Kaira API Reference

Note

Kaira version 0.2.1 documentation. For older versions, please refer to the version selector above.

This documentation provides a comprehensive reference of Kaira’s components organized by functional category. Each component is documented with its parameters, methods, and usage examples.

Overview

Kaira is a modular toolkit for communication systems built on PyTorch. The library is organized into several key modules that handle different aspects of communication systems:

• Channels: Model transmission mediums with various noise and distortion characteristics

• Constraints: Enforce practical limitations on transmitted signals

• Metrics: Evaluate quality and performance of communication systems

• Models: Implement neural network architectures for encoding/decoding and end-to-end communication systems

• Modulations: Implement digital modulation schemes for wireless transmission

• Losses: Provide objective functions for training neural networks

• Utilities: Helper functions and tools for common operations

Base Components

Base classes define the fundamental interfaces that all implementations must adhere to. These abstract classes establish the contract that derived classes must fulfill.

channels.BaseChannel	Base abstract class for communication channel models.
constraints.BaseConstraint	Abstract foundation for implementing signal constraints in PyTorch-compatible format.
metrics.BaseMetric	Base Metric Module.
models.BaseModel	Abstract base class for all models in the Kaira framework.
modulations.BaseModulator	Abstract base class for all modulators.
modulations.BaseDemodulator	Abstract base class for all demodulators.
losses.BaseLoss	Base class for all Kaira loss functions.

Channels

Channel models for communication systems.

This package provides various channel models for simulating communication systems, including analog and digital channels, with support for various noise models, distortions, and fading patterns.

AWGNChannel	Additive white Gaussian noise (AWGN) channel for signal transmission.
BaseChannel	Base abstract class for communication channel models.
BinaryErasureChannel	Binary Erasure Channel (BEC) with symbol erasure probability.
BinarySymmetricChannel	Binary Symmetric Channel (BSC) with symmetric bit flip probability.
BinaryZChannel	Z Channel (asymmetric binary channel) with one-sided error probability.
ChannelRegistry	A registry for channels in Kaira.
FlatFadingChannel	Flat fading channel with configurable distribution and coherence time.
GaussianChannel	alias of AWGNChannel
IdealChannel	alias of PerfectChannel
IdentityChannel	alias of PerfectChannel
LambdaChannel	Customizable channel that applies user-defined functions to signals.
LaplacianChannel	Channel with additive Laplacian (double-exponential) noise.
LogNormalFadingChannel	Log-normal fading channel with configurable shadowing standard deviation.
NonlinearChannel	General nonlinear channel with configurable transfer function.
PerfectChannel	Identity channel that passes signals through unchanged.
PhaseNoiseChannel	Channel that introduces random phase noise.
PoissonChannel	Channel with signal-dependent Poisson noise.
RayleighFadingChannel	Specialized channel for Rayleigh fading in wireless communications.
RicianFadingChannel	Rician fading channel with configurable K-factor and coherence time.
UplinkMACChannel	Uplink Multiple Access Channel (MAC) for modeling multi-user uplink communications.

Constraints

Constraints module for Kaira.

This module contains various constraints that can be applied to transmitted signals in wireless communication systems. These constraints ensure signals meet practical requirements such as power limitations, hardware capabilities, and regulatory specifications.

Available constraint categories: - Base constraint definitions: Abstract base classes for all constraints - Power constraints: Control total power, average power, and PAPR - Antenna constraints: Manage power distribution across multiple antennas - Signal constraints: Handle amplitude limitations and spectral properties - Constraint composition: Combine multiple constraints sequentially

The module also provides factory functions for creating common constraint combinations and utilities for testing and validating constraint effectiveness.

