the residue method for a contour integral, remember that the formula for calculating a residue changes if the pole is of higher order (multiple poles). Using the simple residue formula for a repeated pole will yield an incorrect result.

Summary

• The inverse z-transform recovers a discrete-time sequence $x[n]$ from its z-domain representation $X(z)$ and is uniquely defined only when paired with a Region of Convergence (ROC).
• Partial fraction expansion is the most practical hand-calculation method for rational $X(z)$, breaking it into simpler terms that correspond to known transform pairs.
• The power series expansion (long division) method generates the sequence values term-by-term and is useful for non-rational transforms or finding initial sequence values.
• Contour integration, implemented via the residue theorem, is the direct mathematical method based on the defining complex integral.
• The ROC is essential for determining whether the resulting $x[n]$ is causal (right-sided), anti-causal (left-sided), or two-sided, as the same $X(z)$ expression can correspond to different sequences.