LTI Systems: Properties and Analysis With easy examples.

An LTI system is a system that is both linear and time-invariant. Linear means it obeys superposition: the response to a sum of inputs is the sum of the individual responses, and scaling the input scales the output by the same factor. Time-invariant means its behavior never changes with time: delaying the input simply delays the output by the same amount, without altering its shape.

LTI systems matter because they are the best-understood class of systems in all of engineering, and a huge number of real devices behave like them: resistors, capacitors, inductors, filters, amplifiers (in their linear region), communication channels and even room acoustics. The entire theory of convolution, frequency response and transfer functions rests on these two properties, so recognizing an LTI system unlocks a powerful toolbox.

Key Concepts

• Superposition: T{a x1 + b x2} = a T{x1} + b T{x2} for any constants a, b and inputs x1, x2.
• Time invariance: if input x(t) produces output y(t), then input x(t - t0) produces output y(t - t0) for any delay t0.
• Impulse response h(t): the output when the input is a unit impulse. For LTI systems it completely characterizes the system: every output is the convolution of the input with h(t).
• Causality: the output never anticipates the input; h(t) = 0 for t less than 0.
• BIBO stability: every bounded input produces a bounded output — the practical definition of a stable system.
• Memory: a system has memory if the output depends on past (or future) inputs; a resistor is memoryless, a capacitor is not.

Formulas

• Linearity test: for inputs x1, x2 giving y1, y2, check T{x1 + x2} = y1 + y2 (additivity) and T{a x} = a y (homogeneity).
• Time-invariance test: shift the input by t0 and check the output shifts identically, with no other change.
• BIBO stability: a continuous LTI system is stable if and only if the integral of |h(t)| over all t is finite; discrete-time if the sum of |h[n]| is finite.
• Eigenfunction property: complex exponentials e^(st) are eigenfunctions of LTI systems — the output is the same exponential scaled by H(s), the transfer function.

Worked Example

Test whether the discrete-time system y[n] = 3 x[n] + 2 is linear and time-invariant.

1. Linearity — pick x1[n] = 1 and x2[n] = 2. Then y1[n] = 3x1 + 2 = 5 and y2[n] = 3x2 + 2 = 8.
2. Now apply the sum x[n] = x1[n] + x2[n] = 3. The system gives y[n] = 3x3 + 2 = 11. But y1 + y2 = 5 + 8 = 13. Since 11 is not equal to 13, additivity fails — the system is NOT linear (the constant +2 is the culprit).
3. Time invariance — delay the input: x[n - 1] = 1 gives y[n] = 3x1 + 2 = 5, which equals the delayed output y[n-1] = 5. The +2 constant does not depend on n, so the system IS time-invariant.
4. Conclusion: time-invariant but nonlinear, hence not LTI. A plain y[n] = 3 x[n] would pass both tests and be LTI.

This shows the strict meaning of "linear" in systems theory: even a harmless-looking constant offset destroys linearity.

Common Misconceptions

• "y = 3x + 2 is linear because its graph is a straight line." In systems theory it is not linear — superposition fails, as shown above. Only straight lines through the origin are linear systems.
• "Time-invariant means the output never changes." Wrong. It means the system's rules never change; the output happily varies as the input varies, and a delayed input gives an equally delayed output.
• "All practical systems are LTI." Many are only approximately LTI over a limited range — amplifiers saturate, components drift with temperature. LTI analysis applies where the approximation holds.

Real-World Applications

• Filter design: every analog and digital filter is analyzed as an LTI system via its impulse or frequency response.
• Audio engineering: equalizers, reverb and room correction all use LTI convolution models.
• Communications: channels are modeled as LTI (plus noise), so equalizers invert the channel's impulse response.
• Control systems: stability of feedback loops is decided by LTI pole locations.
• Image processing: blurring and sharpening are 2-D LTI convolution operations.

A system is LTI when it respects superposition and never changes its rules over time. Then its entire behavior is captured by one impulse response, stability reduces to checking that the impulse response is absolutely summable, and analysis moves to the elegant worlds of convolution and transforms.