Crest Factor Reduction (CFR)

The most fundamental CFR technique. The signal magnitude is compared against a predefined threshold, and any sample exceeding this limit is hard-limited to the threshold value. This non-linear operation causes severe spectral regrowth, necessitating a subsequent filtering stage to suppress out-of-band emissions. However, filtering causes peak regrowth, often requiring multiple iterations of clip-and-filter to meet both PAPR and adjacent channel leakage ratio (ACLR) targets.

Instead of hard-clipping, peak windowing multiplies the signal by a smooth window function (e.g., Gaussian, Kaiser, or raised-cosine) centered around each detected peak. This shapes the clipping noise, concentrating its spectrum more effectively than hard clipping and reducing the filtering burden. The window's width and shape are critical design parameters: a wider window better suppresses out-of-band emissions but corrupts more adjacent samples, increasing EVM.

A sophisticated method that subtracts a pre-designed, spectrally-shaped cancellation pulse from the signal at each detected peak. The cancellation pulse is engineered to occupy the same bandwidth as the original signal, ensuring that the correction energy falls strictly in-band and does not cause spectral regrowth. This eliminates the need for iterative filtering. The pulse is typically a scaled sinc function or a digitally pre-distorted kernel optimized for the specific carrier configuration.

A distortion-free technique used in multi-carrier systems like OFDM (Orthogonal Frequency-Division Multiplexing). A subset of subcarriers is reserved and dedicated exclusively to carrying a peak-canceling signal. These reserved tones are orthogonal to the data-carrying tones, meaning the cancellation signal does not interfere with the data payload. The challenge lies in optimizing the cancellation signal on the reserved tones in real-time to effectively suppress peaks without exceeding their allocated power budget.

A non-uniform quantization technique adapted for CFR. The signal is passed through a compressor with a non-linear transfer function (e.g., µ-law or A-law) that amplifies low-amplitude samples while limiting high-amplitude peaks. At the receiver, an expander with the inverse function restores the original dynamic range. While simple to implement, companding introduces non-linear distortion across the entire signal, not just the peaks, leading to a uniform degradation of EVM.

An intelligent CFR method for modulated signals that exploits the decision boundaries of the constellation diagram. Outer constellation points are dynamically moved outward—away from the decision region—to reduce signal peaks without crossing the decision thresholds. This smart clipping introduces no symbol errors, making it highly power-efficient. ACE is particularly effective for dense QAM constellations and is often combined with iterative clipping for aggressive PAPR targets.

The fundamental metric that CFR aims to reduce. PAPR quantifies the ratio between a signal's peak power and its average power, typically expressed in dB. Orthogonal Frequency Division Multiplexing (OFDM) signals in 5G and Wi-Fi exhibit PAPR values of 10-13 dB, forcing power amplifiers to operate with significant back-off from their compression point.

• High PAPR directly reduces PA efficiency to 15-25% in linear operation
• CFR typically targets 6-8 dB PAPR reduction before the PA
• Every 1 dB of PAPR reduction can yield 2-3% efficiency improvement

The complementary linearization technique that follows CFR in the transmit chain. While CFR reduces signal peaks to prevent hard clipping, DPD applies an inverse nonlinearity to compensate for the PA's remaining soft compression and memory effects. The two techniques are co-designed because aggressive CFR can alter the PA's nonlinear behavior that DPD must then correct.

• CFR handles peak excursions; DPD handles in-band distortion
• Joint CFR-DPD optimization improves ACLR by 3-5 dB over independent tuning
• Modern FPGA implementations combine both in a single processing pipeline

A widely used CFR algorithm that multiplies the signal by a smooth window function around detected peaks, rather than hard-clipping. Common window functions include Kaiser, Hamming, and raised-cosine, each trading off PAPR reduction against spectral regrowth.

• Kaiser window: Adjustable side-lobe suppression via beta parameter
• Raised-cosine: Smooth roll-off with controlled excess bandwidth
• Peak windowing preserves the signal's spectral mask better than clipping
• Window length determines the trade-off between EVM and ACLR

An advanced CFR technique that detects peaks exceeding a threshold and subtracts a pre-designed cancellation pulse at each peak location. The cancellation pulse is spectrally shaped to match the transmit filter, ensuring that the correction energy falls within the allocated channel bandwidth rather than spilling into adjacent channels.

• Iterative pulse cancellation achieves deeper PAPR reduction
• Cancellation pulses are pre-computed and stored in LUTs for real-time operation
• Multi-stage implementations cascade 2-3 cancellation stages for progressive peak reduction

The primary signal quality metric that degrades as CFR aggressiveness increases. EVM measures the deviation of transmitted symbols from their ideal constellation positions, and excessive CFR introduces in-band distortion that directly increases EVM. Standards like 3GPP specify maximum EVM limits (e.g., 3.5% for 256-QAM in 5G NR) that constrain how much CFR can be applied.

• CFR-induced EVM comes from peak clipping and windowing distortion
• 256-QAM requires EVM below 3.5%; 1024-QAM below 1.5%
• Adaptive CFR algorithms dynamically adjust thresholds based on modulation order

The spectral containment metric that CFR must preserve. ACLR measures the ratio of transmitted power within the assigned channel to power leaking into adjacent channels. Poorly designed CFR generates spectral regrowth that violates regulatory masks. Modern CFR algorithms incorporate spectral shaping to ensure that peak reduction energy remains within the channel.

• 3GPP specifies ACLR limits of 45 dB for base stations
• Spectral shaping filters in CFR constrain out-of-band emissions
• Joint CFR-DPD optimization can improve ACLR by 5-10 dB over CFR alone