Signal Processing Glossary of Terms

Adaptive Filters

Adaptive filters are digital filters that automatically adjust their coefficients in real-time to optimize performance based on changing input conditions. They are essential for applications where signal statistics are unknown or time-varying, such as noise cancellation, echo removal, and channel equalization.

Example: Students implement adaptive filters using the LMS algorithm to cancel acoustic echo in a speakerphone simulation, observing how the filter converges to the optimal solution.

Adaptive filtering

Adaptive filtering is a type of digital filter that self-adjusts its parameters based on the input signal to minimize a certain error criterion. It is used in environments where signal characteristics change over time.

Example: In the course, students implement an adaptive filter to cancel noise in real-time audio recordings by continuously adjusting the filter coefficients to match the changing noise profile.

Aliasing

Aliasing is a phenomenon that occurs when a signal is sampled at a rate below twice its highest frequency component, causing high-frequency components to appear as lower frequencies in the sampled signal. It results in distortion that cannot be corrected after sampling.

Example: Students observe aliasing by sampling a high-frequency sine wave at an insufficient rate, watching the sampled signal appear as a lower frequency than the original.

Amplitude Modulation (AM)

Amplitude Modulation is a modulation technique where the amplitude of a carrier wave is varied in proportion to the instantaneous amplitude of the input signal. It is widely used in radio broadcasting.

Example: Students explore AM by transmitting audio signals over simulated radio waves, observing how changes in amplitude affect signal transmission and reception.

Analog Signals

Analog signals are continuous signals that vary smoothly over time and can take any value within a continuous range. They represent physical quantities such as sound, light, temperature, or pressure in their natural continuous form.

Example: Students analyze analog audio signals from a microphone, observing how the continuous waveform represents the varying air pressure of sound waves before digital conversion.

Autoencoders

Autoencoders are a type of artificial neural network used to learn efficient codings of input data by compressing and then reconstructing the input. They are commonly used for dimensionality reduction and feature learning.

Example: In assignments, students use autoencoders to compress image data, demonstrating how the network can learn to retain essential features while reducing data size.

Autocorrelation

Autocorrelation measures the similarity between a signal and a time-shifted version of itself, quantifying how correlated the signal is with its own past values. It is used to detect repeating patterns, periodicity, and to analyze signal characteristics.

Example: Students compute the autocorrelation of a speech signal to identify its fundamental frequency and detect periodic components like pitch in voiced speech.

Band-pass filter (BPF)

A Band-pass filter allows frequencies within a specific range to pass through while attenuating frequencies outside that range. It is used to isolate desired frequency components in a signal.

Example: Students design a BPF to isolate specific frequency bands in an EEG signal, enabling the analysis of particular brain wave activities.

Band-stop filter (BSF)

A Band-stop filter attenuates frequencies within a specific range while allowing frequencies outside that range to pass. It is used to eliminate unwanted frequency components from a signal.

Example: In lab sessions, students apply a BSF to remove power line interference from a biomedical signal, improving the clarity of the data.

Bessel filter

A Bessel filter is a type of analog or digital filter with a maximally flat group delay, ensuring minimal signal distortion in the time domain. It is ideal for applications requiring linear phase response.

Example: Students compare different filter types and observe how Bessel filters preserve the waveform shape of transient signals compared to other filters.

Big data analytics

Big data analytics involves examining large and complex data sets to uncover hidden patterns, correlations, and other insights. It leverages advanced computational techniques to process and analyze data efficiently.

Example: In projects, students use big data analytics to process large datasets from sensor networks, extracting meaningful trends and patterns relevant to signal processing applications.

Bilinear transform

The Bilinear transform is a mathematical technique used to convert analog filter designs into digital filters while preserving the frequency response characteristics. It maps the s-plane to the z-plane in filter design.

Example: Students apply the Bilinear transform to convert an analog prototype filter into a digital filter, ensuring the digital filter maintains the desired frequency response.

Butterworth filter

A Butterworth filter is a type of analog or digital filter with a maximally flat frequency response in the passband, ensuring no ripples. It provides a smooth transition from passband to stopband.

Example: Students design Butterworth filters and compare their frequency responses to other filter types, analyzing the trade-offs between smoothness and transition sharpness.

Calculus

Calculus is a branch of mathematics focused on limits, functions, derivatives, integrals, and infinite series. It is fundamental for analyzing and modeling dynamic systems in signal processing.

Example: Students apply calculus to derive the continuous-time Fourier transform, understanding how differentiation and integration affect signal representations.

Causality

Causality refers to a property of systems where the output at any time depends only on past and present inputs, not on future inputs. It is essential for real-time signal processing applications.

Example: In system analysis, students ensure that their designed filters are causal to guarantee that they can be implemented in real-time processing scenarios.

Channel coding

Channel coding involves adding redundancy to transmitted information to protect against errors during transmission. It is essential for reliable communication in noisy channels.

Example: Students implement error-correcting codes like Hamming codes to improve data integrity in simulated digital communication systems.

Chebyshev filter

A Chebyshev filter is a type of analog or digital filter characterized by a steeper roll-off and ripple in either the passband or stopband. It offers a sharper transition between passband and stopband compared to Butterworth filters.

Example: Students design Chebyshev filters and analyze the trade-offs between passband ripple and filter sharpness in different signal processing applications.

CircuiTekz

An extension to LaTeX that allows circuits to be added to LaTeX drawing using the Tekz drawing system.

Example: Generative AI tools can be used to generate Circuit

Classification

Classification is a machine learning task where the goal is to assign input data into predefined categories based on learned patterns. It is widely used in signal processing for tasks like speech recognition and image classification.

