In stationary random processes, the autocorrelation function (ACF) and power spectral density (PSD) are related through the Fourier transform. The ACF provides a measure of similarity between a signal and a shifted version of itself, and the PSD, being the Fourier transform of the ACF, shows how power is distributed over frequency. The relationship is mathematically represented as \( S(f) = \int_{-\infty}^{\infty} R(\tau) e^{-j2\pi f \tau} d\tau \), where \( S(f) \) is the PSD and \( R(\tau) \) the autocorrelation function. This relationship highlights the correspondence between time and frequency domain analyses .

Gaussian random variables in communication systems are defined by two parameters, the mean, and variance, corresponding to the location and spread of the distribution. Key properties include its bell-shaped, symmetrical nature and the Central Limit Theorem, which states the sum of many independent random variables converges to a Gaussian distribution. This is important in communications because many noise processes approximate Gaussian distributions, making them foundational in modeling and analysis. Additionally, Gaussian variables can model signals subjected to minimal distortion, leading to optimal signal processing .

The cumulative distribution function (CDF) of a random variable describes the probability that the variable will take a value less than or equal to a particular value. The fundamental properties of the CDF are: it is non-decreasing, right-continuous, and it ranges from 0 to 1. These properties imply that as the input value increases, the probability does not decrease, ensuring consistency in the probability interpretation. This influences the distribution and statistical behavior of random variables by ensuring that probabilities are always bounded and cumulative, thereby providing a complete description of the distribution of a random variable .

Ergodicity in random processes refers to the property where time averages and ensemble averages are identical. If a process is ergodic, then the statistical properties (mean, variance, autocorrelation) can be determined from a single, sufficiently long realization of the process. For example, in measuring the average temperature over a year, ergodicity implies the time average from one year equals the average from multiple similar systems. This is crucial in signal processing and communications, as it allows system behavior analysis without needing multiple samples .

The expected value of a random variable is a measure of its central tendency, calculated as the sum or integral of all possible values each weighted by their probability. Variance measures the spread or dispersion of the random variable values around the expected value, calculated as the expectation of the squared deviations from the mean. These measures are critical in statistical analysis as they provide insights into the location and variability of data, essential for making predictions and conducting probabilistic assessments in fields like finance, engineering, and science .

The cross-correlation function measures the similarity between two time-series signals as a function of the time-lag applied to one of them. This is particularly useful in identifying time lags at which the signals are synchronized, revealing underlying dependencies. In applications such as signal processing or econometrics, analyzing the cross-correlation function helps in detecting lead-lag relationships, synchrony, or causal influences between signals, thus aiding in prediction and control strategies .

In multivariate analysis, joint distribution describes the probability of simultaneous occurrence of two or more random variables. Marginal distribution provides probabilities of a subset, ignoring others, and is derived by summing (discrete) or integrating (continuous) over the variables not of interest. Conditional distribution describes the probability of one variable given known values of others, derived by dividing the joint probability by the marginal probability. These distributions are vital in understanding the relationships and dependencies between variables, facilitating complex analyses in fields such as econometrics, machine learning, and statistical inference .

Power spectral density (PSD) describes how the power of a signal is distributed with frequency, typically used for signals that are stationary in nature. Energy spectral density (ESD), on the other hand, provides a frequency distribution of the energy of a signal, usually applied to signals of finite energy. PSD is relevant in scenarios dealing with power, like communications systems, where it helps in understanding how system bandwidth affects signal power. ESD is important in analyzing transient signals, providing insights into how energy concentrates across frequencies, essential for applications such as pulse analysis .

The autocorrelation function (ACF) has properties such as symmetry (R(τ) = R(-τ)), maximum at zero lag, and non-negative definite property. These characteristics allow distinguishing deterministic and random processes; deterministic signals have a predictable ACF form, often periodic if the signal is periodic, whereas random processes display broader, less predictable ACF features. In communications, understanding the ACF helps to determine signal predictability and plan system designs that optimize filtering and detection processes .

A Poisson process is characterized by events occurring independently and at constant average rates over time. Key applications include modeling rare events such as arrival of packets in network traffic or radioactive decay. The properties include a mean equal to the variance, and memorylessness, which makes it useful in scenarios where future events are independent of past events. In communications, Poisson models help in designing and predicting load and performance expectations for systems like call centers and network traffic .