Signal detection method based on prior information
By constructing a multidimensional feature analysis framework based on prior information and a method of dynamically adjusting detection parameters, the problem of traditional signal detection in identifying weak signals under extremely low signal-to-noise ratio conditions is solved, and efficient and accurate signal detection in complex noise environments is achieved.
Patent Information
- Application Number
- CN202510461859.0
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-04-14
- Publication Date
- 2025-09-16
AI Technical Summary
Traditional signal detection methods have difficulty effectively identifying weak signals under extremely low signal-to-noise ratio conditions, and their performance degrades especially in complex noise environments. The failure to fully utilize prior information limits the development potential of detection algorithms.
A multi-dimensional feature joint analysis framework based on prior information is constructed, and the prior distribution of signals and noise is mined through a two-dimensional likelihood detector. The detection coefficient and threshold value are dynamically adjusted, and the detection parameters are optimized to adapt to complex environments.
The accuracy and flexibility of signal detection are improved, and it can accurately identify target signals under conditions of large noise interference, reduce misjudgments, and ensure high detection performance.
Smart Images
Figure CN120653905A_ABST
Abstract
Description
Technical Field
[0001] The invention relates to a signal detection method based on prior information, and belongs to the field of radio communication signal detection. Technical Background
[0002] Signal detection technology plays an indispensable role in wireless communications. Its core task is to determine whether a signal of interest exists within a scene through advanced signal processing techniques, enabling subsequent operations such as parameter estimation, code recognition, and modulation analysis to extract valuable information. In complex electromagnetic environments and severe noise interference, signal detection technology is crucial for improving system performance and ensuring data accuracy. For fields such as national security and scientific research, efficient signal detection capabilities are directly related to the reliability of communication systems.
[0003] Traditional signal detection methods, such as energy detectors, are favored for their simple structure, mature theory, and ease of implementation. However, these methods typically offer a general-purpose solution designed to accommodate a wide range of applications, rather than optimizing detection performance for specific signal types. This design approach often proves inadequate under extremely low signal-to-noise ratio (SNR) conditions, especially for weak signal detection. For example, in the presence of high background noise power or complex interference, the performance of energy detectors can significantly degrade or even fail. As application scenarios become more complex and technical requirements increase, traditional methods are increasingly unable to meet the high standards of sensitivity and accuracy required by modern signal detection.
[0004] Many existing detection technologies tend to adopt general-purpose approaches designed to accommodate a wide range of signal types. While this improves the technology's applicability, it also means that the known signal pattern distribution—that is, prior information—is not fully utilized in specific scenarios. In fact, fully utilizing this prior information can significantly improve detection performance, especially in weak signal processing. However, most existing technologies either ignore this valuable resource or utilize it only in limited ways. This approach not only wastes valuable prior knowledge but also limits the potential of detection algorithms. Summary of the Invention
[0005] In order to solve the technical problem that target signals are difficult to detect in complex environmental scenarios, the purpose of the present invention is to provide a signal detection method based on prior information. This method constructs a detection framework for multi-dimensional feature joint analysis by deep mining and making full use of known prior information, and integrates the signal model distribution into the detection process. The present invention breaks through the limitations of traditional likelihood detection methods and expands them from one dimension to two dimensions, so that the changing characteristics of signals in multiple dimensions can be captured more comprehensively. This multi-dimensional analysis method enables the detection algorithm to identify target signals more accurately, and can effectively avoid misjudgment even in cases where the signal is weak or the noise interference is large. On this basis, the corresponding discrimination threshold is flexibly selected according to different environmental changes to adapt to the complex and changeable signal detection environment and improve the efficiency and reliability of signal detection.
[0006] The purpose of the present invention is achieved through the following technical solutions.
[0007] The present invention discloses a signal detection method based on prior information, comprising the following steps:
[0008] Step 1: Obtain the observation signal sequence received by the receiver.
[0009] The received observation sequence Y={y1,y2,...,y i ,...,y m There are two cases: the observation sequence contains only noise, and the observation sequence contains target signal and noise;
[0010] When the observation sequence Y contains only noise, each observation value y i ,i=1,2,...,m is represented by the corresponding random noise value n i Composition, that is: y i =n i ; Where N={n1,n2,...,n i ,...,n m} represents a random noise sequence;
[0011] When the observation sequence Y contains the target signal, each observation value y i The corresponding random noise value n i and target signal s i Together, they constitute: i =n i +s i ; Where S={s1,s2,...,s i ,...,s m} indicates the target signal sequence.