Example:

>>> from kaira.constraints import TotalPowerConstraint, PAPRConstraint
>>> from kaira.constraints.utils import combine_constraints
>>>
>>> # Create individual constraints
>>> power_constr = TotalPowerConstraint(total_power=1.0)
>>> papr_constr = PAPRConstraint(max_papr=4.0)
>>>
>>> # Combine constraints into a single operation
>>> combined = combine_constraints([power_constr, papr_constr])
>>>
>>> # Apply to a signal
>>> constrained_signal = combined(input_signal)

AveragePowerConstraint	Scales signal to achieve specified average power per sample.
BaseConstraint	Abstract foundation for implementing signal constraints in PyTorch-compatible format.
CompositeConstraint	Applies multiple constraints in sequence.
ConstraintRegistry	A registry for constraints in Kaira.
IdentityConstraint	Identity constraint that returns the input signal unchanged.
LambdaConstraint	Constraint that applies a user-defined function to the signal.
PAPRConstraint	Reduces peak-to-average power ratio using soft clipping to minimize signal distortion.
PeakAmplitudeConstraint	Enforces maximum signal amplitude by clipping values that exceed threshold.
PerAntennaPowerConstraint	Distributes power budget across multiple antennas to ensure per-antenna power limits.
SpectralMaskConstraint	Restricts signal frequency components to comply with regulatory spectral masks.
TotalPowerConstraint	Normalizes signal to achieve exact total power regardless of input signal power.

Utils

Utility functions for constraints.

This module provides helper functions for creating, testing, validating, and working with constraints in wireless communication systems. These utilities streamline the process of configuring common constraint combinations and verifying constraint effectiveness.

apply_constraint_chain	Apply a list of constraints in sequence and optionally print debug info.
combine_constraints	Combine multiple constraints into a single constraint.
create_mimo_constraints	Create constraints commonly used in MIMO systems.
create_ofdm_constraints	Create constraints commonly used in OFDM systems.
measure_signal_properties	Measure common signal properties for a given tensor.
verify_constraint	Verify that a constraint produces the expected property in the output.

Metrics

Metrics module for Kaira.

This module contains various metrics for evaluating the performance of communication systems.

BaseMetric	Base Metric Module.
CompositeMetric	A metric that combines multiple metrics with optional weighting.
MetricRegistry	A registry for metrics in Kaira.

Image

Image metrics module.

This module contains metrics for evaluating image quality.

LPIPS	alias of LearnedPerceptualImagePatchSimilarity
LearnedPerceptualImagePatchSimilarity	Learned Perceptual Image Patch Similarity (LPIPS) Module.
MultiScaleSSIM	Multi-Scale Structural Similarity Index Measure (MS-SSIM) Module.
PSNR	alias of PeakSignalNoiseRatio
PeakSignalNoiseRatio	Peak Signal-to-Noise Ratio (PSNR) Module.
SSIM	alias of StructuralSimilarityIndexMeasure
StructuralSimilarityIndexMeasure	Structural Similarity Index Measure (SSIM) Module.

Signal

Signal metrics module.

This module contains metrics for evaluating signal processing performance.

BER	alias of BitErrorRate
BLER	alias of BlockErrorRate
BitErrorRate	Bit Error Rate (BER) metric.
BlockErrorRate	Block Error Rate (BLER) metric.
EVM	alias of ErrorVectorMagnitude
ErrorVectorMagnitude	Error Vector Magnitude (EVM) metric.
FER	alias of BlockErrorRate
FrameErrorRate	alias of BlockErrorRate
SER	alias of BlockErrorRate
SNR	alias of SignalToNoiseRatio
SignalToNoiseRatio	Signal-to-Noise Ratio (SNR) metric.
SymbolErrorRate	alias of BlockErrorRate

Models

Models module for Kaira.

BaseModel	Abstract base class for all models in the Kaira framework.
ChannelCodeModel	A specialized model for Channel Code.
ConfigurableModel	Model that supports dynamically adding and removing steps.
DeepJSCCModel	Deep Joint Source-Channel Coding model.
FeedbackChannelModel	A model that models communication with a feedback channel.
ModelRegistry	A registry for models in Kaira.
MultipleAccessChannelModel	A model simulating a Multiple Access Channel (MAC).
WynerZivModel	A model for Wyner-Ziv coding with decoder side information.