Example: In a project, students develop a classifier to categorize audio signals into different genres based on their spectral features extracted using FFT.

Cognitive signal processing

Cognitive signal processing integrates principles from cognitive science to develop intelligent systems that can adapt and learn from their environment. It focuses on creating signal processing techniques that mimic human cognitive abilities.

Example: Students explore cognitive signal processing by designing systems that can adaptively recognize and respond to changing signal patterns in real-time applications.

Colored noise

Colored noise refers to noise signals with a power spectral density that varies with frequency, unlike white noise which has a constant power spectral density. Examples include pink noise and brown noise.

Example: Students simulate colored noise in their signal processing experiments to study its impact on filter performance and signal detection algorithms.

Complex numbers

Complex numbers are numbers that have both a real and an imaginary component, typically expressed in the form \(a + bi\). They are essential in signal processing for representing and analyzing oscillatory signals and transformations.

Example: Students use complex numbers to represent phasors in AC circuit analysis, facilitating the calculation of impedances and signal interactions.

Compressed Sensing

Compressed sensing is a signal processing technique that enables the reconstruction of sparse signals from far fewer samples than traditional methods require. It exploits signal sparsity in some transform domain to recover signals with high fidelity from undersampled measurements.

Example: Students apply compressed sensing algorithms to reconstruct MRI images from reduced scan data, demonstrating how sparsity assumptions enable faster medical imaging.

Continuous Wavelet Transform (CWT)

The Continuous Wavelet Transform is a signal processing technique that decomposes a signal into wavelets, providing both time and frequency information. It is useful for analyzing non-stationary signals.

Example: Students use CWT to analyze transient features in biomedical signals, such as detecting spikes in EEG data.

Continuous-time signals

Continuous-time signals are defined for every instant of time and can take any value within a range. They are represented mathematically as functions of a continuous variable.

Example: In lectures, students work with continuous-time signals like sine waves and analog audio signals to understand fundamental signal properties before discretization.

Convolution

Convolution is a mathematical operation that combines two signals to produce a third signal, representing the amount of overlap between one signal as it is shifted over another. It is fundamental in system analysis and filter design.

Example: In assignments, students perform convolution of a signal with an impulse response to determine the output of a linear time-invariant (LTI) system.

Convolution Theorem

The convolution theorem states that convolution in the time domain is equivalent to multiplication in the frequency domain, and vice versa. This fundamental property enables efficient computation of convolutions using the Fast Fourier Transform.

Example: Students verify the convolution theorem by computing a convolution both directly and via FFT multiplication, comparing the results to understand the computational efficiency gains.

Convolutional Neural Networks (CNNs)

Convolutional Neural Networks are deep learning architectures that use convolutional layers to automatically learn spatial hierarchies of features from input data. They are particularly effective for image and signal processing tasks.

Example: Students implement a CNN to classify spectrograms of audio signals, learning how convolutional layers extract frequency and temporal patterns for sound recognition.

Correlation

Correlation is a statistical measure that quantifies the strength and direction of the relationship between two signals or variables. It indicates how one signal changes in relation to another and is used for pattern matching and signal analysis.

Example: Students calculate the correlation between two sensor signals to determine if they are measuring related phenomena, such as temperature and humidity in environmental monitoring.

Cross-correlation

Cross-correlation measures the similarity between two signals as a function of the time lag applied to one of them. It is used for signal alignment and pattern detection.

Example: Students use cross-correlation to identify the time delay between two sensor signals, aiding in applications like radar and sonar.

Cursor IDE

An agentic integrated development environment that is ideal for creating and maintaining intelligent textbooks.

Cursor immediately creates embeddings for all the files in your project so when you make a request it knows how to proceed with changes. A request such as "Create a new MicroSim" will automatically create new directories and files that conform to your rules.

Declarative

An abstract form of representation that states the intent of what is to be done, but not the precise details of how it should be done.

Example: Schemadraw allows us to place electrical components relative to each other using terms like left, right, above and below without having to specify the exact x and y coordinates of each component in a circuit diagram.

Deep Learning (DL)

Deep Learning is a subset of machine learning that uses neural networks with many layers to model complex patterns in data. It is applied in signal processing for tasks like image and speech recognition.

Example: In projects, students implement deep learning models to classify speech signals into different spoken words based on their spectral features.

Differential equations

Differential equations are mathematical equations involving derivatives of functions. They are used to model the behavior of dynamic systems in signal processing.

Example: Students derive the differential equation governing an RLC circuit and analyze its response to different input signals.

Differentiation

Differentiation is a calculus operation that computes the derivative of a function, representing the rate of change of the function with respect to a variable. It is used in signal analysis to determine signal slopes and rates.

Example: In lab exercises, students differentiate a signal to find its velocity from position data, applying numerical differentiation techniques.

Digital Signal Processing (DSP)

Digital Signal Processing involves the manipulation of digital signals using computational algorithms. It encompasses filtering, analysis, compression, and transformation of signals.

Example: Students implement digital filters in MATLAB to process and enhance audio signals, applying concepts learned in lectures to practical scenarios.

Digital modulation

Digital modulation involves varying one or more properties of a carrier signal (such as amplitude, frequency, or phase) to encode digital information. It is fundamental in digital communication systems.

Example: Students implement various digital modulation schemes like QAM and PSK in simulated communication systems, analyzing their performance under different noise conditions.

Digital signals

Digital signals are discrete-time signals that have quantized amplitude levels. They are essential for digital communication and processing systems.