[0012] Step 2: First, model the prior distribution of the target signal and noise; secondly, use the prior distribution model of the random noise sequence and the target signal sequence to construct a two-dimensional likelihood accumulation detector.
[0013] Step 2.1: First, model the prior distribution of the target signal and noise, which includes the following:
[0014] For a received random noise sequence of N, where each noise term n i It has a mean of zero and a variance of σ 2 The Gaussian distribution of is expressed as follows:
[0015]
[0016] For the target signal sequence S, where the signal symbol s i ∈{1,-1}, its probability density function is expressed as:
[0017]
[0018] where δ(·) is the Dirac delta function.
[0019] If the observation sequence Y contains only noise, its probability density function is the same as the background noise, which can be expressed as:
[0020]
[0021] If there is a target signal sequence in the observation sequence Y, its probability density function is expressed as:
[0022]
[0023] Step 2.2: Next, use the prior distribution model of the random noise sequence and the target signal sequence to construct a two-dimensional likelihood accumulation detector, which specifically includes the following:
[0024] If the observation sequence Y contains only noise, the observed y i The conditional probability density is:
[0025]
[0026] The cumulative likelihood is written as:
[0027]
[0028] The cumulative log-likelihood is written as:
[0029]
[0030] If the target signal sequence exists in the observation sequence Y, the observed yi The conditional probability density is:
[0031]
[0032] The cumulative likelihood is written as:
[0033]
[0034] The cumulative log-likelihood is written as:
[0035]
[0036] Step 3: Construct a detection statistic for the obtained received signal sequence based on a two-dimensional likelihood accumulation detector.
[0037] Based on the two-dimensional likelihood accumulation detector constructed in step 2, that is, equations (7) and (8), a detection statistic T is constructed for the acquired observation signal sequence Y. The statistic T is the cumulative log-likelihood LL in the presence of the target signal sequence. signal The cumulative log-likelihood LL is the same as the log-likelihood LL when only noise is included. noise The difference is:
[0038] T=LL signal -αLL noise (11)
[0039] Here, α represents the detection coefficient. By introducing the coefficient α, the weight between the target signal sequence and the random noise sequence can be flexibly balanced. When α>1, the suppression of noise fluctuations will be enhanced; when α<1, the emphasis will be on the accumulation of signal features. It is especially suitable for complex non-stationary noise environments.
[0040] Step 4: Calculate the corresponding threshold value based on the optimal detection probability.
[0041] The optimal detection probability in signal detection is the highest sum of the accuracy rates for detecting the target signal sequence and the random noise sequence, that is, maximizing the sum of the true positive rate (TPR) and the true negative rate (TNR). The true positive rate is the probability of correctly detecting the target signal when the target signal sequence is present, while the true negative rate is the probability of correctly detecting the noise when the target signal is absent. The present invention simultaneously determines the optimal detection coefficient α and threshold value τ by solving the optimization problem in equation (12):
[0042] (a,τ)=argmax a,τ {F(a,τ)} (12)
[0043] in,
[0044] F(α,τ)=TPR(α,τ)+TNR(α,τ) (13)
[0045] In formula (13), TPR(α,τ)=P(T>τ|Y=N+S) represents the true positive rate when the detection coefficient α and the threshold value τ are given, that is, the probability that the detection statistic T exceeds the threshold value τ;
[0046] TNR(α,τ)=P(T≤τ|Y=N) represents the true negative rate when α and τ are given, that is, the probability that the detection statistic T is less than or equal to the threshold value τ.
[0047] Step 5: Compare the statistic with the threshold value to determine the detection result of the target signal.
[0048] Compare and analyze the size of the statistic T and the threshold value τ constructed in the above steps: if the value of the statistic T is less than the threshold value τ, then it is determined that the source signal does not exist in the current observation sequence; conversely, if the value of the statistic T is greater than or equal to the threshold value τ, then it is considered that the source signal exists in the current observation sequence.
