Inferensys

Glossary

Cubic Metric (CM)

A figure of merit estimating the power de-rating required for a power amplifier to handle a given signal's envelope statistics, accounting for third-order nonlinearity.
Stylish WeWork-like workspace with hot desks and document wall, professional searching through enterprise knowledge base on a mounted ultrawide display, warm industrial pendants overhead.
POWER DE-RATING METRIC

What is Cubic Metric (CM)?

Cubic Metric is a figure of merit that estimates the power back-off required for a power amplifier to handle a specific signal's envelope statistics, accounting for third-order nonlinearity.

The Cubic Metric (CM) is a figure of merit that quantifies the power de-rating, or back-off, required for a power amplifier (PA) to transmit a given signal without exceeding a specified adjacent channel leakage ratio (ACLR) limit. Unlike the Peak-to-Average Power Ratio (PAPR), which only considers the signal's peak excursions, the CM specifically accounts for the third-order nonlinearity of the amplifier, making it a more accurate predictor of the PA's efficiency degradation for a particular waveform.

CM is derived from the statistical distribution of the signal's instantaneous power, calculating the power of the signal's envelope cubed. This third-order term directly correlates to the dominant source of spectral regrowth in a weakly nonlinear PA. A higher CM value indicates that the signal's envelope statistics will excite more nonlinear distortion, requiring greater input back-off and thus reducing the amplifier's power efficiency compared to a signal with a lower CM but identical PAPR.

SIGNAL METRIC

Key Characteristics of Cubic Metric

Cubic Metric (CM) is a figure of merit that estimates the power de-rating required for a power amplifier to handle a given signal's envelope statistics, specifically accounting for third-order nonlinearity.

01

Third-Order Nonlinearity Sensitivity

Unlike Peak-to-Average Power Ratio (PAPR), which only considers amplitude statistics, CM directly models the impact of third-order intermodulation distortion on amplifier performance.

  • Derived from the third-order moment of the signal envelope voltage
  • Correlates strongly with Adjacent Channel Leakage Ratio (ACLR) degradation
  • Accounts for the fact that signals with identical PAPR can produce different levels of spectral regrowth
  • Defined in the 3GPP specification as a standard metric for UE power de-rating
02

Mathematical Definition

CM is calculated from the normalized raw cubic metric (RCM) of the signal waveform, referenced against a 12.2 kbps Adaptive Multi-Rate (AMR) voice reference signal.

  • RCM is computed as the root-mean-cubed value of the instantaneous power normalized to RMS: RCM = 20 * log10(rms[(v_norm^3)])
  • CM is then expressed as: CM = RCM - RCM_reference
  • The reference RCM for the AMR 12.2 kbps signal is approximately 1.52 dB
  • A CM of 2 dB indicates the signal requires 2 dB more power back-off than the reference to achieve equivalent ACLR performance
03

Relationship to Crest Factor Reduction

Crest Factor Reduction (CFR) algorithms directly influence CM by modifying the signal's envelope statistics. However, not all CFR techniques affect CM equally.

  • Hard clipping introduces sharp discontinuities that generate strong third-order products, potentially increasing CM despite reducing PAPR
  • Peak windowing and peak cancellation with spectrally confined pulses better control CM growth
  • Effective CFR must be evaluated against both PAPR and CM to ensure real-world amplifier efficiency gains
  • A signal with low PAPR but high CM may still cause unacceptable spectral regrowth
04

CM vs. PAPR: Practical Distinction

While PAPR characterizes the envelope peakiness, CM captures the nonlinear distortion potential that actually determines amplifier back-off requirements.

  • Two signals with identical 9 dB PAPR can have CM values differing by 1-2 dB
  • CM is a better predictor of power amplifier efficiency in real operating conditions
  • Modern 3GPP and ETSI standards specify CM limits for user equipment to ensure consistent network performance
  • Power Amplifier Back-off determined by CM rather than PAPR alone results in more accurate efficiency optimization
05

Measurement and Compliance

CM is measured using signal envelope statistics captured from the baseband I/Q waveform before digital-to-analog conversion.

