In-Band Distortion

The primary metric for quantifying in-band distortion. EVM measures the vector difference between the ideal reference constellation point and the actual transmitted point. CFR-induced clipping directly increases this error.

• Mechanism: Clipping truncates the signal envelope, distorting both amplitude and phase of the modulated symbols.
• Impact: Higher-order modulation schemes (e.g., 256-QAM, 1024-QAM) are significantly more sensitive to EVM degradation than QPSK.
• Measurement: Expressed as a percentage of the ideal symbol magnitude or in decibels (dB).

A visual representation of in-band distortion observed on a vector signal analyzer. Instead of tight, distinct clusters, the constellation points spread into diffuse clouds.

• Cause: The nonlinear clipping process adds uncorrelated noise-like interference to each symbol.
• Result: The decision boundaries between adjacent symbols become ambiguous, increasing the probability of a symbol error at the receiver.
• Visual Indicator: A 'fuzzy' constellation diagram is the definitive sign of excessive in-band distortion.

In-band distortion directly reduces the transmitter's modulation accuracy, limiting the achievable data rate and link reliability.

• Bit Error Rate (BER) Floor: Even with a high signal-to-noise ratio (SNR) in the channel, in-band distortion creates an irreducible BER floor that cannot be overcome by increasing transmit power.
• Throughput Loss: Adaptive modulation and coding (AMC) algorithms in modern systems will downgrade to a more robust, lower-order modulation scheme to compensate, reducing peak data throughput.

In-band distortion can be mathematically modeled as the addition of an uncorrelated, nonlinear noise component to the clean signal.

• Spectral Nature: Unlike thermal noise, this distortion is signal-dependent and deterministic for a given clipping threshold.
• EVM vs. PAPR Trade-off: This is the fundamental engineering compromise. Aggressive clipping reduces PAPR and improves power amplifier efficiency but injects more nonlinear noise, degrading EVM.
• Regulatory Constraint: Standards like 3GPP define strict EVM limits (e.g., 3.5% for 64-QAM in 5G NR) that directly constrain the maximum allowable in-band distortion from CFR.

The severity of in-band distortion is a direct function of the Clipping Ratio (CR), defined as the ratio of the clipping threshold to the RMS signal level.

• Low CR (Aggressive Clipping): A threshold set very close to the average power level clips frequently, causing severe in-band distortion and high EVM.
• High CR (Light Clipping): A threshold set far above the average level clips only rare, extreme peaks, resulting in minimal EVM but also minimal PAPR reduction.
• Design Goal: The CFR algorithm must find the optimal CR that satisfies both the EVM specification and the power amplifier efficiency target.

In-band distortion corrupts the signal in ways that stress the receiver's synchronization and channel estimation algorithms.

• Phase Noise Addition: Clipping introduces random phase perturbations that can degrade the performance of phase-locked loops in the receiver.
• Pilot Contamination: In OFDM systems, distortion on pilot subcarriers corrupts the channel estimate, causing a multiplicative error that affects all data subcarriers.
• Common Phase Error (CPE): A bulk rotation of the entire constellation caused by the average phase distortion from clipping, which must be tracked and corrected by the receiver.

The primary metric for quantifying in-band distortion. EVM measures the vector difference between the ideal constellation point and the actual transmitted point after CFR. High EVM directly degrades modulation accuracy and bit error rate (BER).

• Expressed as a percentage or in dB relative to the ideal signal power
• 3GPP defines strict EVM limits per modulation scheme (e.g., 3.5% for 64QAM)
• CFR algorithms must balance EVM degradation against PAPR reduction gain

An iterative CFR process where hard clipping truncates signal peaks, followed by frequency-domain filtering to suppress out-of-band spectral regrowth. However, filtering causes peak regrowth, where previously suppressed peaks reappear.

• Multiple iterations are required to converge on the target PAPR
• Each iteration introduces additional in-band distortion
• The trade-off between EVM, ACLR, and PAPR defines the algorithm's performance envelope

The ability of the receiver to correctly demodulate the transmitted symbols. In-band distortion disperses the constellation points, reducing the noise margin and increasing the probability of symbol errors.

• Higher-order modulations (256QAM, 1024QAM) are more sensitive to EVM
• 5G NR requires extremely low EVM for high data rate modulation schemes
• CFR-induced distortion must stay within the receiver's equalization capability

The phenomenon where filtering a clipped signal causes previously suppressed amplitude peaks to reappear above the clipping threshold. This occurs because filtering removes out-of-band spectral components but redistributes energy in the time domain.

• Necessitates iterative clipping and filtering stages
• Each iteration trades additional EVM for reduced peak regrowth
• Multi-stage CFR architectures progressively tighten the clipping ratio to manage this effect

Defined as the ratio of the maximum permitted signal amplitude after clipping to the RMS level of the unclipped signal. A lower CR indicates more aggressive PAPR reduction but introduces more in-band distortion.

• Typical CR values range from 4 dB to 8 dB depending on the modulation scheme
• CR directly controls the probability of clipping events
• Must be selected based on the CCDF target and acceptable EVM budget

A CFR approach that applies a smooth saturation function (e.g., hyperbolic tangent or polynomial) to limit signal peaks instead of a hard threshold. This reduces the sharp discontinuities that cause severe spectral splatter.

• Produces less out-of-band emission than hard clipping
• Introduces a more gradual, controlled in-band distortion profile
• Often combined with filtering for a balanced EVM-ACLR trade-off