Soft Bit Thresholding

Soft bit thresholding module for binary data processing.

This module provides various thresholding techniques for converting soft bit representations (probabilities, LLRs, etc.) to hard decisions. These thresholders can be used with soft decoders or as standalone components in signal processing pipelines.

Soft bit processing is crucial in modern communication systems to extract maximum information from the received signals. The techniques implemented here are based on established methods in communication theory.

AdaptiveThresholder	Adaptive thresholder for soft bit values.
DynamicThresholder	Thresholder with dynamically adjusting threshold for non-stationary signals.
FixedThresholder	Simple fixed threshold for soft bit values.
HysteresisThresholder	Thresholder with hysteresis for robust decision making in noisy environments.
InputType	Enumeration of supported input types for soft bit thresholders.
LLRThresholder	Specialized thresholder for Log-Likelihood Ratio (LLR) values.
MinDistanceThresholder	Thresholder based on minimum distance calculations.
OutputType	Enumeration of supported output types for soft bit thresholders.
RepetitionSoftBitDecoder	Enhanced decoder for repetition coding with flexible soft bit processing.
SoftBitEnsembleThresholder	Ensemble thresholder that combines decisions from multiple thresholders.
SoftBitThresholder	Base class for soft bit thresholding techniques.
WeightedThresholder	Thresholder that applies weights to input values before thresholding.

Components

Components module for Kaira models.

AFModule	AFModule: Attention-Feature Module [Xu et al., 2021].
ConvDecoder	Convolutional Neural Network (CNN) Decoder for image transmission systems.
ConvEncoder	Convolutional Neural Network (CNN) Encoder for image transmission systems.
MLPDecoder	Multi-Layer Perceptron (MLP) Decoder for communication systems.
MLPEncoder	Multi-Layer Perceptron (MLP) Encoder for communication systems.
Projection	Projection layer for dimensionality reduction in communication systems [Yilmaz et al., 2025, Yilmaz et al., 2025].
ProjectionType	Enum for different types of projections.

Decoders

Forward Error Correction (FEC) decoders for Kaira.

This module provides various decoder implementations for forward error correction codes. The decoders in this module are designed to work seamlessly with the corresponding encoders from the kaira.models.fec.encoders module.

Decoders

• BlockDecoder: Base class for all block code decoders

• SyndromeLookupDecoder: Decoder using syndrome lookup tables for efficient error correction

• BerlekampMasseyDecoder: Implementation of Berlekamp-Massey algorithm for decoding BCH and Reed-Solomon codes

• ReedMullerDecoder: Implementation of Reed-Muller decoding algorithm for Reed-Muller codes

• WagnerSoftDecisionDecoder: Implementation of Wagner’s soft-decision decoder for single-parity check codes

• BruteForceMLDecoder: Maximum likelihood decoder that searches through all possible codewords

• BeliefPropagationDecoder: Implementation of belief propagation algorithm [Kschischang et al., 2002] for decoding LDPC codes

• MinSumLDPCDecoder: Min-Sum decoder [Chen et al., 2005] for LDPC codes with reduced computational complexity

These decoders can be used to recover original messages from possibly corrupted codewords that have been transmitted over noisy channels. Each decoder has specific strengths and is optimized for particular types of codes or error patterns.