[0049] Substitute the optimized detection coefficient α in formula (12) into formula (11) to calculate the statistic T, and compare it with the threshold value τ: if the value of the statistic T is less than or equal to the threshold value τ, then it is determined that the target signal sequence does not exist in the current observation sequence; conversely, if the value of the statistic T is greater than the threshold value τ, then it is determined that the target signal sequence exists in the current observation sequence.
[0050] Beneficial effects:
[0051] 1. The signal detection method based on prior information of the present invention constructs a two-dimensional likelihood accumulation detector by modeling the prior distribution of target signals and noise, analyzes the statistical differences between signal characteristics and noise characteristics, effectively distinguishes target signals from noise, and improves the accuracy and reliability of signal detection.
[0052] 2. The present invention's signal detection method based on prior information overcomes the limitations of traditional likelihood detection methods by expanding them from one dimension to two, thereby more comprehensively capturing the signal's changing characteristics across multiple dimensions. This multidimensional analysis enables the detection algorithm to more accurately identify target signals, effectively avoiding misjudgments even in weak signals or with significant noise interference.
[0053] 3. The signal detection method based on prior information disclosed in this invention dynamically calculates detection coefficients and thresholds through a joint optimization function, enabling the detection system to adapt to different signal environments. This method allows the system to automatically adjust detection parameters based on the current signal environment, improving detection flexibility and adaptability, ensuring high detection performance under various conditions.
[0054] 4. The signal detection method based on prior information disclosed in the present invention determines the optimal detection probability by maximizing the sum of the true positive rate and the true negative rate, thereby achieving high-accuracy signal existence judgment and reducing the possibility of false positives and missed detections. BRIEF DESCRIPTION OF THE DRAWINGS
[0055] Figure 1 This is a flow chart of a signal detection method based on prior information disclosed in the present invention;
[0056] Figure 2 This is a schematic diagram of a signal detection method based on prior information disclosed in the present invention. DETAILED DESCRIPTION
[0057] In order to further explain the implementation ideas of the present invention, the following will be combined with the accompanying drawings in the embodiments of the present invention to carefully and clearly describe the technical solutions in the embodiments of the present invention. The aforementioned and other technical contents, features and effects of the present invention can be clearly presented in the following detailed description of the specific embodiments with the accompanying drawings. Through the description of the specific embodiments, a more in-depth and specific understanding of the technical means and effects adopted by the present invention to achieve the intended purpose can be obtained.
[0058] Example:
[0059] like Figure 1 As shown, the present embodiment discloses a signal detection method based on prior information, and the specific implementation steps are as follows:
[0060] Step 1: Obtain the observation signal sequence received by the receiver.
[0061] In the signal detection scenario, when the signal sequence transmitted by the known satellite is
[0062] S={s1,s2,...,s i ,...,s m}, where the signal symbol s i ∈{1,-1}, during the receiver reception process, the transmitted signal sequence S will be interfered by the random noise sequence superposition, that is, S+N, where N={n1,n2,...,n i ,...,n m} represents a random noise sequence.
[0063] The observation sequence Y received by the current detector is {y1,y2,...,y i ,...,y m There are two cases: the observation sequence contains only noise, and the observation sequence contains the target signal and noise. The detector needs to determine whether the observation sequence Y contains the target signal sequence S.
[0064] When the observation sequence Y contains only noise, each observation value y i The corresponding random noise value n i Composition, that is: y i =n i .
[0065] When the observation sequence Y contains the target signal, each observation value y i The corresponding random noise value n i and target signal s i Together, they constitute: i =n i +s i .
[0066] Step 2: First, model the prior distribution of the target signal and noise; secondly, use the prior distribution model of the random noise sequence and the target signal sequence to construct a two-dimensional likelihood accumulation detector.
[0067] Step 2.1: First, model the prior distribution of the target signal and noise, which includes the following:
[0068] For a received random noise sequence of N, where each noise term n i It has a mean of zero and a variance of σ 2 The Gaussian distribution of is expressed as follows:
[0069]
[0070] For the target signal sequence S, its probability density function is expressed as:
[0071]
[0072] where δ(·) is the Dirac delta function.
