Ontological Layer: Measurement and Detection
Signal Operations & Analysis
Separating Information from Chaos
The Observer
We have generated signals, transformed them to the frequency domain, and filtered them through systems. But now we must observe them. Realm 12 is the domain of measurement and detection. Here, we slide signals against each other to find mathematical similarity (Correlation), we hunt for hidden periodicity, we calculate the absolute limit of communication clarity (SNR), and we use the Power Spectral Density (PSD) to characterize the random, interfering chaos known as Noise.
The 14 Analysis Chambers
Ontological Mapping: From Measurement to Detection
Why Analyze?
- Measurement vs filtering/processing
- Extracting meaning and structure
- Detecting similarity and echoes
- Characterizing signal quality
Signal Correlation
- Measuring mathematical similarity
- Sliding one signal against another
- The concept of 'Lag' (delay)
- Template matching intuition
Autocorrelation
- Comparing a signal with itself: Rxx[k]
- Maximum peak always at lag k=0
- Detecting hidden periodicity/pitch
- Even symmetry property
Cross-Correlation
- Comparing x[n] with y[n]: Rxy[k]
- Finding a template within a stream
- Radar/Sonar time-delay estimation
- Not commutative: Rxy[k] ≠ Ryx[k]
Properties
- Autocorrelation zero-lag = Total Energy
- Symmetry: Rxx[k] = Rxx[-k]
- Correlation vs Convolution distinction
- x[n] * y[-n] equivalence
Energy & Power
- Energy: Total squared sum (Transient)
- Power: Average energy over time (Continuous)
- Parseval's Theorem in practice
- Root Mean Square (RMS) value
Power Spectral Density
- Distribution of power across frequencies
- Wiener-Khinchin Theorem connection
- Fourier Transform of Autocorrelation
- Essential for random signals/noise
ESD vs PSD
- ESD: Applied to Energy signals (finite)
- PSD: Applied to Power signals (infinite)
- |X(ω)|² for deterministic transients
- Choosing the right spectral tool
Noise Models
- Unwanted, random interference
- White Noise: Flat PSD across all frequencies
- Colored Noise: Frequency-dependent PSD
- Thermal, Quantization, and Impulsive types
Signal-to-Noise Ratio
- SNR = P_signal / P_noise
- Decibel scale: 10 log10(Ps/Pn)
- The fundamental metric of signal quality
- Filtering to improve SNR margins
Detection Theory
- Matched Filter concept (Template matching)
- Extracting known signals from high noise
- Setting threshold values for detection
- Trade-off: False Alarms vs Missed Detections
Spectral Estimation
- Estimating PSD from finite measured data
- The Periodogram method
- Welch's Method: Windowing and Averaging
- Trading frequency resolution for variance reduction
Analysis Classification
- Time Domain vs Frequency Domain metrics
- Deterministic signals vs Random processes
- White vs Colored interference
- High SNR vs Low SNR environments
14 — The Analysis Laboratory
In Realm 11, we built the convolution engine to filter signals. Here, we built the correlator to detect information hidden within them. By measuring Energy, calculating SNR, and analyzing the Power Spectral Density of noise, we can pull truth from chaos. Our next step is to alter the very fabric of the discrete timeline itself: Realm 13, Multirate DSP.