Channel Impairment Simulation

Simulates the stochastic amplitude and phase fluctuations caused by a signal traversing multiple reflective paths. This involves convolving the transmitted waveform with a time-varying impulse response.

• Rayleigh Fading: Models non-line-of-sight environments where no single path dominates, creating deep fades.
• Rician Fading: Models environments with a dominant line-of-sight component plus scattered paths, characterized by the K-factor.
• Power Delay Profile (PDP): Defines the relative power and delay of each resolvable multipath tap, often standardized by models like Tapped Delay Line (TDL) profiles from 3GPP.

Introduces frequency-domain distortion to mimic relative motion between transmitter and receiver. This is critical for training models for high-mobility platforms.

• Doppler Shift: A deterministic frequency offset proportional to radial velocity and carrier frequency.
• Doppler Spread: The spectral broadening caused by multiple paths arriving with different Doppler shifts, characterized by the maximum Doppler frequency (f_d).
• Jakes' Model: A widely used statistical model for generating a Rayleigh fading waveform with a specific Doppler spectrum, often used to simulate Clarke's spectrum for isotropic scattering.

Injects statistical noise to replicate the thermal agitation of electrons in receiver front-end components and external environmental interference.

• Additive White Gaussian Noise (AWGN): The fundamental model for thermal noise, characterized by a flat power spectral density and Gaussian amplitude distribution.
• Phase Noise: Short-term, random fluctuations in the phase of a signal caused by imperfections in local oscillators, critical for simulating realistic hardware impairments.
• Impulsive Noise: Non-Gaussian noise with high-amplitude, short-duration bursts, common in industrial and automotive environments, often modeled using Middleton Class A or Bernoulli-Gaussian distributions.

Augments signals with non-ideal characteristics of analog front-ends to bridge the sim-to-real gap. Models trained without these impairments often fail when deployed on real hardware.

• IQ Imbalance: Introduces gain and phase mismatches between the in-phase (I) and quadrature (Q) branches of a direct-conversion receiver, causing mirror-frequency interference.
• Power Amplifier Non-Linearity: Simulates amplitude-to-amplitude (AM-AM) and amplitude-to-phase (AM-PM) distortion using models like Rapp or Saleh to replicate spectral regrowth.
• Carrier Frequency Offset (CFO): Simulates the mismatch between transmitter and receiver local oscillator frequencies, causing a rotating constellation.

Models the large-scale attenuation of signal power over distance and the slow variations caused by obstructions in the environment.

• Path Loss: A deterministic, distance-dependent attenuation modeled by log-distance path loss models with a path loss exponent (n).
• Shadow Fading (Log-Normal Shadowing): A stochastic, large-scale variation modeled as a zero-mean Gaussian process in the dB domain, capturing the effect of buildings and terrain.
• Correlated Shadowing: Models the spatial correlation of shadow fading, where nearby locations experience similar attenuation, often using a Gudmundson correlation model.

Simulates the presence of other transmitters sharing the same or adjacent frequency resources, a dominant impairment in dense spectral environments.

• Co-Channel Interference (CCI): Injects signals from other users operating on the exact same frequency, often modeled with different modulation schemes and symbol rates.
• Adjacent Channel Interference (ACI): Simulates leakage from transmitters in neighboring frequency bands due to imperfect filtering and spectral regrowth from non-linear power amplifiers.
• Narrowband Jamming: Models intentional or unintentional interference as a high-power tone or swept-frequency signal within the receiver's bandwidth.

The algorithmic application of statistical models to replicate rapid amplitude and phase fluctuations caused by multipath propagation. This process applies Rayleigh or Rician distributions to baseband IQ samples to emulate environments with or without a dominant line-of-sight path.

• Rayleigh Fading: Models dense urban or indoor non-line-of-sight scenarios.
• Rician Fading: Models environments with a strong direct path, defined by a K-factor.
• Nakagami-m Fading: A generalized model that can emulate conditions more severe than Rayleigh.

The augmentation of RF signals with frequency offsets to mimic relative motion between transmitter and receiver. This is critical for training models for high-mobility deployments such as satellite links or vehicular communications.

• Introduces a frequency offset proportional to velocity and carrier frequency.
• Causes Doppler spread, a spectral broadening that induces inter-carrier interference in OFDM systems.
• Often modeled using a Jakes spectrum for isotropic scattering environments.

A characterization of a multipath channel describing received signal power as a function of time delay. It serves as the foundational parameter for generating realistic synthetic channel impulse responses.

• Defines the delay spread, which dictates the coherence bandwidth of the channel.
• Standardized profiles like ITU Vehicular A or EPA (Extended Pedestrian A) are used for reproducible testing.
• Directly influences the design of the cyclic prefix in OFDM waveforms to prevent inter-symbol interference.

The deliberate introduction of gain and phase mismatches between the in-phase (I) and quadrature (Q) branches of a signal. This trains models to be robust to hardware front-end imperfections in low-cost direct-conversion receivers.

• Gain Imbalance: Amplitude mismatch between I and Q branches, measured in dB.
• Phase Imbalance: Deviation from the ideal 90-degree orthogonality, measured in degrees.
• Results in an unwanted image signal that mirrors the desired spectrum, degrading the Error Vector Magnitude (EVM).

The fundamental noise model used to simulate thermal noise in the receiver. It adds a random process with a flat power spectral density and a Gaussian amplitude distribution to the clean signal.

• Defined by the Signal-to-Noise Ratio (SNR) in dB.
• Serves as the baseline impairment before adding more complex distortions like fading.
• Essential for benchmarking the fundamental sensitivity limits of a neural receiver.

The modeling of short-term, random frequency fluctuations in a local oscillator. This impairment causes a rotation of the constellation diagram and generates inter-carrier interference.

• Characterized by a power spectral density profile with slopes of -30, -20, -10, and 0 dB/decade.
• Critically degrades high-order QAM (e.g., 1024-QAM) and OFDM systems.
• Often simulated using a Wiener process or by filtering white noise through a specific phase noise mask.