This article started as an operations guide I wrote for factory engineers. While the concepts are familiar to someone with a decade of experience, explaining them clearly for readers who lack basic background requires unpacking many interrelated ideas. Below I list common RF metrics and summarize their practical meanings.
1. Rx Sensitivity
Receive sensitivity is one of the fundamental concepts. It represents the minimum signal level that a receiver can detect while meeting a specified error-rate criterion. Historically, circuit-switched systems used a general error-rate concept; in practice, BER (bit error rate) or PER (packet error rate) are often used to evaluate sensitivity. In LTE-era systems throughput is sometimes used to define sensitivity, since LTE lacks circuit-switched voice channels and user-perceived throughput became a more practical metric.
2. SNR (Signal-to-Noise Ratio)
SNR is often discussed alongside sensitivity. Demodulation SNR is defined as the SNR threshold at which a demodulator meets a specified error-rate. S stands for the useful signal and N for noise, which broadly denotes any undesired signal content. A commonly cited baseline is the natural noise floor of -174 dBm/Hz, a thermal-noise-derived power spectral density that is independent of system type but depends on temperature. Noise power is the noise power density integrated over the signal bandwidth, so wider signal bandwidths collect more noise power.
3. Tx Power
Transmit power matters because the transmitted signal undergoes path loss before reaching the receiver; higher transmit power generally enables longer link distances. However, whether the transmitter SNR matters at the receiver depends on the natural noise floor and path loss. For example, if path loss attenuates both signal and transmitter-generated noise by the same amount and the received transmitter-generated noise falls well below the ambient noise floor, the receiver noise is dominated by the ambient thermal noise rather than the transmitter's internal noise.
4. ACLR / ACPR
ACLR and ACPR measure parts of the transmitter noise that leak into adjacent channels, commonly called adjacent-channel leakage. ACLR and ACPR are essentially the same metric but are used in different contexts (for example, terminal vs. base station testing). They quantify interference by measuring leaked power into adjacent channel bandwidths, reflecting how a transmitter's leakage can interfere with receivers of similar systems operating on adjacent channels.
In LTE, ACLR measurements have two settings: EUTRA ACLR describes LTE-to-LTE interference, while UTRA ACLR considers LTE-to-UMTS interference. The measurement bandwidths match the victim system's occupied bandwidth, which makes ACLR/ACPR a paired definition with receiver adjacent-channel selectivity.
5. Modulation Spectrum / Switching Spectrum
In GSM systems, modulation spectrum and switching spectrum play roles similar to adjacent-channel leakage metrics. Their measurement bandwidths differ from the occupied bandwidth of GSM. Modulation spectrum evaluates interference between synchronous systems, while switching spectrum evaluates interference between asynchronous systems. In GSM, cells are asynchronous, so switching-spectrum components from rising/falling edges can fall into neighboring cell payloads; modulation spectrum captures interference when payloads overlap in time.
6. SEM (Spectrum Emission Mask)
SEM is a band-limited metric used to define a spectral template for in-band emissions. It differs from broader spurious-emission measurements, which emphasize out-of-band leakage and are often considered from an EMC perspective. SEM uses smaller measurement bandwidths (often 100 kHz to 1 MHz) to detect discrete spectral points that exceed limits within or near the operating band. SEM violations can occur even when ACLR is acceptable, for example when narrow spurs from LO or clock feed into the transmitter chain.
7. EVM (Error Vector Magnitude)
EVM is a vector quantity with magnitude and phase components; it measures the error between an actual signal point and the ideal constellation point. EVM effectively represents transmitter signal quality: larger deviations from ideal points correspond to larger EVM magnitude.
Transmit SNR is often not the primary concern because (1) transmitter SNR is usually much higher than the receiver demodulation SNR requirement, and (2) when computing receiver sensitivity we assume large path loss such that transmitter noise is below the ambient noise floor. However, transmitter-internal SNR matters for short-range wireless links, such as Wi?Fi (802.11) systems. With higher-order modulations like 256QAM in 802.11ac, high SNR and low EVM are required for reliable demodulation.