[0073] If the observation sequence Y contains only noise, its probability density function is the same as the background noise, which can be expressed as:
[0074]
[0075] If there is a target signal sequence in the observation sequence Y, its probability density function is expressed as:
[0076]
[0077] Step 2.2: Next, use the prior distribution model of the random noise sequence and the target signal sequence to construct a two-dimensional likelihood accumulation detector, which specifically includes the following:
[0078] If the observation sequence Y contains only noise, the observed y iThe conditional probability density is:
[0079]
[0080] The cumulative likelihood is written as:
[0081]
[0082] The cumulative log-likelihood is written as:
[0083]
[0084] If the target signal sequence exists in the observation sequence Y, the observed y i The conditional probability density is:
[0085]
[0086] The cumulative likelihood is written as:
[0087]
[0088] The cumulative log-likelihood is written as:
[0089]
[0090] Step 3: Construct a detection statistic for the obtained received signal sequence based on a two-dimensional likelihood accumulation detector.
[0091] Based on the two-dimensional likelihood accumulation detector constructed in step 2, that is, equations (20) and (21), a detection statistic T is constructed for the acquired observation signal sequence Y. The statistic T is the cumulative log-likelihood LL in the presence of the target signal sequence. signal The cumulative log-likelihood LL is the same as the log-likelihood LL when only noise is included. noise The difference is:
[0092] T=LL signal -αLL noise (twenty four)
[0093] Here, α represents the detection coefficient. By introducing the coefficient α, the weight between the target signal sequence and the random noise sequence can be flexibly balanced. When α>1, the suppression of noise fluctuations will be enhanced; when α<1, the emphasis will be on the accumulation of signal features. It is especially suitable for complex non-stationary noise environments.
[0094] Step 4: Calculate the corresponding threshold value based on the optimal detection probability.
[0095] The optimal detection probability in signal detection is the highest sum of the accuracy rates for detecting the target signal sequence and the random noise sequence, that is, maximizing the sum of the true positive rate (TPR) and the true negative rate (TNR). The true positive rate is the probability of correctly detecting the target signal when the target signal sequence is present, while the true negative rate is the probability of correctly detecting the noise when the target signal is absent. The present invention simultaneously determines the optimal detection coefficient α and threshold value τ by solving the optimization problem in equation (12):
[0096] (a,τ)=argmax a,τ {F(a,τ)}(25)
[0097] in,
[0098] F(α,τ)=TPR(α,τ)+TNR(α,τ)(26)
[0099] In formula (25), TPR(α,τ)=P(T>τ|Y=N+S) represents the true positive rate when the detection coefficient α and the threshold value τ are given, that is, the probability that the detection statistic T exceeds the threshold value τ;
[0100] TNR(α,τ)=P(T≤τ|Y=N) represents the true negative rate when α and τ are given, that is, the probability that the detection statistic T is less than or equal to the threshold value τ.
[0101] Step 5: Compare the statistic with the threshold value to determine the detection result of the target signal.
[0102] Compare and analyze the size of the statistic T and the threshold value τ constructed in the above steps: if the value of the statistic T is less than the threshold value τ, then it is determined that the source signal does not exist in the current observation sequence; conversely, if the value of the statistic T is greater than or equal to the threshold value τ, then it is considered that the source signal exists in the current observation sequence.
[0103] Substitute the optimized detection coefficient α in formula (25) into formula (24) to calculate the statistic T, and compare it with the threshold value τ: if the value of the statistic T is less than or equal to the threshold value τ, then it is determined that the target signal sequence does not exist in the current observation sequence; conversely, if the value of the statistic T is greater than the threshold value τ, then it is considered that the target signal sequence exists in the current observation sequence.
[0104] The above specific description further illustrates the purpose, technical solutions and beneficial effects of the invention in detail. It should be understood that the above description is only a specific embodiment of the present invention and is not intended to limit the scope of protection of the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the scope of protection of the present invention.
Claims
1. A signal detection method based on prior information, characterized in that: The following steps are included: Step 1: Obtain the observation signal sequence received by the receiver; Step 2: First, the prior distribution of the target signal and noise is modeled; secondly, a two-dimensional likelihood accumulation detector is constructed using the prior distribution model of the random noise sequence and the target signal sequence; Step 3: construct a detection statistic for the obtained received signal sequence based on a two-dimensional likelihood accumulation detector; Step 4: Calculate the corresponding threshold value based on the optimal detection probability; Step 5: Compare the statistic with the threshold value to determine the detection result of the target signal.
