WO2009097755A1

WO2009097755A1 - A method and apparatus for locally detecting signal, a method and apparatus for detecting signal at center, and a system for detecting signal

Info

Publication number: WO2009097755A1
Application number: PCT/CN2009/070075
Authority: WO
Inventors: Lei Chen; Jun Wang; Shaoqian Li; Linjun Lu
Original assignee: Huawei Technologies Co., Ltd.
Priority date: 2008-02-01
Filing date: 2009-01-08
Publication date: 2009-08-13
Also published as: CN101499859A

Abstract

A method and apparatus for locally detecting signal, a method and apparatus for detecting signal at center, and a system for detecting signal are disclosed. The method for locally detecting signal comprises: according to the determined quantification threshold corresponding to the local node, detection statistic corresponding to the signal to be detected is quantified and the quantification result is obtained. The quantification result is used for judging whether the signal to be detected is valid signal. The method for detecting signal at center comprises: obtaining the quantification result which is acquired through quantifying detection statistic by local node. According to the quantification result, global decision result indicating whether the signal to be detected is valid signal is obtained. Applying the solutions above, detection performance can be expediently and simply improved.

Description

Local and central detection signal method, device and detection signal system The present application claims to be submitted to the Chinese Patent Office on February 1, 2008, the application number is 200810006087.8, the invention name is "local, central detection signal method and local, central detection The priority of the Chinese Patent Application, the entire disclosure of which is incorporated herein by reference. TECHNICAL FIELD The present invention relates to the field of detection technologies, and in particular, to a method for locally detecting a signal, a method for centrally detecting a signal, a local detecting device and a center detecting device, and a system for detecting a signal. BACKGROUND In a wireless sensor network and a cognitive communication (Cognitive Radio, CR) communication field, when detecting an effective signal, a distributed detection method is often used, and at least one local node detects the detected signal, and obtains it locally. A detection statistic is then reported to the central node, and the central node obtains a global decision result of "whether the signal to be detected is a valid signal" according to each detection statistic. For example, in a CR network, in order to improve spectrum utilization, cognitive wireless communication devices can operate in an authorized frequency band in an "Opportunistic Way" manner. Here, an authorized user can also be called a Primary User (PU). Therefore, the CR system needs to detect whether there is a PU signal on the target frequency band in order to provide the cognitive radio with a candidate channel that is not occupied by the PU and is accessible to the CR system.

Generally, the system for implementing a distributed detection signal includes a local node and a central node, and the local node is configured to obtain a local decision result of whether a valid signal exists according to a local detection algorithm or obtain a detection statistic of a signal to be detected for determining, the center Node is used to fuse local decisions of various local nodes The result or the detected statistic is finally judged to determine whether there is a detection result of the valid signal. For example, the spectrum sensing technology for detecting PU signals can be divided into two categories: single-node sensing and cooperative sensing. In cooperative sensing, there are two types according to different transmission overheads: one is that each local detecting node makes a "PU". After the local decision of the signal exists, the local decision result is transmitted to the central node, and then the central node uses the "and" and "or" criteria to perform data fusion on each local node.

The global decision result of "whether there is a PU signal". Since the local node only provides a 1-bit decision result, it cannot provide more information, such as an energy signal for decision, etc., so the performance of the cooperative scheme is limited. This scheme is suitable for situations where the system has strict transmission overhead requirements. The second is that each local detecting node does not make a decision, and transmits some or all of the detected continuous data to the central node, such as the energy data of the signal to be detected. Then, the central node can use Bayesian criteria for data fusion, and The comprehensive judgment yields a global judgment result. The detection performance of this method is good, but because the dynamic range of the transmission detection data is large, the transmission overhead is often large, which is difficult to implement in practice.

It can be seen that the current method of detecting signals cannot effectively improve the signal detection performance. SUMMARY OF THE INVENTION Embodiments of the present invention provide a method for locally detecting a signal, which can effectively improve signal detection performance.

Embodiments of the present invention also provide a method for detecting a center, which can effectively improve signal detection performance.

Embodiments of the present invention additionally provide a local detecting device that can effectively improve signal detection performance.

Embodiments of the present invention additionally provide a center detecting device that can effectively improve signal detection performance. Embodiments of the present invention further provide a system for detecting a signal, which system can effectively improve signal detection performance.

To achieve the above objective, the technical solution of the embodiment of the present invention is specifically implemented as follows:

An embodiment of the present invention provides a method for locally detecting a signal, where the method includes:

And determining, according to the determined quantization threshold corresponding to the local node, the detection statistic corresponding to the detection signal, to obtain a quantization result, where the quantization result is used to determine whether the to-be-detected signal is a valid signal.