Examples

>>> from kaira.models.fec.encoders import BCHCodeEncoder
>>> from kaira.models.fec.decoders import BerlekampMasseyDecoder
>>> encoder = BCHCodeEncoder(15, 7)
>>> decoder = BerlekampMasseyDecoder(encoder)
>>> # Example decoding
>>> received = torch.tensor([1, 0, 1, 1, 0, 1, 0, 0, 1, 1, 0, 0, 1, 0, 1])
>>> decoded = decoder(received)

BaseBlockDecoder	Base class for block code decoders.
BeliefPropagationDecoder	Belief propagation decoder for linear block codes and LDPC codes.
BeliefPropagationPolarDecoder	Decoder for Polar code using Belief Propagation (BP) method.
BerlekampMasseyDecoder	Berlekamp-Massey decoder for BCH and Reed-Solomon codes.
BruteForceMLDecoder	Brute Force Maximum Likelihood decoder for block codes.
MinSumLDPCDecoder	Min-Sum decoder for LDPC codes [Chen et al., 2005].
ReedMullerDecoder	Reed-Muller decoder using majority-logic decoding.
SuccessiveCancellationDecoder	Decoder for Polar code using Successive Cancellation (SC) method [Arıkan, 2009].
SyndromeLookupDecoder	Syndrome lookup decoder for linear block codes.
WagnerSoftDecisionDecoder	Wagner soft-decision decoder for single parity-check codes.

Encoders

Forward Error Correction encoders for Kaira.

This module provides various encoder implementations for forward error correction, including: - Block codes: Fundamental error correction codes that operate on fixed-size blocks - Linear block codes: Codes with linear algebraic structure allowing matrix operations - LDPC codes: Low-Density Parity-Check codes with sparse parity-check matrices - Cyclic codes: Special class of linear codes with cyclic shift properties - BCH codes: Powerful algebraic codes with precise error-correction capabilities - Reed-Solomon codes: Widely-used subset of BCH codes for burst error correction - Hamming codes: Simple single-error-correcting codes with efficient implementation - Repetition codes: Basic codes that repeat each bit multiple times - Golay codes: Perfect codes with specific error correction properties - Single parity-check codes: Simple error detection through parity bit addition

These encoders can be used to add redundancy to data for enabling error detection and correction in communication systems, storage devices, and other applications requiring reliable data transmission over noisy channels [Lin and Costello, 2004, Moon, 2005].

BCHCodeEncoder	Encoder for BCH (Bose–Chaudhuri–Hocquenghem) codes.
BaseBlockCodeEncoder	Base class for block code encoders.
CyclicCodeEncoder	Encoder for cyclic codes.
GolayCodeEncoder	Encoder for binary Golay codes.
HammingCodeEncoder	Encoder for Hamming codes.
LDPCCodeEncoder	Encoder for LDPC code [Gallager, 1962], [Gallager, 1963].
LinearBlockCodeEncoder	Encoder for linear block coding.
PolarCodeEncoder	Encoder for Polar code [Arıkan, 2008].
ReedMullerCodeEncoder	Encoder for Reed-Muller codes.
ReedSolomonCodeEncoder	Encoder for Reed-Solomon (RS) codes.
RepetitionCodeEncoder	Encoder for repetition coding that extends LinearBlockCodeEncoder.
SingleParityCheckCodeEncoder	Encoder for single parity-check codes.
SystematicLinearBlockCodeEncoder	Encoder for systematic linear block coding.

Generic

Generic model implementations for Kaira.

This module provides generic model implementations that can be used as building blocks for more complex models, such as sequential, parallel, and branching models.

BranchingModel	Model that routes inputs through different paths based on conditions.
IdentityModel	Identity Model.
LambdaModel	Lambda Model.
ParallelModel	A model that processes steps in parallel.
SequentialModel	A model that processes steps sequentially.

Image

Image model implementations for Kaira.

This module provides models specifically designed for image data transmission.

