What Are Linearization Techniques?

Linearization techniques are methods used to approximate or transform a nonlinear system, device, or function so that it can be analyzed and controlled using linear theory. Because most physical systems exhibit some degree of nonlinearity, engineers across control, circuit design, and communications routinely apply linearization to make design tractable without sacrificing accuracy within a defined operating range. The approach exploits the observation that a smooth nonlinear function behaves approximately like a straight line over a sufficiently small region, and much of linear systems theory then applies within that region.

The motivation for linearization is partly analytical and partly practical. Linear models support exact closed-form solutions, stability guarantees, and a rich toolbox of frequency-domain and state-space methods. Nonlinear models often require numerical simulation and offer far fewer guarantees on stability or performance. Linearization bridges the two regimes.

Control System Synthesis

The most common linearization method in control engineering is Jacobian linearization, also called small-signal or operating-point linearization. Given a nonlinear state-space model, the designer selects an equilibrium point and expands the dynamics using a first-order Taylor series. The resulting linear model, parameterized by the Jacobian matrices evaluated at that equilibrium, is valid for small perturbations around the chosen point. Standard tools for controller synthesis, including PID tuning, pole placement, linear-quadratic regulator design, and H-infinity methods, can then be applied directly to the linearized model. When the operating range is large or the system has multiple equilibria, gain scheduling extends this approach by computing separate linear models at several operating points and blending controllers between them.

A more exact alternative is feedback linearization, a nonlinear control technique that uses a state transformation and a nonlinear control law to cancel the nonlinear terms algebraically, producing a closed-loop system that is exactly linear in the new coordinates. Unlike Jacobian linearization, feedback linearization does not restrict validity to a small neighborhood. Detailed treatments of feedback linearizing control appear in graduate-level texts on nonlinear control, including the lecture notes maintained by Michael Henson at the University of South Carolina.

MOSFET Circuit Linearization

In analog and RF integrated circuits, nonlinearity arises from the transconductance characteristics of MOSFETs, which produce harmonic distortion and intermodulation products when driven with large signals. Several circuit-level linearization techniques address this. Derivative superposition places multiple transistors in parallel with different bias points, exploiting cancellation between their third-order nonlinear terms. IM2 injection deliberately introduces a second-order intermodulation signal that cancels the dominant distortion component at the output. Noise and distortion cancellation techniques, analyzed in detail for CMOS low-noise amplifier design in the IEEE Transactions on Circuits and Systems, achieve simultaneous improvement in noise figure and linearity by using a complementary signal path. Optimal biasing, sometimes called class-AB biasing in power amplifiers, places the transistor at a quiescent point where the third-order Volterra coefficient passes through zero, giving a local linearity maximum without external correction circuitry.

RF Transmitter Linearization

In RF transmitters, power amplifiers are a primary source of nonlinearity. Three main approaches address this: predistortion, feedback, and feedforward. Digital predistortion (DPD) characterizes the amplifier's nonlinear behavior and applies an inverse function to the baseband signal before upconversion, so that the cascade of predistorter and amplifier approximates a linear response. Cartesian feedback wraps a baseband feedback loop around the amplifier in in-phase and quadrature branches, directly suppressing distortion through negative feedback. Feedforward uses a cancellation path to subtract the error signal at the output. A comprehensive review of analog linearization methods for integrated transmitters published in IEEE conference proceedings covers these techniques and their implementation tradeoffs.

Applications

Linearization techniques have applications across a wide range of engineering contexts, including:

• Closed-loop control of robotic manipulators, aircraft, and process control systems
• RF power amplifier design in cellular base stations and handsets
• Low-noise amplifier design for radar and satellite receivers
• Analog-to-digital converter characterization and calibration
• Small-signal circuit models used in SPICE simulation and analog IC design