Ontological Layer: Channel Impairment
Noise & Probability
What happens when randomness enters the system?
The Mathematics of Chaos
A real communication system never operates in a perfectly clean universe. Unwanted random disturbances alter the transmitted signal, burying the truth inside jitter and static. Because noise is fundamentally unpredictable, we cannot use ordinary algebra to describe it. Instead, we must use the mathematical language of Probability and Random Variables. Realm 02 establishes the statistical tools required to measure, model, and predict the impact of Thermal Noise, calculating the ultimate boundary of performance: the Signal-to-Noise Ratio.
The 24 Statistical Nodes
Ontological Mapping: From Random Variables to SNR
What is Noise?
- Random unwanted disturbance
- Noise vs Distortion vs Interference
- Limits communication range and data rate
- Unavoidable physical reality
Randomness
- Deterministic vs Random phenomena
- Sample space and Events
- Mathematical measure of likelihood
- Basic probability rules
Random Variables
- Mapping outcomes to numerical values
- Discrete vs Continuous RVs
- Notation (X for variable, x for value)
- Complex random variables (I/Q)
Probability Mass (PMF)
- For discrete random variables
- P_X(x) = P(X = x)
- Probabilities assigned to specific values
- Sum of all probabilities = 1
Probability Density (PDF)
- For continuous random variables
- f_X(x) describes density, not direct probability
- P(a ≤ X ≤ b) = Integral of PDF
- Total area under curve = 1
Cumulative Dist. (CDF)
- F_X(x) = P(X ≤ x)
- Accumulated probability up to value x
- Non-decreasing function from 0 to 1
- Derivative of CDF yields PDF
Expectation & Mean
- E[X], the statistical mean (μ)
- Long-run average value
- Center of mass of the PDF
- Linearity of expectation
Statistical Moments
- E[Xⁿ] captures distribution shape
- 1st Moment: Mean
- 2nd Moment: Average Power (E[X²])
- Central moments (variance, skewness)
Variance & Std Dev
- Var(X) = E[(X − μ)²] = σ²
- Measure of spread / dispersion
- Standard Deviation (σ) = √Variance
- Variance = AC Noise Power (if mean=0)
Gaussian Distribution
- The "Normal" bell curve: N(μ, σ²)
- Fundamental model for thermal/random noise
- Central Limit Theorem justification
- Characterized entirely by Mean and Variance
Gaussian Noise
- Noise amplitude follows Gaussian PDF
- Typically modeled as Zero-Mean (μ=0)
- Power is equal to Variance (σ²)
- Crucial for probability-of-error math
Noise Power
- Average power of random noise signal
- Proportional to E[n²(t)] (Mean-Square value)
- RMS Noise Voltage = √Variance
- Direct link between statistics and physics
Autocorrelation R_X(τ)
- Statistical similarity of a signal with its shifted self
- Reveals temporal structure in random noise
- R_X(0) = Total Average Power
- Gateway to Power Spectral Density
Random Processes
- A family of random variables indexed by time
- X(t) notation
- Ensemble of possible waveforms
- Required to model continuous noise streams
Stationary & Ergodic
- Stationary: Statistics do not change over time
- Wide-Sense Stationary (WSS): Constant mean, R_X(τ)
- Ergodic: Time average equals statistical average
- Allows practical measurement of noise
Power Spectral Density
- S_X(f) describes power distribution over frequency
- Units: Watts/Hz
- Wiener-Khinchin: Fourier Transform of Autocorrelation
- Essential for filter and channel analysis
White Noise
- Idealized flat PSD across all frequencies (N₀/2)
- Contains equal power at every frequency
- Autocorrelation is a pure impulse (unpredictable)
- White Gaussian Noise (AWGN) model basis
Band-Limited Noise
- Real noise passed through a receiver filter
- PSD is restricted to a finite bandwidth (B)
- Total Noise Power = PSD × Bandwidth
- Why wider bandwidths let in more noise
Thermal Noise
- Caused by random electron agitation in resistors
- Johnson-Nyquist noise
- N = kTB (Power depends on Temp and Bandwidth)
- Modeled as White Gaussian Noise
Shot & Flicker Noise
- Shot: Discrete arrival of carriers at junctions (diodes)
- Flicker (1/f): PSD rises at low frequencies
- Internal hardware noise sources
- Dominates different frequency bands
Signal-to-Noise Ratio
- SNR = Signal Power / Noise Power
- SNR(dB) = 10 log₁₀(S/N)
- The ultimate metric of signal clarity
- Determines maximum achievable capacity
Noise Factor (F)
- F = SNR_in / SNR_out
- Measures SNR degradation caused by a device
- Noise Figure (NF) = 10 log₁₀(F)
- Friis formula for cascaded amplifiers
Probability of Error (Pe)
- Noise pushes signal across decision threshold
- Overlap of Gaussian probability tails
- Foundation of digital receiver detection
- P_e decreases as SNR increases
24 — The Statistical Laboratory
By applying probability, we have tamed the chaos of the real world into predictable mathematics. We know how to measure noise power, and we understand that SNR is the ultimate currency of communication. Now we must learn how to protect our information before sending it into this noisy channel. This prepares you for Realm 03: Amplitude Modulation.