Bourtsoulatze2019DeepJSCCDecoder	DeepJSCC decoder model from [Bourtsoulatze et al., 2019].
Bourtsoulatze2019DeepJSCCEncoder	DeepJSCC encoder model from [Bourtsoulatze et al., 2019].
DeepJSCCFeedbackDecoder	Decoder network for DeepJSCC with Feedback [Kurka and Gündüz, 2020].
DeepJSCCFeedbackEncoder	Encoder network for DeepJSCC with Feedback [Kurka and Gündüz, 2020].
DeepJSCCFeedbackModel	Deep Joint Source-Channel Coding with Feedback implementation [Kurka and Gündüz, 2020].
Tung2022DeepJSCCQ2Decoder	DeepJSCCQ2 Decoder Module [Tung et al., 2022].
Tung2022DeepJSCCQ2Encoder	DeepJSCCQ2 Encoder Module [Tung et al., 2022].
Tung2022DeepJSCCQDecoder	DeepJSCCQ Decoder Module [Tung et al., 2022].
Tung2022DeepJSCCQEncoder	DeepJSCCQ Encoder Module [Tung et al., 2022].
Xie2023DTDeepJSCCDecoder	Discrete Task-Oriented Deep JSCC decoder.
Xie2023DTDeepJSCCEncoder	Discrete Task-Oriented Deep JSCC encoder.
Yilmaz2023DeepJSCCNOMADecoder	DeepJSCC-NOMA Decoder Module [Yilmaz et al., 2023].
Yilmaz2023DeepJSCCNOMAEncoder	DeepJSCC-NOMA Encoder Module [Yilmaz et al., 2023].
Yilmaz2023DeepJSCCNOMAModel	Distributed Deep Joint Source-Channel Coding over a Multiple Access Channel [Yilmaz et al., 2023].
Yilmaz2024DeepJSCCWZConditionalDecoder	DeepJSCC-WZ Conditional Decoder Module [Yilmaz et al., 2024].
Yilmaz2024DeepJSCCWZConditionalEncoder	DeepJSCC-WZ Conditional Encoder Module [Yilmaz et al., 2024].
Yilmaz2024DeepJSCCWZDecoder	DeepJSCC-WZ Decoder Module [Yilmaz et al., 2024].
Yilmaz2024DeepJSCCWZEncoder	DeepJSCC-WZ Encoder Module [Yilmaz et al., 2024].
Yilmaz2024DeepJSCCWZModel	A specialized Wyner-Ziv model for neural joint source-channel coding with side information.
Yilmaz2024DeepJSCCWZSmallDecoder	DeepJSCC-WZ-sm Decoder Module [Yilmaz et al., 2024].
Yilmaz2024DeepJSCCWZSmallEncoder	DeepJSCC-WZ-sm Encoder Module [Yilmaz et al., 2024].

Compressors

Image compressor models, including standard and neural network-based methods.

BPGCompressor	BPG (Better Portable Graphics) image compression based on bpgenc and bpgdec.
NeuralCompressor	Neural network-based image compression model.

Modulations

Digital modulation schemes for wireless communications.

This package provides implementations of common digital modulation and demodulation techniques used in modern communication systems, including PSK, QAM, PAM, and differential modulation schemes.

BPSKDemodulator	Binary Phase-Shift Keying (BPSK) demodulator.
BPSKModulator	Binary Phase-Shift Keying (BPSK) modulator.
BaseDemodulator	Abstract base class for all demodulators.
BaseModulator	Abstract base class for all modulators.
DBPSKDemodulator	Differential Binary Phase-Shift Keying (DBPSK) demodulator.
DBPSKModulator	Differential Binary Phase-Shift Keying (DBPSK) modulator.
DPSKDemodulator	Differential Phase-Shift Keying (DPSK) demodulator.
DPSKModulator	Differential Phase-Shift Keying (DPSK) modulator.
DQPSKDemodulator	Differential Quadrature Phase-Shift Keying (DQPSK) demodulator.
DQPSKModulator	Differential Quadrature Phase-Shift Keying (DQPSK) modulator.
IdentityDemodulator	Identity demodulator that passes input data through unchanged.
IdentityModulator	Identity modulator that passes input data through unchanged.
ModulationRegistry	A registry for modulations in Kaira.
OQPSKDemodulator	Offset Quadrature Phase-Shift Keying (OQPSK) demodulator.
OQPSKModulator	Offset Quadrature Phase-Shift Keying (OQPSK) modulator.
PAMDemodulator	Pulse Amplitude Modulation (PAM) demodulator.
PAMModulator	Pulse Amplitude Modulation (PAM) modulator.
PSKDemodulator	General M-ary Phase-Shift Keying (PSK) demodulator.
PSKModulator	General M-ary Phase-Shift Keying (PSK) modulator.
Pi4QPSKDemodulator	Π/4-QPSK demodulator.
Pi4QPSKModulator	Π/4-QPSK (π/4 shifted QPSK) modulator.
QAMDemodulator	Quadrature Amplitude Modulation (QAM) demodulator.
QAMModulator	Quadrature Amplitude Modulation (QAM) modulator.
QPSKDemodulator	Quadrature Phase-Shift Keying (QPSK) demodulator.
QPSKModulator	Quadrature Phase-Shift Keying (QPSK) modulator.

