Out-of-Band Emission

Out-of-band emission primarily arises from spectral regrowth caused by nonlinear amplification. When a signal with high peak-to-average power ratio (PAPR) passes through a power amplifier operating near saturation, the third-order intermodulation products spread energy into adjacent channels. This regrowth is not present in the original baseband signal but is generated by the amplitude-to-amplitude (AM-AM) and amplitude-to-phase (AM-PM) distortion of the amplifier.

The primary metric for quantifying out-of-band emission is Adjacent Channel Leakage Ratio (ACLR). ACLR measures the ratio of transmitted power within the assigned channel to the power leaking into adjacent frequency channels. Regulatory bodies such as 3GPP and ETSI specify minimum ACLR requirements—typically 45 dBc for the first adjacent channel in LTE and 5G NR systems. Failure to meet ACLR limits results in certification rejection.

A spectral mask defines the maximum allowable emission power as a function of frequency offset from the carrier. Masks are specified by standards organizations and vary by radio access technology. Key mask parameters include:

• In-band emission limits within the occupied bandwidth
• Out-of-band emission limits immediately outside the channel
• Spurious emission limits at far-out frequency offsets Nonlinear distortion from crest factor reduction (CFR) and power amplifier compression must be controlled to stay within these boundaries.

Crest factor reduction (CFR) techniques deliberately clip or window signal peaks to improve power amplifier efficiency, but this nonlinear operation inherently generates out-of-band emission. There is a fundamental trade-off: more aggressive peak-to-average power ratio reduction yields better amplifier efficiency but produces more spectral regrowth. Advanced CFR algorithms like peak windowing and pulse injection are designed to minimize this regrowth while achieving target PAPR reduction.

The bandwidth of out-of-band emission typically extends 3 to 5 times the original signal bandwidth due to nonlinear distortion. For a 20 MHz LTE carrier, third-order intermodulation products can spread energy across 60 MHz or more. This expansion is particularly problematic in carrier aggregation and multi-standard radio deployments where multiple carriers operate in close spectral proximity, requiring sophisticated digital predistortion (DPD) to contain the emission footprint.

Out-of-band emission is characterized using spectrum analyzers and vector signal analyzers with specific measurement techniques:

• ACLR measurements using integrated power in adjacent channels
• Spectrum emission mask (SEM) tests across defined frequency offsets
• CCDF curves to correlate PAPR statistics with emission levels
• Error vector magnitude (EVM) measurements to assess in-band distortion alongside out-of-band leakage These measurements are critical during regulatory pre-compliance testing and power amplifier characterization.

The primary regulatory metric quantifying out-of-band emission. ACLR measures the ratio of transmitted power within the assigned channel to power leaking into adjacent frequency channels. Nonlinearities from crest factor reduction and power amplifier compression directly degrade ACLR. Regulatory bodies like 3GPP and ETSI specify strict ACLR limits—typically -45 dBc or better for base stations—that must be met for certification. Excessive out-of-band emission causes adjacent channel interference, degrading capacity in neighboring cells.

A spectral mask defines the maximum allowable out-of-band emission as a function of frequency offset from the carrier. Unlike ACLR which measures integrated power, the mask specifies absolute or relative power limits at specific frequency offsets. CFR algorithms must be designed to suppress peaks while ensuring the resulting spectrum remains under the mask. Violations typically occur at the mask corners where emission limits tighten. Iterative clipping and filtering is often tuned specifically to meet mask requirements without over-constraining PAPR reduction.

Spectral regrowth is the physical mechanism behind out-of-band emission. When a band-limited signal passes through a nonlinear device—such as a clipper or power amplifier—the nonlinearity generates intermodulation products that spread energy beyond the original bandwidth. Third-order nonlinearities produce products at 2f₁ - f₂ and 2f₂ - f₁, falling directly into adjacent channels. Higher-order nonlinearities affect alternate channels. The sharper the clipping transition, the wider the spectral regrowth. Peak windowing smooths these transitions to confine regrowth.

The fundamental tension in CFR design: aggressive clipping reduces PAPR but generates severe out-of-band emission. Subsequent filtering suppresses the emission but causes peak regrowth, partially undoing the PAPR reduction. This iterative loop—clip, filter, measure—must be balanced. Key parameters include:

• Clipping ratio (CR): lower values increase both PAPR reduction and out-of-band emission
• Filter bandwidth: wider filters preserve more PAPR reduction but allow more emission
• Filter order: steeper filters better contain emission but introduce more latency

Unlike hard clipping which creates sharp signal discontinuities, peak windowing multiplies detected peaks by a smooth time-domain window—typically raised-cosine, Kaiser, or Gaussian functions. The window's spectral properties directly determine the out-of-band emission profile. Wider windows produce narrower spectral spreading but may overlap adjacent peaks. The window shape is a design parameter optimized for the specific spectral mask. Peak windowing achieves comparable PAPR reduction to clipping with significantly lower ACLR degradation.

Peak cancellation subtracts a pre-designed cancellation pulse from the signal at each detected peak. The cancellation pulse is engineered to have spectral confinement—its energy is concentrated within the transmit channel, minimizing out-of-band emission. The pulse shape is typically derived from the transmit filter impulse response or designed via optimization. Unlike clipping-and-filtering, peak cancellation does not require explicit filtering stages, though pulse overlap at high peak densities can cause cumulative distortion. Modern implementations use multi-stage cancellation with scaled pulses.