2. The signal detection method based on prior information according to claim 1, wherein: The implementation method of step one is: The received observation sequence Y={y1,y2,...,y i ,...,y m There are two cases: the observation sequence contains only noise, and the observation sequence contains target signal and noise; When the observation sequence Y contains only noise, each observation value y i ,i=1,2,...,m is represented by the corresponding random noise value n i Composition, that is: y i =n i ; Where N={n1,n2,...,n i ,...,n m } represents a random noise sequence; When the observation sequence Y contains the target signal, each observation value y i The corresponding random noise value n i and target signal s i Together, they constitute: i =n i +s i ; Where S={s1,s2,...,s i ,...,s m } indicates the target signal sequence.
3. The signal detection method based on prior information according to claim 2, characterized in that: The method for modeling the prior distribution of target signal and noise in step 2 is: For a received random noise sequence of N, where each noise term n i It has a mean of zero and a variance of σ 2 The Gaussian distribution of is expressed as follows: For the target signal sequence S, where the signal symbol s i ∈{1,-1}, its probability density function is expressed as: where δ(·) is the Dirac delta function; If the observation sequence Y contains only noise, its probability density function is the same as the background noise, which can be expressed as: If there is a target signal sequence in the observation sequence Y, its probability density function is expressed as:
4. The signal detection method based on prior information according to claim 2, wherein: The method for constructing a two-dimensional likelihood accumulation detector using the prior distribution model of the random noise sequence and the target signal sequence in step 2 is: If the observation sequence Y contains only noise, the observed y i The conditional probability density is: The cumulative likelihood is written as: The cumulative log-likelihood is written as: If the target signal sequence exists in the observation sequence Y, the observed y i The conditional probability density is: The cumulative likelihood is written as: The cumulative log-likelihood is written as:
5. The signal detection method based on prior information according to claim 4, characterized in that: The implementation method of step three is: Based on the two-dimensional likelihood accumulation detector constructed in step 2, namely, equations (7) and (8), a detection statistic T is constructed for the acquired observation signal sequence Y; the statistic T is the cumulative log-likelihood LL in the presence of the target signal sequence. signal The cumulative log-likelihood LL is the same as the log-likelihood LL when only noise is included. noise The difference is: T=LL signal -αLL noise (11) Here, α represents the detection coefficient. By introducing the coefficient α, the weight between the target signal sequence and the random noise sequence can be flexibly balanced. When α>1, the suppression of noise fluctuations will be enhanced; when α<1, the emphasis will be on the accumulation of signal features.
6. The signal detection method based on prior information according to claim 5, characterized in that: The implementation method of step 4 is: The optimal detection probability in signal detection refers to the maximum sum of the accuracy of detecting the target signal sequence and the random noise sequence, that is, maximizing the sum of the true positive rate TPR and the true negative rate TNR. The true positive rate refers to the probability of correctly detecting the target signal when the target signal sequence exists, while the true negative rate refers to the probability of correctly detecting the noise when the target signal does not exist. The optimal detection coefficient α and threshold value τ are determined by weighing the optimization problem in equation (12): (a,τ)=argmax a,τ {F(a,τ)} (12) where, F(α,τ)=TPR(α,τ)+TNR(α,τ) (13) In formula (13), TPR(α,τ)=P(T>τ|Y=N+S) represents the true positive rate when the detection coefficient α and the threshold value τ are given, that is, the probability that the detection statistic T exceeds the threshold value τ; TNR(α,τ)=P(T≤τ|Y=N) represents the true negative rate when α and τ are given, that is, the probability that the detection statistic T is less than or equal to the threshold value τ.
7. The signal detection method based on prior information according to claim 6, characterized in that: The implementation method of step five is: Substitute the optimized detection coefficient α in formula (12) into formula (11) to calculate the statistic T, and compare it with the threshold value τ: if the value of the statistic T is less than or equal to the threshold value τ, it is determined that the target signal sequence does not exist in the current observation sequence; conversely, if the value of the statistic T is greater than the threshold value τ, it is determined that the target signal sequence exists in the current observation sequence.
8. The signal detection method based on prior information according to claim 7, characterized in that: By modeling the prior distribution of target signals and noise, and based on the statistical differences between signal characteristics and noise characteristics, an asymmetric two-dimensional likelihood accumulation detector is constructed to improve the accuracy and reliability of signal detection.