System Definition

Introduction

In signal processing, a system is any process that transforms an input signal into an output signal according to some rule or set of rules. This definition is deliberately broad because systems appear everywhere: an amplifier transforms a weak electrical signal into a stronger one, a communication channel transforms a transmitted signal into a received (possibly distorted) version, a digital filter transforms a sequence of numbers into another sequence, and even the human ear transforms pressure waves into neural signals.

The mathematical abstraction of a system allows us to analyze, design, and predict the behavior of these diverse physical processes using a common framework. Rather than studying each device from first principles of physics, we characterize its input-output behavior and apply powerful system-theoretic tools.

Mathematical Representation

A system $T$ is represented as an operator that maps an input signal $x$ to an output signal $y$:

$$y(t) = T\{x(t)\}$$

For discrete-time systems: $$y[n] = T\{x[n]\}$$

The operator $T$ encapsulates everything the system does — amplification, filtering, delay, distortion, or any combination thereof. The system is completely characterized by specifying what output it produces for every possible input.

Multiple Representations

The same system can be described in several equivalent ways:

Time domain: Impulse response $h(t)$ or difference equation relating input and output

Transform domain: Transfer function $H(s)$, $H(z)$, or $H(j\omega)$

State space: State equations $\dot{x} = Ax + Bu$, $y = Cx + Du$

Block diagram: Graphical interconnection of basic elements (adders, multipliers, delays)

Each representation offers different analytical advantages, but they all describe the same underlying system behavior.

System Classification Framework

Systems are classified along several independent axes. A given system may be:

These classifications are independent — a system can be linear but time-varying, causal but unstable, or any other combination. The most analytically tractable systems are Linear Time-Invariant (LTI) systems, which satisfy both linearity and time-invariance simultaneously.

Input-Output Relationships

Systems can be described by explicit input-output equations:

Algebraic (memoryless): $y(t) = 3x(t) + 5$ — output at time $t$ depends only on input at time $t$

Differential equation: $\frac{dy}{dt} + 2y(t) = x(t)$ — output depends on input and past output history

Difference equation: $y[n] = 0.5y[n-1] + x[n]$ — discrete-time version with memory

Integral equation: $y(t) = \int_0^t x(\tau) d\tau$ — accumulates input over time

Convolution: $y(t) = \int_{-\infty}^{\infty} h(\tau)x(t-\tau) d\tau$ — general LTI system representation

Continuous-Time vs. Discrete-Time Systems

Continuous-time systems process signals defined for all real values of $t$. Examples include analog circuits (amplifiers, RLC filters), mechanical systems (mass-spring-damper), and electromagnetic wave propagation.

Discrete-time systems process sequences defined only at integer indices $n$. Examples include digital filters (implemented in software or DSP hardware), recursive algorithms, and any system operating on sampled data.

Hybrid systems contain both continuous and discrete-time components, connected through ADCs and DACs. A modern control system typically senses analog signals, processes them digitally, and produces analog actuator commands.

SISO vs. MIMO Systems

Single-Input Single-Output (SISO): One input signal produces one output signal. Most introductory analysis focuses on SISO systems.

Multiple-Input Multiple-Output (MIMO): Multiple input signals produce multiple output signals simultaneously. The system is characterized by a matrix of transfer functions. MIMO systems arise in multi-antenna communications, multi-variable control, and multi-channel audio processing.

The Concept of System Response

The response of a system to a specific input is called the system response to that input:

• Impulse response $h(t)$: Response when input is $\delta(t)$
• Step response $s(t)$: Response when input is $u(t)$
• Frequency response $H(j\omega)$: Steady-state response to sinusoidal inputs at each frequency
• Zero-state response: Output due to input alone (zero initial conditions)
• Zero-input response: Output due to initial conditions alone (zero input)

For LTI systems, the impulse response completely characterizes the system — the response to any arbitrary input can be computed via convolution with $h(t)$.

Real-World System Examples

Amplifier: $y(t) = Gx(t)$ where $G$ is the gain. Linear, time-invariant, memoryless, causal, stable, invertible.

Communication channel: $y(t) = h(t) * x(t) + n(t)$ where $h(t)$ models multipath propagation and $n(t)$ is additive noise.

Digital averaging filter: $y[n] = \frac{1}{M}\sum_{k=0}^{M-1}x[n-k]$ — FIR filter that smooths the input.

Thermostat: Nonlinear (bang-bang control), time-invariant, causal, stable, with memory (hysteresis).

Key Takeaways

• A system transforms input signals into output signals according to a defined rule: $y = T\{x\}$
• Systems are classified along multiple independent axes: linearity, time-invariance, causality, stability, memory, invertibility
• The same system can be represented as an equation, transfer function, state-space model, or block diagram
• For LTI systems, the impulse response $h(t)$ completely characterizes the system
• Real systems span the full range from simple (amplifier) to complex (communication channels)
• The system classification framework determines which analysis tools apply to a given system