Example: In assignments, students convert analog audio signals into digital form through sampling and quantization, then process them using digital filters.

Digital Signal Processors (DSPs)

Digital Signal Processors are specialized microprocessors designed specifically for performing digital signal processing operations efficiently. They feature architectures optimized for repetitive mathematical operations like multiply-accumulate, making them ideal for real-time signal processing applications.

Example: Students program a DSP to implement a real-time audio filter, learning how the specialized hardware enables low-latency processing that general-purpose processors cannot achieve.

Discrete Cosine Transform

The Discrete Cosine Transform (DCT) is a transform that expresses a signal as a sum of cosine functions oscillating at different frequencies. Unlike the DFT, it uses only real numbers and is widely used in image and audio compression algorithms like JPEG and MP3.

Example: Students apply the DCT to image blocks in a JPEG compression exercise, observing how most image energy concentrates in low-frequency coefficients enabling efficient compression.

Discrete Fourier Transform (DFT)

The Discrete Fourier Transform is a mathematical transform that converts a finite sequence of time-domain samples into a sequence of frequency-domain components. It is fundamental for digital signal processing.

Example: Students compute the DFT of a sampled audio signal to analyze its frequency content and identify dominant frequencies.

Discrete Wavelet Transform (DWT)

The Discrete Wavelet Transform is a wavelet-based transform that analyzes discrete signals at different frequency bands with different resolutions. It is useful for signal compression and denoising.

Example: Students apply DWT to compress image data, observing how different wavelet levels capture various image features.

Discrete-time signals

Discrete-time signals are defined only at discrete time intervals and are typically obtained by sampling continuous-time signals. They are essential for digital signal processing.

Example: In assignments, students sample a continuous-time sine wave to create a discrete-time signal for further digital analysis using FFT.

Edge detection

Edge detection is an image processing technique used to identify and locate sharp discontinuities in intensity within an image. It is crucial for feature extraction and image analysis.

Example: In projects, students apply edge detection algorithms like the Sobel filter to identify boundaries in grayscale images, facilitating object recognition tasks.

Eclipse Layout Kernel

A collection of javascript layout algorithms, and an infrastructure that bridges the gap between layout algorithms and diagram viewers and editors.

We can use tools Eclipse Layout Kernel (ELK) to automate the layout of signal processing circuits after we have used generative AI to generate a text description of a circuit.

Note that ELK itself doesn't render the drawing but only computes positions (and possibly dimensions) for the diagram elements. Other downstream frameworks are then used to render a circuit drawing and execute a simulation of the circuit.

Note that ELK was originally written to support the Java-based Eclipse IDE system and it still uses legacy Java code.

• See also: ELK documentation website
• ELK JSON Layout Format - a JSON format for storing nodes, ports, labels, edges, and edge sections of a layout diagram.

Elliptic filter

An Elliptic filter, also known as a Cauer filter, is a type of analog or digital filter with equalized ripple in both the passband and stopband. It provides the steepest transition between passband and stopband for a given filter order.

Example: Students design an Elliptic filter to achieve a specific cutoff frequency with minimal filter order, comparing its performance to other filter types.

Energy Signals

Energy signals are signals that have finite total energy, calculated as the integral of the squared magnitude over all time. They typically decay to zero as time approaches infinity and are contrasted with power signals which have infinite energy but finite average power.

Example: Students classify different signal types by computing their energy and power, learning that a decaying exponential pulse is an energy signal while a constant sinusoid is a power signal.

Energy Spectral Density (ESD)

Energy Spectral Density represents the distribution of signal energy over frequency. It is used to analyze how energy is distributed across different frequency components of a signal.

Example: Students calculate the ESD of a vibration signal to identify dominant frequencies associated with mechanical resonances.

Ergodicity

Ergodicity is a property of a stochastic process where time averages are equal to ensemble averages. It implies that a single, sufficiently long realization of the process can represent the entire statistical behavior.

Example: In signal analysis, students assess whether a given noise signal is ergodic by comparing its time-averaged statistics to theoretical ensemble averages.

Error correction

Error correction involves adding redundancy to transmitted information to protect against errors during transmission. It is essential for reliable communication in noisy channels.

Example: Students implement Reed-Solomon codes to correct burst errors in a simulated digital transmission system, enhancing data reliability.

Error detection

Error detection is the process of identifying errors in transmitted or stored data. It ensures data integrity by identifying corrupted information.

Example: In lab sessions, students use parity checks and CRC (Cyclic Redundancy Check) methods to detect errors in data packets during transmission.

Euler's formula

Euler's formula establishes a fundamental relationship between complex exponentials and trigonometric functions, expressed as \(e^{j\theta} = \cos(\theta) + j\sin(\theta)\). It is essential for analyzing sinusoidal signals and phasors in signal processing.

Example: Students use Euler's formula to convert time-domain sinusoidal signals into their phasor representations, simplifying the analysis of AC circuits.

Even Signals

Even signals are signals that exhibit symmetry about the vertical axis, meaning \(x(t) = x(-t)\) for all time \(t\). A cosine wave is a classic example of an even signal, and even signals have purely real Fourier transforms.

Example: Students decompose arbitrary signals into even and odd components, learning that any signal can be expressed as the sum of an even part and an odd part.

FIR filters

Finite Impulse Response filters are a type of digital filter with a finite number of coefficients. They are inherently stable and can have linear phase characteristics.

Example: Students design FIR filters using the window method to create low-pass filters for audio signal processing applications.

