Skip to content
Tutorials Center
My learning

Topics

Lesson 10 of 36 ยท LTI Systems and Convolution

Impulse Response and LTI Systems

See why a linear time-invariant system is fully described by its impulse response, how to measure it, and how it decides causality and stability.

Linear time-invariant systems have a remarkable property: once you know the output for one special input, the unit impulse, you know the output for every input. That special output is the impulse response h[n] (or h(t)). This lesson explains why, and how the impulse response also tells you whether the system is causal and stable.

The idea in discrete time

Any discrete signal can be written as a sum of scaled, shifted impulses:

x[n] = sum over k of x[k] delta[n - k].

For x = [2, 3, 1] starting at n = 0 this means x[n] = 2 delta[n] + 3 delta[n-1] + 1 delta[n-2]: three impulses with heights 2, 3, 1.

Now feed this into an LTI system with impulse response h[n]:

  1. Time-invariance: delta[n - k] produces h[n - k], the same response delayed by k.
  2. Linearity: a scaled input x[k] delta[n - k] produces the scaled response x[k] h[n - k].
  3. Superposition: the sum of inputs gives the sum of outputs.

So y[n] = sum over k of x[k] h[n - k]. This sum is the convolution of x with h, written y = x * h, and it is the topic of the next lessons.

flowchart LR
    X["x[n] = sum of x[k] delta[n-k]"] --> H["LTI system h[n]"]
    H --> Y["y[n] = sum of x[k] h[n-k]"]

Decompose the input into impulses, respond to each with a shifted copy of h, and add the results.

Measuring an impulse response

To find h[n] of a black-box LTI system, feed it delta[n] and record the output. The step response s[n] (the response to u[n]) carries the same information: since u[n] is the running sum of delta[n], s[n] is the running sum of h[n], and h[n] = s[n] - s[n-1].

Python Measure h[n] of a black-box system

Press Run (or Ctrl + Enter) to see the result here.

Program input ยท what you'd type while the program runs

Type the answers before you press Run, one per line. Each time the program reads a line (with input()), it gets the next line from this box.

The impulse response is [1, 0.5, 0.25, 0, 0, ...]: three nonzero values, so this is a finite impulse response (FIR) system. A system whose impulse response never ends, like y[n] = 0.9 y[n-1] + x[n] with h[n] = 0.9^n u[n], is infinite impulse response (IIR).

Predicting any output from h

Python Output from the impulse response alone

Press Run (or Ctrl + Enter) to see the result here.

Program input ยท what you'd type while the program runs

Type the answers before you press Run, one per line. Each time the program reads a line (with input()), it gets the next line from this box.

What h tells you about the system

Causality. An LTI system is causal if and only if h[n] = 0 for n < 0. Intuition: the response to an impulse at n = 0 cannot start before the impulse arrives.

Memory. An LTI system is memoryless only if h[n] = K delta[n].

Stability. An LTI system is BIBO stable if and only if its impulse response is absolutely summable: sum of |h[n]| < infinity (continuous: integral of |h(t)| dt < infinity). The rule makes sense because the worst-case input flips signs to line up with h, producing output sum |h|.

Python Stability from the impulse response

Press Run (or Ctrl + Enter) to see the result here.

Program input ยท what you'd type while the program runs

Type the answers before you press Run, one per line. Each time the program reads a line (with input()), it gets the next line from this box.

For h[n] = a^n u[n] the sum is 1 / (1 - |a|) when |a| < 1, so the system is stable only for |a| < 1. That is the first appearance of a theme you will see again for poles in the Laplace and Z domains.

Continuous time

The continuous version replaces sums with integrals. Any signal is x(t) = integral of x(tau) delta(t - tau) d tau (the sifting property), so y(t) = integral of x(tau) h(t - tau) d tau. For example, an RC low-pass circuit has h(t) = (1/RC) e^(-t / RC) u(t): causal, and stable because the integral of |h| is 1.

Check your understanding

Quick check ยท 4 questions

of 4 answered You scored / 4 ยท perfect!

  1. 1. What does an LTI system's impulse response tell you?

    Because every input is a weighted sum of shifted impulses, linearity and time-invariance give the output for any input via convolution.
  2. 2. An LTI system is causal if and only if:

    A causal system cannot respond before the impulse arrives.
  3. 3. h[n] = 0.9^n u[n] is:

    The geometric series 1 + 0.9 + 0.81 + ... sums to 1 / (1 - 0.9) = 10, so h is absolutely summable and the system is stable. It is IIR and causal.
  4. 4. How do you get h[n] from the step response s[n]?

    The step is the running sum of the impulse, so the difference of the step response is the impulse response.

Summary

  • Any signal is a weighted sum of shifted impulses; LTI systems respond to each with a shifted, scaled copy of h.
  • The output is the convolution y = x * h.
  • Measure h with an impulse, or difference the step response.
  • Causal: h = 0 for negative time. Stable: h is absolutely summable (integrable).
  • FIR has finite-length h; IIR has an infinitely long h.

Next: Convolution Sum