Utils

Utility functions for digital modulation schemes.

binary_array_to_gray	Convert binary array to Gray code.
binary_to_gray	Convert binary number to Gray code.
calculate_spectral_efficiency	Calculate spectral efficiency of a modulation scheme in bits/s/Hz.
calculate_theoretical_ber	Calculate theoretical Bit Error Rate (BER) for common modulations.
gray_array_to_binary	Convert Gray-coded array to binary.
gray_to_binary	Convert Gray code to binary number.
plot_constellation	Plot a constellation diagram.

Losses

Kaira Losses Package.

This package provides various loss functions for different modalities.

BaseLoss	Base class for all Kaira loss functions.
CompositeLoss	A loss that combines multiple loss functions with optional weighting.
LossRegistry	A registry for loss functions in Kaira.

Adversarial

Adversarial Losses module for Kaira.

This module contains various adversarial loss functions for GAN-based training.

FeatureMatchingLoss	Feature Matching Loss Module for GANs.
HingeLoss	Hinge Loss Module for GANs.
LSGANLoss	Least Squares GAN Loss Module.
R1GradientPenalty	R1 Gradient Penalty Module for GANs.
VanillaGANLoss	Vanilla GAN Loss Module.
WassersteinGANLoss	Wasserstein GAN Loss Module.

Audio

Audio Losses module for Kaira.

This module contains various loss functions for training audio-based communication systems.

AudioContrastiveLoss	Audio Contrastive Loss Module.
FeatureMatchingLoss	Feature Matching Loss Module.
L1AudioLoss	L1 Audio Loss Module.
LogSTFTMagnitudeLoss	Log STFT Magnitude Loss Module.
MelSpectrogramLoss	Mel-Spectrogram Loss Module.
MultiResolutionSTFTLoss	Multi-Resolution STFT Loss Module.
STFTLoss	STFT Loss Module.
SpectralConvergenceLoss	Spectral Convergence Loss Module.

Image

Losses module for Kaira.

This module contains various loss functions for training communication systems, including MSE loss, LPIPS loss, and SSIM loss. These loss functions are widely used in image processing and computer vision tasks [Wang and Bovik, 2009] [Zhang et al., 2018].

CombinedLoss	Combined Loss Module.
FocalLoss	Focal Loss Module for dealing with class imbalance.
GradientLoss	Gradient Loss Module.
L1Loss	L1 (Mean Absolute Error) Loss Module.
LPIPSLoss	Learned Perceptual Image Patch Similarity (LPIPS) Loss Module.
MSELPIPSLoss	MSELPIPSLoss Module.
MSELoss	Mean Squared Error (MSE) Loss Module.
MSSSIMLoss	Multi-Scale Structural Similarity Index Measure (MS-SSIM) Loss Module.
PSNRLoss	Peak Signal-to-Noise Ratio (PSNR) Loss Module.
SSIMLoss	Structural Similarity Index Measure (SSIM) Loss Module.
StyleLoss	Style Loss Module based on Gram matrices.
TotalVariationLoss	Total Variation Loss Module.
VGGLoss	VGG Perceptual Loss Module.

Multimodal

Multimodal Losses module for Kaira.