Fast Fourier Transform (FFT)

The Fast Fourier Transform is an efficient algorithm that computes the discrete Fourier transform (DFT) and its inverse, reducing the computational complexity from \(O(N^2)\) to \(O(N \log N)\) for a sequence of \(N\) points. This transformation decomposes a time-domain signal into its constituent frequencies, enabling rapid analysis and processing of frequency components.

Example: We use the FFT to analyze audio signals by converting a time-domain recording into its frequency spectrum. This allows them to identify dominant frequencies, filter out noise, and visualize the signal's frequency content for applications such as music analysis or noise reduction.

Feature extraction

Feature extraction involves transforming raw data into a set of characteristics that capture essential information for analysis. It is a crucial step in pattern recognition and machine learning.

Example: Students extract Mel-frequency cepstral coefficients (MFCCs) from audio signals to use as features for speech recognition tasks.

Filter banks

Filter banks consist of multiple filters that decompose a signal into different frequency bands. They are used in applications like audio processing and image compression to analyze and process signals in parallel frequency channels.

Example: Students design filter banks to separate an audio signal into various frequency bands, allowing independent processing and manipulation of each band for effects like equalization.

Filter design

Filter design is the process of creating filters with specific frequency responses to achieve desired signal processing objectives. It involves selecting appropriate filter types, orders, and parameters.

Example: Students engage in filter design projects where they create low-pass, high-pass, and band-pass filters tailored to remove specific noise frequencies from sensor data.

Filter Order

Filter order refers to the number of reactive elements or the degree of the transfer function polynomial in a filter design. Higher-order filters provide steeper roll-off transitions between passband and stopband but require more computational resources and may introduce more phase distortion.

Example: Students compare filters of different orders, observing how increasing the order of a Butterworth filter from 2nd to 8th order creates a sharper cutoff but increases computational complexity.

Filter Stability

Filter stability is a property ensuring that a filter produces bounded outputs for all bounded inputs, preventing runaway oscillations or unbounded growth in the output signal. For digital IIR filters, stability requires all poles of the transfer function to lie inside the unit circle in the z-plane.

Example: Students analyze the pole-zero plots of various IIR filter designs, identifying which configurations produce stable filters and which lead to unstable behavior.

Fourier Transform (FT)

The Fourier Transform is a mathematical transform that decomposes a continuous-time signal into its constituent frequencies. It provides a frequency-domain representation of time-domain signals.

Example: Students apply the FT to analyze the frequency content of an analog signal, understanding how different frequency components contribute to the overall signal.

Fourier series

Fourier series decompose periodic signals into sums of sine and cosine functions with discrete frequencies. It is fundamental for analyzing periodic signals in both time and frequency domains.

Example: Students represent a square wave using Fourier series, analyzing how adding higher harmonics affects the signal's approximation to the ideal waveform.

Frequency Domain

The frequency domain is a representation of signals or systems as functions of frequency rather than time. It reveals the frequency components present in a signal and their relative amplitudes and phases, enabling analysis and design of filters and systems.

Example: Students transform time-domain audio signals into the frequency domain using FFT, visualizing the spectral content to identify noise frequencies for removal.

Frequency Modulation (FM)

Frequency Modulation is a modulation technique where the frequency of the carrier wave is varied in accordance with the input signal. It is widely used in radio broadcasting and communication systems.

Example: In experiments, students generate FM signals and demodulate them to recover the original audio, studying the effects of modulation index on signal quality.

Frequency response

Frequency response describes how a system or filter responds to different frequency components of an input signal. It characterizes the gain and phase shift introduced by the system across frequencies.

Example: Students plot the frequency response of designed filters to evaluate their effectiveness in attenuating unwanted frequencies and preserving desired ones.

Generative Adversarial Networks (GANs)

Generative Adversarial Networks are a class of machine learning frameworks where two neural networks, a generator and a discriminator, compete to produce realistic data. GANs are used for tasks like image generation and data augmentation.

Example: In advanced projects, students use GANs to generate synthetic audio signals, exploring their applications in data augmentation for training signal processing models.

Gradient descent

Gradient descent is an optimization algorithm used to minimize the loss function by iteratively moving in the direction of the steepest descent as defined by the negative of the gradient. It is widely used in training machine learning models.

Example: Students implement gradient descent to train a neural network for signal classification, observing how learning rates and iterations affect convergence and accuracy.

High-pass filter (HPF)

A High-pass filter allows frequencies above a certain cutoff frequency to pass through while attenuating lower frequencies. It is used to remove low-frequency noise or to isolate high-frequency components.

Example: Students design an HPF to eliminate baseline wander in ECG signals, enhancing the detection of heartbeats by removing slow-varying trends.

IIR filters

Infinite Impulse Response filters are a type of digital filter that, unlike FIR filters, have feedback and an infinite impulse response. They are capable of achieving sharp frequency cutoffs with fewer coefficients but may have stability concerns.

Example: Students design IIR filters using the bilinear transform method, comparing their performance and computational efficiency to FIR filters in real-time audio processing.

Impulse Response

The impulse response is the output of a linear time-invariant (LTI) system when the input is a unit impulse function. It completely characterizes the system's behavior, as the response to any input can be computed by convolving the input with the impulse response.

Example: Students measure the impulse response of an audio system by playing a short click and recording the output, then use it to predict the system's response to music signals.

ISO Definition

A term definition is considered to be consistent with ISO metadata registry guideline 11179 if it meets the following criteria:

1. Precise
2. Concise
3. Distinct
4. Non-circular
5. Unencumbered with business rules

Integration

Integration is a calculus operation that computes the area under a curve defined by a function. It is used in signal processing for tasks such as finding signal energy and performing convolution.

