Cyclostationary Signal Detection via Cyclic Correlation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current methods for detecting cyclostationary signals, particularly in opportunistic radio systems and radio watermarking, face challenges in reliably determining the presence of signals with low signal-to-noise ratios and varying transmission parameters, leading to high false alarm rates and inefficiencies in blind and semi-blind contexts.
Innovation Solution
A method involving a discrimination function based on the quadratic form of cyclic correlation coefficients, utilizing the normalized incomplete lower gamma function and cyclic correlation coefficients, to estimate the presence of cyclostationary signals, even in low signal-to-noise conditions, by comparing the estimated value against a predetermined threshold for reliable detection.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Device complexity
If simple power detection is used to detect radio signals, then the detection process is simple, but the detection reliability is poor because spread signals are drowned out in noise
Solution Approach 1:
The invention changes the detection parameter from simple power level to cyclic autocorrelation characteristics. By computing the cyclic autocorrelation function at specific cyclic frequencies and time shifts, the detector exploits the periodic statistical properties of spread signals to distinguish them from noise, thereby improving detection reliability while maintaining reasonable complexity
Solution Approach 2:
The invention replaces the mechanical/power-based detection approach with a statistical signal processing approach. Instead of measuring signal power directly, the system uses cyclic autocorrelation analysis to detect the presence of spread signals by identifying their characteristic cyclic statistical properties, which are absent in noise
2Reliability
If semi-blind detection methods using autocorrelation function are used to test for cyclic frequency characteristics, then detection capability is improved, but the method does not work in blind context and cannot achieve predetermined false alarm rates
Solution Approach 1:
The invention creates a universal detection method that functions in both blind and semi-blind contexts. The cyclic autocorrelation-based detector can operate without prior knowledge of signal parameters (blind mode) or with partial knowledge (semi-blind mode), achieving predetermined false alarm rates in both scenarios through appropriate threshold selection based on the statistical distribution of the test statistic
3Reliability
If cyclic frequency testing is performed to detect radio watermarks, then watermark detection is enabled, but the method cannot determine presence with predetermined false alarm rate across various audio signals
Solution Approach 1:
The invention introduces feedback through statistical threshold selection. The detector computes the cyclic autocorrelation and compares it against a threshold determined by the desired false alarm rate and the statistical properties of the test statistic under the null hypothesis. This feedback mechanism enables precise control of false alarm rates across different audio signals while maintaining reliable watermark detection
Data Source
Figure 1
Figure 2
Figure 3
AI summary
The present invention relates to a method of detecting a cyclo-stationary signal in a signal to be analysed, received from a sender by a receiver or read from a storage medium by a reading device. According to this method, the value of a discrimination function J s' is estimated, expressing as a quadratic form cyclic correlation coefficients of the signal to be analysed for a set (I* M) of non-zero time offsets and a set ( /K) of cyclic frequencies, and the value (formula (I)) is compared to a bad detection level η to determine whether said cyclo-stationary signal is present in the signal to be analysed, where γ (v,x) is the normalized incomplete lower gamma function, V is the product of the cardinal (M) of said set of non-zero time offsets and of the cardinal (K) of said set of cyclic frequencies, U is the width of the self-correlation window over which are calculated the cyclic correlation coefficients, and σ4 is the square of the noise variance.