802.11 engineers commonly use EVM to assess transmitter linearity, while cellular engineers tend to focus on ACLR/ACPR and spectrum metrics for transmitter linearity. Cellular systems evolved with an emphasis on minimizing interference between cells, so adjacent-channel leakage metrics are central. Wi?Fi evolved for local-area packet data and often adapts modulation order to channel conditions; it tolerates interference differently and can trade rate for robustness.
In practice there are two main ways to improve received performance: lower modulation order to reduce demodulation threshold, or improve transmitter EVM to increase SNR. Because EVM degradation is often caused by nonlinearity (e.g., PA AM-AM distortion), EVM is commonly used as a transmitter linearity indicator.
7.1 Relationship Between EVM and ACPR/ACLR
Quantitatively relating EVM and ACPR/ACLR is difficult. From amplifier nonlinearity considerations, EVM and ACLR/ACPR are often positively correlated: AM-AM and AM-PM distortion increase EVM and are major sources of ACLR/ACPR. However, the correlation is not strict. For example, digital clipping at intermediate frequencies reduces peak-to-average ratio (PAR) and can improve ACLR/ACPR after the PA, but clipping damages waveform quality and worsens EVM.
7.2 Origins of PAR
PAR (peak-to-average ratio) is typically represented by a CCDF curve showing the probability of exceeding a given instantaneous power relative to the average. PAR significantly affects transmitter spectrum-regeneration metrics because high peaks can drive amplifiers into nonlinear regions, producing large spectral regrowth. Different air interfaces have different PAR characteristics: GMSK used in classic GSM has low envelope fluctuation and allowed pushing PAs toward P1dB for efficiency; 8PSK and modern OFDM-based systems have higher PAR and thus require more back-off or PAR mitigation techniques. LTE uplink SC-FDMA reduces PAR relative to OFDM to improve uplink PA efficiency.
8. Interference Metrics Summary
Interference metrics cover sensitivity under imposed interference beyond static sensitivity. Common interference tests include Blocking, Desense, and Channel Selectivity.
8.1 Blocking
Blocking is an RF metric describing the effect of a large unwanted signal injected into the receiver, often saturating an early-stage LNA and pushing amplifiers into nonlinearity or saturation. This reduces gain for the desired signal and generates strong intermodulation and distortion. Another blocking mechanism occurs when a large interferer triggers AGC to reduce gain, leaving the desired signal too small for proper demodulation.
Blocking tests can be in-band or out-of-band. In practice, blocking stimuli are often single-tone approximations of narrowband interferers. Mitigation focuses on improving receiver input IP3 and increasing dynamic range; for out-of-band blocking, front-end filtering attenuation is also important.
8.2 AM Suppression
AM suppression is a GSM-specific metric. It characterizes interference from neighboring cells with TDMA-like signals that are synchronous and have fixed delay. It reflects receiver tolerance to neighbor-cell interference typical in analog-era GSM deployments.
8.3 Adjacent (Alternative) Channel Suppression / Selectivity
Adjacent-channel selectivity measures receiver ability to tolerate leakage from neighboring-frequency cells. Transmitter spectral regrowth in adjacent frequencies can appear as in-band interference to a receiver of the same or similar system. Protocols and spectrum regulations pair transmitter leakage metrics (ACLR/ACPR) with receiver neighbor-channel selectivity to ensure coexistence.
8.4 Co-Channel Suppression (Selectivity)
Co-channel selectivity describes absolute same-frequency interference patterns, typically between two co-channel cells. Both signals may be valid but with differing strengths. Receiver co-channel selectivity indicates how well the receiver can avoid being overwhelmed when multiple valid signals are present.