Example: In assignments, students integrate the squared magnitude of a signal over time to calculate its total energy, applying numerical integration techniques.

Interpolation

Interpolation is the process of estimating unknown values between known data points. In signal processing, it is used to reconstruct continuous signals from discrete samples or to increase the sampling rate.

Example: Students apply interpolation techniques to upsample a discrete-time signal, filling in additional samples to achieve a higher resolution representation.

Inverse Fourier Transform (IFT)

The Inverse Fourier Transform converts a frequency-domain representation of a signal back into its time-domain form. It is essential for reconstructing time-domain signals from their frequency components.

Example: Students perform IFT on a frequency spectrum obtained via FFT to verify the accuracy of the signal reconstruction process.

Kalman filter

The Kalman filter is an algorithm that provides estimates of unknown variables by predicting a system's future states and updating them based on new measurements. It is widely used in navigation and tracking systems.

Example: Students implement a Kalman filter to estimate the position and velocity of a moving object using noisy sensor data, demonstrating its effectiveness in real-time tracking.

Laplace Transform

The Laplace Transform is a mathematical transform that converts a time-domain function into a complex frequency-domain representation using the complex variable \(s = \sigma + j\omega\). It is essential for analyzing continuous-time systems, solving differential equations, and studying system stability.

Example: Students use the Laplace Transform to solve RLC circuit differential equations, converting them to algebraic equations in the s-domain for easier manipulation and solution.

Least Mean Squares (LMS) algorithm

The Least Mean Squares algorithm is an adaptive filter algorithm that iteratively adjusts filter coefficients to minimize the mean square error between the desired and actual outputs. It is simple and widely used in adaptive filtering applications.

Example: In lab exercises, students apply the LMS algorithm to develop an adaptive noise canceller, effectively reducing background noise in audio signals.

Linear algebra

Linear algebra is a branch of mathematics dealing with vectors, matrices, and linear transformations. It is fundamental for understanding and implementing signal processing algorithms.

Example: Students use linear algebra concepts to solve systems of equations arising in filter design and to perform operations like matrix multiplication in signal transformations.

Linear Systems

Linear systems are systems that satisfy the principles of superposition (additivity) and homogeneity (scaling), meaning the response to a sum of inputs equals the sum of individual responses, and scaling the input scales the output proportionally. Linear systems are fundamental to signal processing analysis.

Example: Students verify linearity by applying multiple inputs to a filter separately and combined, confirming that the combined output equals the sum of individual outputs.

Low-pass filter (LPF)

A Low-pass filter allows frequencies below a certain cutoff frequency to pass through while attenuating higher frequencies. It is used to remove high-frequency noise or to smooth signals.

Example: Students design an LPF to eliminate high-frequency components from a sensor signal, ensuring that only the relevant low-frequency information is retained for analysis.

Machine Learning (ML)

Machine Learning is a field of artificial intelligence focused on developing algorithms that enable computers to learn patterns from data and make decisions. It is applied in signal processing for tasks like classification, regression, and prediction.

Example: In projects, students train machine learning models to classify different types of heartbeats from ECG data, utilizing features extracted through signal processing techniques.

Matrices

Matrices are rectangular arrays of numbers arranged in rows and columns. They are used in signal processing for operations like transformations, system representations, and solving linear equations.

Example: Students use matrix multiplication to perform transformations in multi-dimensional signal spaces, such as rotating image data in computer vision applications.

Mean

Mean is a statistical measure representing the average value of a set of numbers. It is used to summarize the central tendency of signal data.

Example: Students calculate the mean of a noisy signal to understand its baseline level and to use it in normalization procedures.

Multirate signal processing

Multirate signal processing involves the use of different sampling rates within a single system, such as in decimation and interpolation. It is essential for efficient signal representation and processing.

Example: Students implement a multirate system to downsample and upsample audio signals, optimizing processing speed while maintaining signal quality.

Multiresolution analysis

Multiresolution analysis decomposes a signal into components at various scales or resolutions. It is fundamental in wavelet transforms and enables detailed analysis of signal features at different levels.

Example: In wavelet projects, students perform multiresolution analysis to separate signal components, facilitating the detection of both coarse and fine features in biomedical signals.

Neural Networks (NN)

Neural Networks are computational models inspired by the human brain, consisting of interconnected layers of nodes (neurons) that process data. They are used in signal processing for tasks like classification, regression, and pattern recognition.

Example: Students build and train neural networks to classify speech signals, learning how network architecture affects performance and accuracy.

Noise cancellation

Noise cancellation is the process of removing unwanted noise from a signal to enhance the desired information. It is commonly used in audio processing and communication systems.

Example: Students develop noise cancellation algorithms to clean up speech recordings, improving clarity by reducing background noise through adaptive filtering techniques.

Nyquist theorem

The Nyquist theorem states that to accurately sample a continuous-time signal without aliasing, the sampling rate must be at least twice the highest frequency present in the signal. It sets the foundation for sampling in digital signal processing.

Example: Students apply the Nyquist theorem to determine appropriate sampling rates for different audio signals, ensuring accurate digital representation without loss of information.

Odd Signals

Odd signals are signals that exhibit anti-symmetry about the origin, meaning \(x(t) = -x(-t)\) for all time \(t\). A sine wave is a classic example of an odd signal, and odd signals have purely imaginary Fourier transforms.

Example: Students analyze the symmetry properties of various signals, learning that the odd component of a signal corresponds to the imaginary part of its Fourier transform.

