Peak Cancellation

The algorithm continuously monitors the complex baseband I/Q signal to identify samples where the instantaneous envelope magnitude exceeds a predefined clipping threshold. Detection typically operates on oversampled signals to accurately capture peak locations between original sample points. Key aspects include:

• Magnitude computation: Calculates sqrt(I² + Q²) for each sample
• Threshold comparison: Compares against the target clipping ratio (CR)
• Peak interpolation: Uses oversampling to locate true peak positions with sub-sample accuracy
• Windowed search: Prevents redundant cancellation on adjacent peaks within a guard interval

The cancellation pulse is a pre-computed, spectrally confined impulse response designed to match the occupied bandwidth of the original signal. Its shape directly determines the trade-off between PAPR reduction and ACLR degradation. Critical design parameters include:

• Spectral mask compliance: Pulse spectrum must fit within regulatory emission limits
• Time-domain compactness: Shorter pulses localize distortion but widen spectral footprint
• Windowing function: Raised-cosine or Kaiser windows shape the pulse for controlled roll-off
• Coefficient storage: Pulse samples are stored in a look-up table for real-time convolution

At each detected peak, the cancellation pulse is amplitude-scaled and phase-rotated to match the excess magnitude of the peak before subtraction. The scaling factor is computed as the difference between the peak amplitude and the clipping threshold. This process involves:

• Complex scaling: Pulse is multiplied by (|x_peak| - Threshold) * e^(j*angle(x_peak))
• Time alignment: Pulse center is aligned precisely with the detected peak sample index
• Accumulative subtraction: Multiple overlapping pulses are summed into a cancellation signal
• Baseband operation: All processing occurs on the complex envelope before digital up-conversion

Aggressive PAPR targets often require cascaded peak cancellation stages with progressively tighter thresholds. Each stage cancels peaks that survive previous stages due to peak regrowth from filtering or pulse overlap. Multi-stage design considerations:

• Stage sequencing: Initial stages use wider pulses for coarse reduction; later stages use narrower pulses for fine suppression
• Inter-stage filtering: Optional band-limiting filters between stages control spectral regrowth accumulation
• Convergence monitoring: CCDF curves are evaluated after each stage to verify progressive PAPR reduction
• Latency budget: Each stage adds processing delay, requiring balance against system latency requirements

Peak cancellation is typically implemented in FPGA or ASIC fabric within the digital front-end of base station transmitters. Real-time constraints demand pipelined architectures with deterministic latency. Implementation factors include:

• Parallel processing: Multiple peaks can be processed simultaneously in wideband signals
• Memory bandwidth: Cancellation pulse LUTs require fast access for real-time convolution
• Fixed-point precision: Coefficient quantization must balance EVM performance against logic resources
• Pipeline depth: Deep pipelines enable high clock rates but increase processing latency
• Power consumption: Cancellation logic contributes to overall digital front-end power budget

Peak cancellation effectiveness is evaluated through a multi-dimensional trade-space analysis. The primary metrics are:

• PAPR reduction gain: Measured in dB at a specific CCDF probability point (e.g., 10^-4)
• EVM degradation: In-band distortion introduced by the cancellation process, measured as % RMS
• ACLR impact: Change in adjacent channel leakage ratio before and after cancellation
• Spectral mask margin: Remaining headroom to regulatory emission limits after processing
• Processing latency: Total delay through the cancellation pipeline, critical for TDD systems

The fundamental trade-off is between PAPR reduction aggressiveness and signal fidelity preservation.

The overarching signal conditioning discipline that peak cancellation belongs to. CFR deliberately limits the peak amplitude of a transmit waveform to improve power amplifier efficiency and prevent compression. While peak cancellation is one implementation, other CFR methods include clipping, windowing, and tone reservation. The choice of CFR algorithm involves a trade-off between PAPR reduction gain, EVM degradation, and ACLR compliance.

A closely related technique often used synonymously with peak cancellation. Pulse injection adds pre-designed, spectrally confined cancellation pulses at detected peak locations to suppress amplitude while controlling ACLR. The key distinction is that pulse injection typically uses a fixed library of pre-computed pulses, while peak cancellation may adaptively scale and shape the cancellation pulse based on the detected peak magnitude and phase. Both methods aim to subtract energy precisely at peak locations.

The phenomenon where filtering a clipped or peak-cancelled signal causes previously suppressed amplitude peaks to reappear. This occurs because band-limiting filters remove the out-of-band cancellation energy, partially restoring the original peak structure. Peak cancellation algorithms must account for regrowth by:

• Applying iterative cancellation stages
• Over-cancelling peaks slightly to compensate
• Using cancellation pulses with inherent spectral containment to minimize filter interaction

A cascaded architecture applying successive stages of peak cancellation or clipping with progressively tighter thresholds. Each stage removes a portion of the peak energy, and intermediate filtering controls spectral regrowth. Multi-stage designs achieve aggressive PAPR reduction targets (8-12 dB) with controlled distortion by distributing the cancellation burden across stages. This prevents any single stage from introducing excessive EVM or ACLR degradation.

The primary in-band distortion metric that constrains peak cancellation aggressiveness. EVM quantifies the deviation of measured constellation points from their ideal reference positions. Peak cancellation introduces EVM because the subtracted cancellation pulse distorts the data-bearing signal within the occupied bandwidth. System designers must balance:

• PAPR reduction (more cancellation = better efficiency)
• EVM budget (regulatory limits like 3GPP specify maximum EVM per modulation scheme)
• ACLR limits (spectral mask compliance)