This module contains various loss functions for training multimodal systems.

AlignmentLoss	Alignment Loss for multimodal embeddings.
CMCLoss	Cross-Modal Consistency Loss Module.
ContrastiveLoss	Contrastive Loss Module.
InfoNCELoss	InfoNCE Loss Module for multimodal contrastive learning.
TripletLoss	Triplet Loss Module for multimodal data.

Text

Text Losses module for Kaira.

This module contains various loss functions for training text-based systems.

CosineSimilarityLoss	Cosine Similarity Loss Module.
CrossEntropyLoss	Cross Entropy Loss Module.
LabelSmoothingLoss	Label Smoothing Loss Module.
Word2VecLoss	Word2Vec Loss Module.

Data

Data utilities for Kaira, including data generation and correlation models.

BinaryTensorDataset	Dataset of randomly generated binary tensors.
UniformTensorDataset	Dataset of uniformly distributed random tensors.
WynerZivCorrelationDataset	Dataset for Wyner-Ziv coding scenarios with correlated sources.

create_binary_tensor	Create a random binary tensor with specified probability of 1s.
create_uniform_tensor	Create a tensor with uniformly distributed random values.
load_sample_images	Load sample images from popular datasets for demonstrations.

Utils

General utility functions for the Kaira library.

PlottingUtils

A comprehensive plotting utility class with static methods for visualization.

add_noise_for_snr	Add Gaussian noise to achieve a target Signal-to-Noise Ratio.
calculate_num_filters_factor_image	Calculate the number of filters in an image based on the number of strided layers and bandwidth ratio.
calculate_snr	Calculate the SNR between original and noisy signals.
estimate_signal_power	Estimate the power of a signal.
noise_power_to_snr	Calculate SNR in dB given signal and noise power.
snr_db_to_linear	Convert Signal-to-Noise Ratio from decibel to linear scale.
snr_linear_to_db	Convert Signal-to-Noise Ratio from linear scale to decibels.
snr_to_noise_power	Convert SNR in dB to noise power given a signal power.
to_tensor	Convert an input data into a torch.Tensor, with an option to move it to a specific device.

Snr

Utility functions for Signal-to-Noise Ratio (SNR) calculations and conversions.

add_noise_for_snr	Add Gaussian noise to achieve a target Signal-to-Noise Ratio.
calculate_snr	Calculate the SNR between original and noisy signals.
estimate_signal_power	Estimate the power of a signal.
noise_power_to_snr	Calculate SNR in dB given signal and noise power.
snr_db_to_linear	Convert Signal-to-Noise Ratio from decibel to linear scale.
snr_linear_to_db	Convert Signal-to-Noise Ratio from linear scale to decibels.
snr_to_noise_power	Convert SNR in dB to noise power given a signal power.

Benchmarks

Kaira Benchmarking System.

This module provides standardized benchmarks for evaluating communication system components and deep learning models in Kaira.

BaseBenchmark	Base class for all benchmarks.
BenchmarkConfig	Configuration for benchmark execution.
BenchmarkRegistry	Registry for managing benchmark classes and instances.
BenchmarkResult	Container for benchmark results.
BenchmarkResultsManager	Manages benchmark results with improved directory structure and organization.
BenchmarkSuite	Collection of benchmarks that can be run together.
BenchmarkVisualizer	Visualizer for benchmark results.
ComparisonRunner	Runner for comparing multiple benchmarks on the same task.
ParallelRunner	Parallel benchmark runner using thread pool.
ParametricRunner	Runner for sweeping parameters across benchmarks.
StandardMetrics	Collection of standard metrics for communication system evaluation.
StandardRunner	Standard sequential benchmark runner.

create_benchmark	Create an instance of a registered benchmark.
get_benchmark	Get a registered benchmark class.
get_config	Get a predefined configuration.
list_benchmarks	List all available benchmark names.
list_configs	List available predefined configurations.
register_benchmark	Decorator to register a benchmark class.