The embodiment of the invention provides a method for detecting a center, the method comprising:

Obtaining a quantized result obtained by quantifying a detection statistic by a local node;

Based on the quantized result, a global decision result indicating whether the signal to be detected is a valid signal is obtained. An embodiment of the present invention provides a local detecting device, where the local detecting device includes:

a detector, configured to detect a detection statistic corresponding to the signal to be detected in the local node, and a quantizer, configured to quantize the detection metric obtained by the detector according to a predetermined quantization threshold.

An embodiment of the present invention provides a center detecting device, where the center detecting device includes:

An obtaining module, configured to obtain a quantized result of the local detecting device, where the quantized result is a result obtained by quantifying a detection statistic corresponding to the signal to be detected;

The execution module obtains, according to the quantized result obtained by the obtaining module, a global decision result indicating whether the signal to be detected is a valid signal.

An embodiment of the present invention provides a system for detecting a signal, where the system for detecting a signal includes: a local node, configured to determine a quantization threshold corresponding to the local node, and quantize the detection statistic corresponding to the detection signal according to the quantization threshold, Obtaining a quantized result, the local node is at least one; the central node is configured to obtain a quantized result obtained by the local node to quantize the detection statistic, And according to the quantized result, a global decision result indicating whether the signal to be detected is a valid signal is obtained. Embodiments of the present invention have the following beneficial effects:

The local, central detection signal method, the local, the central detection device, and the detection signal system provided by the embodiments of the present invention can adjust the number of bits transmitted to the central node by quantifying the detection statistics of each local node to the center. The node provides more detection statistic information, which can effectively improve detection performance. BRIEF DESCRIPTION OF DRAWINGS FIG. 1 is a schematic structural diagram of a system for detecting a signal according to an embodiment of the present invention;

2 is a schematic flow chart of a method for detecting a signal in Embodiment 1 of the present invention;

3 is a schematic flowchart of a method for detecting a signal according to Embodiment 2 of the present invention;

3-1 is a schematic flowchart of step 303 in the second embodiment of the present invention;

4 is a schematic structural diagram of a local detecting apparatus according to an embodiment of the present invention;

FIG. 5 is a schematic structural diagram of a center detecting apparatus according to an embodiment of the present invention; FIG.

Figure 6 is a graph showing the relationship between the error probability and the number of locally detected nodes;

Graph, which is a graph of the relationship between the probability of error and the number of locally detected nodes. BEST MODE FOR CARRYING OUT THE INVENTION Hereinafter, the present invention will be described in further detail with reference to the accompanying drawings.

FIG. 1 is a schematic structural diagram of a system for detecting a signal according to an embodiment of the present invention. As shown in FIG. 1, the system includes: at least one local node and one central node, which are connected by a channel between the local node and the central node. Here, the local node may be a physical entity that implements detection locally, and the central node is a physical entity that aggregates detection results of at least one local node. Wherein, the local node includes: a detector that locally detects the detected signal and a quantizer that quantizes the detected statistic obtained by the detector; and the central node includes: an estimator for estimating the likelihood ratio of each local node based on the quantized result from each local node And an execution module that fuses the likelihood ratios obtained by the respective estimators and the respective quantized results.

In the system for detecting signals in this embodiment, the quantizer in the local node further compresses the locally determined decision data, and uses β(·) to represent the quantization process of the quantizer, expressed as equation (1):

Q (x) = _Vi , xe A _i , i = l, 2,- - - , q ( 1 )

Where iff means if and only if, JC denotes the quantized data, ^ denotes the quantizer output, q denotes the number of quantization levels, is the quantization interval, and Δ, = [ί3⁄4, _αί+1 ;) , where Represents the quantization threshold.

The quantizer used in this embodiment may be a local optimum quantizer or a uniform quantizer.

For the first quantizer, it is designed to be optimal at the lowest signal-to-noise ratio required by the system, because in signal detection, it is more challenging to detect weak signals, but higher in signal-to-noise. Signals, even with some non-optimal techniques, can easily meet system requirements. When designing the quantizer, its Signal to Noise Ratio (SNR) is exactly the lowest level required by the system. Under this assumption, the optimal quantizer is actually the local optimal quantizer (Locally Optimal). Quantizer, LOQ). The local optimum quantizer can be designed using the deflection criteria, and the deviation is used to indicate that the detection statistic is at H. And the statistical distance between the two cases, the greater the deviation, indicating the better the detection performance, Q represents the quantized detection statistic, then the deviation D (β) can be expressed as the formula ( 2 ):

Where Α denotes the mathematical expectation under the assumption, £. Represents mathematical expectations under hypothetical conditions. V. Indicated under the assumption H. The variance below.

