Z-Transform Properties and Inverse Techniques

Core Idea

Z-transform properties (linearity, time shift, convolution, scaling) simplify analysis of discrete-time systems. Inverse Z-transform techniques include partial fractions, long division, and residue calculation to recover time-domain sequences from X(z).

Explainer

The Z-transform converts a discrete-time sequence x[n] into a rational function X(z) in the complex variable z. The power of this representation comes from its properties, which let you manipulate sequences algebraically in the Z-domain rather than working sample-by-sample in the time domain. You already know how to compute the Z-transform from the definition; now you'll learn to exploit its structure.

The most important properties are linearity and the time-shift property. Linearity means that if you scale or add sequences, their Z-transforms scale and add identically — this lets you decompose complicated signals into simpler pieces. The time-shift property states that delaying a sequence by k samples multiplies X(z) by z^-k. This is the Z-domain analog of the Laplace differentiation property, and it is why difference equations — which relate outputs to delayed inputs — transform into simple polynomial equations in z. The convolution property completes the toolkit: convolution of two sequences in time corresponds to multiplication of their Z-transforms. This turns the operation of filtering (a convolution) into a multiplication, which is far easier to analyze.

To go in the other direction — recovering x[n] from X(z) — you need the inverse Z-transform. The most practical technique is partial fraction expansion. You factor the denominator of X(z)/z into first-order terms, expand into partial fractions, and look up each term in a Z-transform table. For example, a term of the form A/(1 - a·z^-1) corresponds to the sequence A·aⁿ·u[n] when the region of convergence (ROC) is outside the pole at z = a. Getting the ROC right is critical: the same rational function X(z) can correspond to multiple different sequences depending on whether you select causal, anti-causal, or two-sided sequences — the ROC determines which interpretation is correct.

Long division provides an alternative when you need the first few terms of x[n] explicitly. Divide the numerator polynomial by the denominator polynomial (using decreasing powers of z for causal sequences, or increasing powers for anti-causal ones), and the coefficients of the quotient are x[0], x[1], x[2], and so on. This is less elegant than partial fractions but useful for checking answers or when the sequence has no closed form. The residue method generalizes partial fractions to higher-order poles by computing contour integral residues — it is the formal basis for all inverse Z-transform calculations, though you rarely need it explicitly for simple pole cases.

Together, these tools let you work entirely in the Z-domain: model a filter as a ratio of polynomials H(z), find its response to an input X(z) by computing Y(z) = H(z)·X(z), then invert to get y[n]. This pipeline — transform, multiply, invert — is the Z-transform workflow that underlies all IIR and FIR filter design and analysis you will encounter next.