Orthogonal Frequency Division Multiplexing (OFDM)

OFDM is a digital modulation technique that splits a signal into multiple orthogonal subcarriers, each carrying a portion of the data. It is widely used in modern communication systems like Wi-Fi and LTE.

Example: Students simulate an OFDM-based communication system, analyzing how subcarriers are multiplexed and demultiplexed to achieve high data rates and robustness against multipath fading.

Pattern recognition

Pattern recognition involves identifying patterns and regularities in data, enabling classification and prediction based on learned models. It is essential in applications like speech and image recognition.

Example: In assignments, students develop pattern recognition systems to identify specific gestures from motion sensor data, applying feature extraction and classification algorithms.

Periodic Signals

Periodic signals are signals that repeat their pattern at regular intervals, satisfying \(x(t) = x(t + T)\) for all time \(t\), where \(T\) is the fundamental period. They can be represented using Fourier series as sums of harmonically related sinusoids.

Example: Students analyze periodic waveforms like square waves and sawtooth waves, computing their Fourier series coefficients to understand their harmonic content.

Phase Modulation (PM)

Phase Modulation is a modulation technique where the phase of the carrier wave is varied in accordance with the input signal. It is used in various communication systems for transmitting information.

Example: Students generate and demodulate PM signals, studying how phase variations encode information and affect signal integrity in noisy environments.

Phase Shift Keying (PSK)

Phase Shift Keying is a digital modulation scheme where the phase of the carrier signal is varied to represent data symbols. It is widely used in wireless and digital communication systems.

Example: Students implement PSK modulation and demodulation schemes, analyzing bit error rates under different noise conditions to evaluate system performance.

Phasors

Phasors are complex numbers representing sinusoidal functions with amplitude and phase, simplifying the analysis of linear time-invariant systems in the frequency domain. They are used to analyze AC circuits and signal interactions.

Example: Students use phasors to solve AC circuit problems, representing voltage and current as rotating vectors to easily calculate impedances and power factors.

Polyphase filters

Polyphase filters are filter structures that decompose filtering operations into multiple phases, improving efficiency in multirate signal processing applications. They are used in implementations like interpolation and decimation.

Example: Students design polyphase filter banks for efficient upsampling and downsampling of audio signals, reducing computational complexity compared to standard filtering approaches.

Poles

Poles are the values of the complex variable \(s\) (in continuous-time) or \(z\) (in discrete-time) for which the transfer function of a system becomes infinite. They determine the system's natural frequencies, stability, and transient response characteristics.

Example: Students plot the poles of various filter transfer functions on the s-plane or z-plane, learning how pole locations affect stability and the shape of the impulse response.

Power Spectral Density (PSD)

Power Spectral Density quantifies how the power of a signal is distributed across different frequency components. It is used to analyze the frequency content and energy distribution of signals.

Example: Students compute the PSD of vibration data to identify dominant frequencies associated with mechanical faults in rotating machinery.

Probability

Probability is a branch of mathematics that deals with the likelihood of events occurring. It is fundamental in signal processing for modeling and analyzing random signals and noise.

Example: Students study the probability distributions of noise in communication systems, applying statistical methods to model and mitigate interference.

Quadrature Amplitude Modulation (QAM)

Quadrature Amplitude Modulation is a modulation scheme that combines both amplitude and phase modulation, allowing the transmission of multiple bits per symbol. It is widely used in digital communication systems for high data rates.

Example: Students implement QAM in a simulated communication system, analyzing its capacity and resilience to noise compared to simpler modulation schemes.

Quantization

Quantization is the process of mapping a continuous range of amplitude values to a finite set of discrete levels in analog-to-digital conversion. It introduces quantization error, which is the difference between the original continuous value and its quantized representation.

Example: Students experiment with different quantization bit depths on audio signals, listening to how reducing bits from 16 to 8 to 4 introduces increasingly audible noise and distortion.

Random processes

Random processes are collections of random variables indexed by time or space, used to model signals that evolve unpredictably. They are fundamental in understanding noise and stochastic signals.

Example: Students analyze random processes by modeling wireless channel noise, studying how it affects signal transmission and reception in communication systems.

Random variables

Random variables are variables that can take on different values based on probabilistic outcomes. They are used to model and analyze stochastic processes and noise in signal processing.

Example: Students model thermal noise in electronic circuits as a Gaussian random variable, applying statistical techniques to predict its impact on signal quality.

Real-time Processing

Real-time processing is the ability to process signals fast enough that the output is available within a specified time constraint, typically before the next input sample arrives. It is essential for applications like live audio processing, control systems, and telecommunications.

Example: Students implement a real-time audio equalizer that processes samples within the sampling period, learning about buffer management and computational constraints.

Recursive Least Squares (RLS) algorithm

The Recursive Least Squares algorithm is an adaptive filter method that recursively finds the filter coefficients minimizing the weighted linear least squares cost function. It provides faster convergence compared to LMS.

Example: In lab projects, students implement the RLS algorithm to adaptively filter out noise from a signal, observing its rapid convergence and performance advantages over LMS.

Regression

Regression is a statistical method for modeling the relationship between a dependent variable and one or more independent variables. It is used in signal processing for prediction and trend analysis.

Example: Students apply linear regression to predict future signal values based on past observations, evaluating the model's accuracy in time-series forecasting.

Sampling

Sampling is the process of converting a continuous-time signal into a discrete-time signal by taking measurements at regular intervals. It is a fundamental step in digital signal processing.