Suppose that for any quantization interval ^, the probability that the detection statistic is within the range is: ₇ (Ζ)^ Ρ[ Ε Δ,] = | _Δ ( >.

"'... ' ( 3 )

Where j=0, l, respectively correspond to H. And two hypotheses, /^x), represent the probability density function. Then, according to formula (8) and the definition of expectation and variance, formula (2) is rewritten as:

Where ^ denotes the coefficient of probability, V is a vector consisting of _1⁄4 , for a given quantization interval Δ, the maximum V of equation (7) is satisfied:

When the formula (10) is satisfied, the deviation is obtained at the quantization interval !! ^Δ) is:

D(A) = l(D)-l ( 6 ) In (11), ( ⁷ )

The final goal of determining the quantization interval is to maximize the deviation at the quantization interval, so the final quantization interval can be derived from equation (12) as Δ: Δ = argmax / (A) ( 8 )

Δ

Assume that the distribution functions (x) and F of the statistic are detected. (X) can be obtained by theoretical analysis or simulation. Then P _Q (1) can be found by the distribution function, namely:

P ₁ (l) = F _l (a _l+1 )-F ₁ (a _l ), P ₀ (I) = F ₀ (a _l+1 ) - F ₀ where ί3⁄4 is the quantization threshold and has _αι = -∞ and _α3⁄4+1 =+ _∞ , where -∞ means negative infinity, and +∞ means positive infinity. Equation (8) can be converted to:

Δ = arg max ~

/=1 ^F o { ^a i ₊ i)- ^F o { ^a i) ( 9 )

For the second quantizer, it is a dynamic range based uniform quantizer (Dynamic Range based Uniform Quantizer, DRUQ): The quantizer uniformly divides the dynamic range of the detection statistic into q quantization intervals according to the determined number of quantization levels, assuming that the detection statistic of the i-th moment of a local detection node is 7; ( _Ζ · = 1,2,· · ·, «), the maximum value T _max and the minimum value 7^ of 7; in this w time are counted, and after the double-quantization, the quantization interval is:

A― T max -T nun

q ( 10 )

Finally, determine the amount 4乜 threshold α _; for: = 7^ + (/-1) Δ, = 1, 2, · · ·^ + 1

To implement the above-mentioned function of determining the quantization threshold, the local node may further include:

A quantization threshold determiner is configured to quantify the number of levels according to a transmission overhead requirement of the local node to the central node. And determining, according to a statistical rule of the detection statistics obtained by the detector, a quantization threshold corresponding to the local node.

For the first quantizer, the quantization threshold determiner includes:

a statistical acquisition unit, configured to acquire a distribution function, a mean value, or a variance of the detection statistic;

a first quantization threshold determining unit, configured to determine, according to a distribution function, a mean value, or a variance of the detection statistic obtained by the statistical acquisition unit, a quantization interval that maximizes a deviation between the detection statistic and a maximum deviation, according to the quantization Interval, determining a quantization threshold corresponding to the local node, where Η. The signal to be detected is a valid signal, indicating that the signal to be detected is not a valid signal.

For the second quantizer, the quantization threshold determiner includes:

The dynamic range determining unit is configured to determine a dynamic range of the detected statistic according to the maximum value and the minimum value of the detected statistic.

a second quantization threshold determining unit, wherein the dynamic range obtained by the dynamic range determining unit is divided by the number of quantization levels, and the quantization interval of the local node is obtained, and the local node quantization interval and the minimum value of the detection statistic are determined. The quantization threshold corresponding to the local node.

In this embodiment, the local check detected by the local node by adding a quantizer to the local node The transmission data of the measured statistic is adjusted to provide more detection statistic information to the central node, and has better synergistic gain, and can obtain better performance under the condition of less local node and lower local detection performance. . Therefore, the design difficulty of the system topology and the design difficulty of the local sensing algorithm can be alleviated, so that the detection performance can be effectively improved.

FIG. 2 is a schematic flow chart of a method for detecting a signal according to Embodiment 1 of the present invention. In this embodiment, the local optimal statistic is used to quantize the local detection statistic. As shown in FIG. 2, the method includes the following steps:

Step 201: Determine the number of quantization levels of the local node according to the requirement of the transmission overhead.