  • Requires computation of the Complementary Cumulative Distribution Function (CCDF) of the cubed envelope
  • Test equipment and signal analyzers provide automated CM measurement capabilities
  • Typical 3GPP CM limits for LTE/NR user equipment range from 1.0 to 3.0 dB depending on modulation scheme and resource block allocation
  • Exceeding CM limits triggers Maximum Power Reduction (MPR) requirements to maintain ACLR compliance
06

Impact on Digital Predistortion Design

Digital Pre-Distortion (DPD) systems must account for the CM of the target signal when designing linearization strategies.

  • High-CM signals require DPD with greater correction bandwidth to address third-order distortion products
  • Memory polynomial models used in DPD are inherently structured around odd-order nonlinearities that CM quantifies
  • CM-aware Crest Factor Reduction can be jointly optimized with DPD to balance efficiency and linearity
  • Signals pre-conditioned for low CM reduce the computational complexity required in the DPD coefficient estimation path
CUBIC METRIC ESSENTIALS

Frequently Asked Questions

Critical questions about Cubic Metric (CM) and its role in estimating power amplifier back-off for modern communication signals with non-constant envelopes.

Cubic Metric (CM) is a figure of merit that estimates the power de-rating, or back-off, required for a power amplifier to handle a given signal's envelope statistics, specifically accounting for third-order nonlinearity. It is mathematically defined as the ratio of the root-mean-cubed (RMC) power of the signal to its root-mean-squared (RMS) power, expressed in decibels. Unlike Peak-to-Average Power Ratio (PAPR), which only considers the peak instantaneous power, CM directly correlates with the amount of third-order intermodulation distortion generated when the signal passes through a nonlinear amplifier. The 3GPP standardization body adopted CM as a more accurate metric than PAPR for predicting the actual power de-rating needed to meet adjacent channel leakage ratio requirements.

POWER DE-RATING METRICS

Cubic Metric vs. Peak-to-Average Power Ratio

Comparison of the two primary figures of merit used to estimate the required power amplifier back-off for a given modulated signal, highlighting the superior accuracy of Cubic Metric in capturing third-order nonlinearity effects.

FeaturePeak-to-Average Power Ratio (PAPR)Cubic Metric (CM)

Definition

Ratio of peak instantaneous power to average power of the signal envelope.

Estimate of power de-rating required relative to a reference signal, accounting for third-order nonlinearity.

Mathematical Basis

max(|x(t)|²) / E[|x(t)|²]

20·log₁₀{rms[(|x(t)|/rms[x(t)])³]} / 1.52

Reference Signal

None (absolute power ratio).

12.2 kbps Adaptive Multi-Rate (AMR) voice signal (normalized to 0 dB).

Captures Amplifier Nonlinearity

Accounts for Signal Envelope Statistics

Sensitivity to Third-Order Distortion

None. Ignores the cubic term of the amplifier transfer function.

Directly models the third-order nonlinearity via the cubed voltage term.

Correlation with Measured Power De-Rating

Poor. Often overestimates or underestimates required back-off.

High. Provides a more accurate prediction of the actual output power back-off needed.

Standardization Body

Generally defined in signal processing literature.

3GPP (TS 25.101, TS 36.101, TS 38.101).

Prasad Kumkar

About the author

Prasad Kumkar

CEO & MD, Inference Systems

Prasad Kumkar is the CEO & MD of Inference Systems and writes about AI systems architecture, LLM infrastructure, model serving, evaluation, and production deployment. Over 5+ years, he has worked across computer vision models, L5 autonomous vehicle systems, and LLM research, with a focus on taking complex AI ideas into real-world engineering systems.

His work and writing cover AI systems, large language models, AI agents, multimodal systems, autonomous systems, inference optimization, RAG, evaluation, and production AI engineering.