Summary: Blocking is "large-signal interfering with small-signal" where RF design has some mitigation options. Metrics such as AM suppression and adjacent/co-channel selectivity represent "small-signal interference with large-signal" and are often addressed more effectively at the physical-layer protocol and algorithm level. Single-tone desense originates from specific coexistence scenarios where a narrow, in-band interferer (single tone) mixes with receiver LO phase noise or nonlinearity to produce baseband interference or intermodulation products inside the receive passband.
Historically, single-tone desense arose when CDMA systems coexisted with legacy analog systems in the same band; the legacy system's narrow tones could produce harmful in-band effects for the later system. Similar practical coexistence considerations have shaped frequency planning and band assignments.
Textbook explanations of blocking focus on gain compression, but two additional mechanisms are worth noting: (1) amplifier-induced spectral regrowth that creates harmonic components overlapping the desired band, and (2) cross-modulation where a large interferer modulates the amplifier behavior and distorts the small desired signal. Practical blocking vulnerability can therefore peak at certain interferer frequencies related to harmonics and intermodulation.
9. Dynamic Range, Temperature Compensation, and Power Control
Dynamic range, temperature compensation, and power control often become visible only under extreme tests, yet they embody subtle design choices in RF systems.
9.1 Transmitter Dynamic Range
Transmitter dynamic range specifies the maximum and minimum transmit power at which the transmitter meets other emission and linearity requirements. Maximum output must not degrade linearity metrics such as ACLR/ACPR/SEM and EVM, while minimum output must maintain adequate output SNR so that the transmitted signal is not submerged in transmitter internal noise. A practical failure mode is when reducing output power worsens ACLR because baseband noise or other internal noise sources dominate the measured adjacent-channel power at low output levels.
9.2 Receiver Dynamic Range
Receiver dynamic range relates to reference sensitivity at the low end and maximum tolerable input level at the high end. Maximum receive level is limited by distortion at any stage, from the front-end LNA to downstream ADC. Designers raise IIP3 for front-end stages and use AGC to keep signals within the dynamic range of subsequent stages. Some front-end LNAs implement multiple gain states with corresponding noise-figure tradeoffs: higher gain yields lower NF but lower input linearity; lower gain yields higher NF but higher IIP3. This technique helps maintain adequate SNR at the ADC while preserving linearity across varying input conditions.
9.3 Temperature Compensation
Transmitters are commonly temperature-compensated. Receiver gain and NF vary with temperature, but small-signal performance usually remains within system margins. For transmitters, temperature compensation can target output power accuracy or RF-chain gain stability. Closed-loop power control common in modern systems reduces the need for coarse temperature compensation of absolute output power, but RF-chain gain stability is still important for functions like digital predistortion (DPD) and precise DAC-level mapping. Typical temperature compensation methods include variable attenuators or variable-gain amplifiers controlled by temperature sensors and factory calibration.
9.4 Transmit Power Control
Transmit power control serves two primary goals: power conservation and interference suppression. For battery-powered devices, maintaining just enough uplink power for successful demodulation saves energy. For interference-limited systems like CDMA, tight and frequent power control balances received powers from different users to limit multiple-access interference. LTE uplink power control also mitigates mutual interference because multiple users share the uplink carrier. GSM implements coarser, stepwise control using power levels at roughly 1 dB steps.
Interference-limited systems, such as CDMA, have capacity constrained by interference rather than raw spectrum or time-slot resources. In contrast, systems like GSM are resource-limited in frequency and time domains, so their power-control requirements are less strict.
9.5 Power Control and RF Design Considerations
From an RF design perspective, closed-loop power control depends on reliable power-detection and feedback loops. One common practical issue is transmit power flatness across the band. Production calibrations typically adjust high/medium/low frequency points, but if transmitter gain is not flat across the band, closed-loop power control calibrated at a few points may produce errors at uncalibrated frequencies. Ensuring acceptable band-edge and in-band power flatness reduces closed-loop power errors in deployed operation.
