Skip to content
Tutorials Center
My learning

Topics

Lesson 9 of 36 ยท LTI Systems and Convolution

System Properties: Linearity, Time-Invariance, Causality, Stability

Learn to test a system for memory, causality, stability, linearity, time-invariance and invertibility, with worked examples and a small Python tester.

Not all systems are equally easy to analyze. Some obey simple rules that let you predict the response to any input from just a few measurements. Knowing which properties a system has is the first step of analysis. There are six to learn: memory, causality, stability, linearity, time-invariance and invertibility.

We write y = H{x} to say that system H turns input x into output y.

Memory

A system is memoryless (static) if the output at time t depends only on the input at the same time. y(t) = 3 x(t) and y[n] = x[n]^2 are memoryless. y[n] = x[n] + x[n-1] has memory: it needs the previous sample. An integrator and a delay both have memory.

Causality

A system is causal if the output at time t0 depends only on the input at times t <= t0: it cannot see the future. All physical real-time systems are causal. y[n] = x[n] + x[n-1] is causal; y[n] = x[n+1] is not; y(t) = x(-t) is not (at t = -1 it needs x(1)). Non-causal systems are fine for offline processing, such as smoothing a recorded signal with a window centred on each sample.

Stability (BIBO)

A system is bounded-input, bounded-output (BIBO) stable if every bounded input produces a bounded output: |x| <= M implies |y| <= M'. y[n] = 0.5 y[n-1] + x[n] is stable. An accumulator y[n] = sum of x[k] is unstable: the bounded input x[n] = 1 gives output n + 1, which grows forever.

Linearity

A system is linear if it obeys superposition: for any signals x1, x2 and constants a, b,

H{a x1 + b x2} = a H{x1} + b H{x2}.

This combines two ideas: additivity and homogeneity (scaling). A quick necessary test: a linear system maps zero input to zero output. y = 2x + 1 fails (input 0 gives output 1) so it is not linear, even though its graph is a straight line: the technical word is affine. y = x^2 is also not linear.

Time-invariance

A system is time-invariant if delaying the input just delays the output: if H{x(t)} = y(t) then H{x(t - t0)} = y(t - t0) for every t0. Its rules do not change with the clock.

Test: (1) delay the input, then apply the system; (2) apply the system, then delay the output; compare. For y[n] = n x[n]: delay the input first, H{x[n-1]} = n x[n-1]; delay the output second, y[n-1] = (n-1) x[n-1]. These differ, so the system is time-varying. Anything that explicitly contains n or t outside of x(...) is a warning sign, as are time scaling and reversal.

Invertibility

A system is invertible if distinct inputs always give distinct outputs, so that an inverse system can recover the input. y = 2x is invertible (divide by 2); y = x^2 is not (inputs 3 and -3 give the same output); a differentiator is not (it loses the constant).

A numerical tester

We cannot prove a property by trying examples, but we can disprove one by finding a counterexample. The tester below checks superposition and shift-invariance on random inputs.

Python Testing linearity and time-invariance

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 table shows square and affine 2x+1 failing linearity, and n * x[n] failing time-invariance. The accumulator is linear and time-invariant but not BIBO stable.

Python The accumulator is not BIBO stable

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.

A summary table

System Memory Causal Stable Linear Time-invariant
y = 3x no yes yes yes yes
y[n] = x[n] + x[n-1] yes yes yes yes yes
y[n] = x[n+1] yes no yes yes yes
y = x^2 no yes yes no yes
y[n] = n x[n] no yes no yes no
accumulator yes yes no yes yes
flowchart TB
    A["all systems"] --> B["linear"]
    B --> C["linear and time-invariant (LTI)"]
    C --> D["LTI and stable"]
    D --> E["LTI, stable, causal: real filters"]

Each extra property narrows the family; LTI is the sweet spot for theory.

Check your understanding

Quick check ยท 5 questions

of 5 answered You scored / 5 ยท perfect!

  1. 1. Which system is memoryless?

    Only y[n] = 5 x[n] uses the input at the same instant and nothing else.
  2. 2. y(t) = x(t + 1) is:

    The output at t depends on the input at t + 1, which is in the future.
  3. 3. Why is y = 2x + 1 not linear?

    Linear systems map zero to zero. An affine system fails scaling: H{2x} = 4x + 1, while 2H{x} = 4x + 2.
  4. 4. y[n] = n x[n] is:

    It obeys superposition, but delaying the input does not simply delay the output because the factor n changes with time.
  5. 5. A bounded input to a BIBO stable system always gives:

    That is the definition of BIBO stability.

Summary

  • Memory: depends on other times than now. Causality: depends only on past and present.
  • BIBO stability: bounded in, bounded out.
  • Linear: superposition holds (and zero in gives zero out). Time-invariant: a delayed input gives a delayed output.
  • Examples and counterexamples can disprove a property; a proof needs algebra.
  • Linear time-invariant (LTI) systems are the class the rest of the chapter studies.

Next: Impulse Response and LTI Systems