OFDM Spectral Correlation Density

The cyclic prefix (CP) creates a distinct cyclostationary signature by copying the end of each OFDM symbol to its beginning. This repetition generates spectral correlation peaks at cyclic frequency α = k/Ts, where Ts is the OFDM symbol duration.

• Enables blind estimation of useful symbol length (Tu) and total symbol length (Ts)
• Correlation pattern distinguishes normal CP from extended CP modes
• Peak magnitude proportional to CP-to-symbol ratio, providing a feature for waveform classification

Known pilot symbols inserted at regular subcarrier intervals and OFDM symbol positions create deterministic spectral correlation patterns. These embedded reference signals produce peaks at cyclic frequencies corresponding to the pilot spacing in time and frequency.

• Comb-type pilots: Generate correlation at cyclic frequencies related to pilot subcarrier spacing
• Block-type pilots: Produce periodicity at the frame or slot repetition rate
• Enables discrimination between LTE, 5G NR, and WiFi standards based on unique pilot grid structures

The SCF is computed as the Fourier transform of the cyclic autocorrelation function along the time axis, producing a two-dimensional frequency-cyclic frequency map.

• FAM (FFT Accumulation Method): Efficient algorithm using channelization and FFT-based smoothing
• SSCA (Strip Spectral Correlation Analyzer): Alternative method trading resolution for computational speed
• Output reveals conjugate vs. non-conjugate correlation features, distinguishing real-valued from complex-valued cyclostationarity

The presence of unused subcarriers at band edges creates a spectral correlation signature distinct from fully-loaded waveforms. The null subcarrier pattern produces characteristic notches in the SCD profile.

• Enables estimation of occupied bandwidth and DC subcarrier location
• Null pattern periodicity reveals the FFT size (Nfft) when combined with active subcarrier count
• Differentiates between carrier aggregation configurations in LTE-Advanced and 5G NR

SCD-based detection exploits the fact that stationary noise exhibits spectral correlation only at α = 0, while modulated signals show non-zero correlation at multiple cyclic frequencies.

• Provides robust detection at SNR levels where energy detection fails
• Cycle frequency domain filtering separates signal features from noise floor
• Enables reliable classification below -10 dB SNR for signals with strong cyclostationary signatures like CP-OFDM

Overlapping OFDM signals with different symbol rates or CP lengths occupy distinct regions in the cyclic frequency domain, enabling separation without spatial filtering.

• Signals with different subcarrier spacing produce peaks at different α values
• Co-channel interference resolved when signals have asynchronous symbol timing
• Enables blind source separation for spectrum monitoring in dense electromagnetic environments

A blind parameter estimation technique that identifies the number of subcarriers in an OFDM signal by analyzing cyclostationary signatures or autocorrelation properties. The spectral correlation density function reveals the subcarrier spacing as the frequency separation between correlated spectral components. By measuring this spacing and knowing the signal bandwidth, the FFT size can be derived. Alternative methods include:

• Autocorrelation-based lag detection
• Cumulant-based order estimation
• Deep learning regression on spectrograms

A compact numerical representation of an OFDM signal's discriminative characteristics used as input to machine learning classifiers. A well-constructed feature vector for spectral correlation density analysis includes:

• Cyclic domain profile: normalized SCD magnitude at key cycle frequencies
• Spectral coherence peaks: alpha frequencies where correlation exceeds threshold
• Cumulant statistics: higher-order moments resistant to Gaussian noise
• Bandwidth and subcarrier spacing estimates: derived from SCD structure These features enable classification of OFDM variants (LTE, WiFi, 5G NR) without demodulation.

The identification of specific OFDM implementation details to determine the wireless standard or vendor-specific configuration of a transmitter. Spectral correlation density analysis reveals protocol-specific features such as:

• Pilot patterns: unique subcarrier allocations for channel estimation
• Preamble structures: training sequences with known correlation properties
• Frame timing: periodicity of synchronization signals
• Cyclic prefix ratios: distinguishing LTE normal CP from extended CP This enables precise identification of LTE, 5G NR, WiFi 6, or proprietary waveforms.