Example: Students sample an analog audio signal at different rates, observing the effects of sampling rate on signal representation and aliasing phenomena.

Sampling rate conversion

Sampling rate conversion changes the sampling rate of a discrete-time signal, either by upsampling (increasing) or downsampling (decreasing). It is used to match different system requirements or standards.

Example: Students implement sampling rate conversion algorithms to adapt audio signals for different playback devices, ensuring compatibility and maintaining quality.

Schemadraw

A Python library that uses a declarative circuit placement algorithm (left, right, above, below) when describing circuits.

• Schemadraw

Short-Time Fourier Transform (STFT)

The Short-Time Fourier Transform is a time-frequency analysis technique that applies the Fourier transform to short, overlapping segments of a signal. It provides localized frequency information over time.

Example: Students use STFT to create spectrograms of speech signals, visualizing how frequency content evolves during spoken words for speech recognition tasks.

Signal compression

Signal compression reduces the amount of data required to represent a signal by removing redundancies and irrelevant information. It is essential for efficient storage and transmission.

Example: Students implement JPEG compression on images, learning how transform coding and quantization reduce file sizes while maintaining visual quality.

Signal decomposition

Signal decomposition breaks down a complex signal into simpler components, such as fundamental frequencies or wavelets. It is essential for analysis, compression, and feature extraction.

Example: In projects, students decompose music signals into harmonic and percussive components, enabling separate processing and enhancement of each part.

Signal detection

Signal detection involves identifying the presence of a signal within noisy data. It is critical in applications like radar, communications, and biomedical monitoring.

Example: Students design detectors to identify ECG signals amidst physiological noise, applying statistical methods to enhance detection reliability.

Signal estimation

Signal estimation refers to the process of inferring the true signal from observed noisy measurements. It is fundamental in applications requiring accurate signal reconstruction.

Example: Students use Wiener filtering techniques to estimate the original speech signal from a noisy recording, evaluating the filter's effectiveness in different noise environments.

Signal filtering

Signal filtering involves manipulating a signal to remove unwanted components or to enhance desired features. It is a core operation in signal processing for noise reduction, feature extraction, and signal shaping.

Example: Students apply various filters to biomedical signals to remove artifacts, improving the quality of data used for diagnosis and analysis.

Signal prediction

Signal prediction estimates future values of a signal based on past and present data. It is used in applications like forecasting, control systems, and communication.

Example: In assignments, students develop predictive models using autoregressive techniques to forecast stock market trends based on historical price data.

Signal Processing

The analysis, manipulation, and interpretation of signals to extract meaningful information, enhance signal quality, or transform signals into more useful forms.

According to Arxiv Taxonomy, the field of Signal Processing is a discipline within the Electrical Engineering and Systems Science discipline. It includes the following research topics.

Theory, algorithms, performance analysis and applications of signal and data analysis, including physical modeling, processing, detection and parameter estimation, learning, mining, retrieval, and information extraction. The term "signal" includes speech, audio, sonar, radar, geophysical, physiological, (bio-) medical, image, video, and multimodal natural and man-made signals, including communication signals and data. Topics of interest include: statistical signal processing, spectral estimation and system identification; filter design, adaptive filtering / stochastic learning; (compressive) sampling, sensing, and transform-domain methods including fast algorithms; signal processing for machine learning and machine learning for signal processing applications; in-network and graph signal processing; convex and nonconvex optimization methods for signal processing applications; radar, sonar, and sensor array beamforming and direction finding; communications signal processing; low power, multi-core and system-on-chip signal processing; sensing, communication, analysis and optimization for cyber-physical systems such as power grids and the Internet of Things.

Signal reconstruction

Signal reconstruction is the process of rebuilding a continuous-time signal from its discrete samples, typically using interpolation methods. It ensures that the original signal can be accurately recovered from its sampled version.

Example: Students reconstruct audio signals from their samples using sinc interpolation, assessing the fidelity of the reconstructed signal compared to the original.

Signal-to-Noise Ratio (SNR)

Signal-to-Noise Ratio is a measure that compares the level of a desired signal to the level of background noise, typically expressed in decibels (dB). Higher SNR indicates a cleaner signal with less noise interference, and it is a key metric for evaluating signal quality in communication and measurement systems.

Example: Students measure the SNR of audio recordings under different conditions, learning how various noise sources degrade signal quality and how filtering can improve SNR.

Sparse representation

Sparse representation involves expressing a signal as a linear combination of a few non-zero elements from a dictionary of possible basis functions. It is useful for efficient signal representation and compression.

Example: Students apply sparse coding techniques to represent natural images with fewer coefficients, achieving compression without significant loss of detail.

Spectral Leakage

Spectral leakage is an artifact that occurs in Fourier analysis when the signal being analyzed is not perfectly periodic within the analysis window, causing energy to spread from the true frequency into adjacent frequency bins. Window functions are used to minimize this effect.

Example: Students observe spectral leakage by computing the FFT of a sine wave that doesn't complete an integer number of cycles in the window, then apply Hamming and Hanning windows to reduce the leakage.

Spectrogram

A spectrogram is a visual representation of the frequency spectrum of a signal as it varies with time, typically displayed as a heat map with time on the horizontal axis, frequency on the vertical axis, and color or intensity representing magnitude. It is created using the Short-Time Fourier Transform (STFT).

Example: Students generate spectrograms of speech signals to visualize how formant frequencies change during vowel sounds, aiding in understanding speech production and recognition.

Stability

Stability in signal processing systems refers to the ability of a system to produce bounded outputs for bounded inputs. It ensures that the system behaves predictably over time.

Example: Students analyze the stability of different filter designs by examining their pole locations, ensuring that all filters used in projects are stable.

Subband coding

Subband coding splits a signal into multiple frequency bands and encodes each band separately. It is used in audio and image compression to exploit frequency-specific redundancies.

Example: Students implement subband coding for audio signals, achieving compression by encoding each frequency band independently based on its perceptual importance.

Supervised learning

Supervised learning is a machine learning paradigm where models are trained on labeled data, learning to map inputs to desired outputs. It is used in signal processing for classification and regression tasks.

Example: Students train supervised learning models to classify different types of signals, using labeled datasets to teach the model to recognize specific patterns.

Support Vector Machines (SVM)

Support Vector Machines are supervised learning models used for classification and regression tasks by finding the optimal hyperplane that separates data into classes. They are effective in high-dimensional spaces.

Example: Students use SVMs to classify EEG signal patterns, distinguishing between different mental states based on their spectral features.

Time domain

The time domain represents signals as functions of time, focusing on how signal amplitude changes over time. It is one of the primary domains for analyzing and processing signals.

Example: Students analyze time-domain waveforms of audio signals to identify temporal features like amplitude envelopes and transient events.

Time-frequency analysis

Time-frequency analysis examines signals in both time and frequency domains simultaneously, providing a more detailed representation of signal characteristics. Techniques include STFT and wavelet transforms.

Example: Students apply time-frequency analysis to analyze transient events in biomedical signals, identifying when specific frequency components occur over time.

Time-Invariant Systems

Time-invariant systems are systems whose behavior and characteristics do not change over time, meaning a time-shifted input produces an identically time-shifted output. Combined with linearity, they form the important class of Linear Time-Invariant (LTI) systems that are central to signal processing.

Example: Students test whether a system is time-invariant by applying the same input at different times and verifying that the output is simply delayed by the same amount.

Transfer function

A transfer function describes the relationship between the input and output of a linear time-invariant system in the frequency domain. It characterizes the system's behavior and response to different frequencies.

Example: Students derive the transfer function of an electronic filter and analyze its impact on various input signals by examining the system's frequency response.

Unsupervised learning

Unsupervised learning is a machine learning paradigm where models find patterns and relationships in unlabeled data without explicit instructions. It is used in signal processing for clustering, dimensionality reduction, and anomaly detection.

Example: Students apply unsupervised learning techniques like k-means clustering to group similar signal patterns, identifying underlying structures without predefined labels.

Unit Impulse Function

The unit impulse function, also known as the Dirac delta function \(\delta(t)\) in continuous time or the Kronecker delta \(\delta[n]\) in discrete time, is a mathematical function that is zero everywhere except at the origin, where it is infinitely large with a total integral of one. It is fundamental for characterizing system responses.

Example: Students use the unit impulse as an input to LTI systems to determine their impulse responses, which completely characterize the system's behavior.

Unit Step Function

The unit step function, denoted \(u(t)\) in continuous time or \(u[n]\) in discrete time, is a function that equals zero for negative arguments and one for non-negative arguments. It is used to represent signals that turn on at a specific time and to analyze system step responses.

Example: Students apply the unit step function as input to a first-order RC circuit, analyzing how the system responds to a suddenly applied voltage.

Variance

Variance is a statistical measure that quantifies the spread of a set of values around the mean. It is used to assess the variability or dispersion in signal data.

Example: Students calculate the variance of a noise signal to understand its power distribution and to design appropriate filters for noise reduction.

Wavelet Transform (WT)

The Wavelet Transform is a signal processing technique that decomposes a signal into wavelets, allowing for both time and frequency localization. It is useful for analyzing non-stationary signals with transient features.

Example: Students apply the Wavelet Transform to analyze heart rate variability data, identifying both slow and rapid changes in heart rhythms.

White noise

White noise is a random signal with a constant power spectral density across all frequencies. It serves as a fundamental noise model in signal processing and communications.

Example: Students simulate white noise to test the robustness of filtering algorithms, ensuring that filters effectively remove noise without distorting the signal.

Wiener filter

The Wiener filter is an optimal linear filter that minimizes the mean square error between the estimated and true signals. It is used for signal restoration and noise reduction.

Example: Students implement Wiener filtering to denoise images, comparing the restored images to the original and evaluating the filter's effectiveness.

Window functions

Window functions are mathematical functions used to select a subset of a signal for analysis, typically in Fourier transforms. They reduce spectral leakage by tapering the signal edges.

Example: In FFT assignments, students apply different window functions like Hamming and Hann windows to minimize spectral leakage and improve frequency resolution.

Z-Transform

The Z-Transform is a mathematical transform that converts discrete-time signals into the complex frequency domain. It is used for analyzing and designing digital filters and systems.

Example: Students use the Z-Transform to analyze the stability and frequency response of digital filters, facilitating the design of effective filtering systems.

Zero-crossing rate

Zero-crossing rate is the rate at which a signal changes sign, indicating the number of times the signal crosses the zero amplitude axis. It is used as a feature in signal classification tasks.

Example: Students calculate the zero-crossing rate of audio signals to distinguish between different types of sounds, such as voiced and unvoiced speech.

Zeros

Zeros are the values of the complex variable \(s\) (in continuous-time) or \(z\) (in discrete-time) for which the transfer function of a system equals zero. They affect the frequency response by creating notches or nulls at specific frequencies and shape the overall filter characteristics.

Example: Students design notch filters by placing zeros at specific frequencies on the unit circle, effectively eliminating unwanted interference like 60 Hz power line noise.