In this embodiment, the local nodes are (≥1). The number of quantization levels can be determined according to the transmission overhead requirement of the local node to the central node, that is, when the system can tolerate a large transmission overhead, the quantization level of the quantizer can be set to a larger number of bits to increase the accuracy of the decision. When the system cannot tolerate a large transmission overhead, the quantization level of the quantizer is set to a small number of bits to reduce the transmission overhead.

Step 202: Determine a quantization threshold of the local node according to the number of quantization levels of the local node.

In this embodiment, a local optimal quantizer is used, and the method for determining the quantization threshold is: according to the distribution function, determining a statistical distance such that the detection statistic is at H ₀ and the two cases, that is, the deviation, the maximum quantization interval And the quantization threshold (/ = 1, 2,...).

Here, the local node uses the theoretical analysis method to find the distribution function and F of the detection statistic. (x) In the present embodiment, it is assumed that (x) and F are determined by theoretical analysis by detecting the analytical expression of the statistic and the information of the noise and the signal. (x) expression. Assuming that the local node uses the energy detection to obtain the detection statistic, then ( ) and F _Q (x) are determined by three factors: noise, signal power and detection time. Usually, thermal noise is used to estimate or measure on the no-load channel. So that can be based on F. For the definition of (x), find F. ^). At the same time, since the local optimum quantization criterion is used in this embodiment, the power of the signal is the lowest level required by the system, so that F x^ can be obtained. Of course, in the case where it is difficult to obtain (x) and F ₀ (x) by the theoretical analysis method, the present embodiment can also be obtained by a simulation method. Since the detection statistic at different moments can be regarded as independent and identical distribution, according to the central limit theorem, the detection statistic can be regarded as obeying the normal distribution. Therefore, it is only necessary to directly obtain the mean and variance of the detected statistics by running the simulation based on the easily available information such as the signal type and noise power. Then, based on the mean and variance of the detected statistics, it is determined that the detection statistic is at H. The quantization interval and the quantization threshold ί3⁄4 (/ = 1, 2, · · · , ) with the largest deviation from the two cases.

Step 203: The local node performs local detection to obtain local detection statistics before quantization.

In this embodiment, the local detection statistic, such as signal energy data, before quantization can be obtained by using the local detection method in the prior art.

Step 204: Quantify the local detection statistic according to the determined quantization threshold.

In this embodiment, the local detection statistic is quantized using equation (1) based on the determined quantization threshold.

Step 205: The central node acquires the quantized result after quantizing each local detection statistic. Step 206: The central node performs data fusion on each quantized result to obtain a global decision balance. In this embodiment, the method of minimum error probability criterion is adopted to perform data fusion on each quantization result.

Suppose the quantization result is ^ where / represents the number of the local node, and / = 1, 2, · · · , «. In the absence of cost information, it is the optimal fusion criterion. The minimum error probability criterion can be expressed as:

Wherein, and respectively indicate that the signal to be detected is a valid signal and there are no two cases, the corpse ( ) indicates a prior probability that the signal to be detected is a valid signal, and Ρ(Η.;) indicates a prior probability that the signal to be detected is not a valid signal. The left side of the formula indicates the likelihood ratio of the test result, and the right side indicates the decision threshold. Decision threshold table When the central node gets a "true" global decision result, when

..., ^Ηι )< ≤, get the "true" global decision result at the center node, when ( _Μι , (Η ₀ )

= When you can get 'true' or 'Ή. true' according to the set policy, corpse..., |Η. ) (Η ₀ ) The global decision result. Convert the formula (11) into a log-likelihood ratio test form, and obtain the formula (12):

If the detection statistic of each detection node is independent of each other, and it is for the sake of convenience, it is assumed that the detection statistic obtained after quantization is 1, 2, ... results, then:

corpse

=ΠΓΜ"'= )

i ⁼¹ (13) where represents the set of all nodes ^ whose detection statistic takes the value j. Similar, get the formula

(14):

Substituting equations (13) and (14) into equation (12) and simplifying it:

Among them, ^Λ . = ^, is the /th detection node

Ρ (Η.)

^The probability of the detection statistic of ^{11 is} the likelihood ratio. Thus, the central node gets the global decision result is The a priori rate and likelihood ratio of the local node are determined.

Of course, if the quantized local detection statistic is one bit, the "and" or "or" criteria used in the prior art can also be used for data fusion.

In this embodiment, after the local detection statistic is quantized by using a local optimal quantizer, and the quantized discrete local detection statistic can be data fusion through the minimum error criterion, thereby estimating the correct one. Global judgment result.

The following is an example of detecting the PU signal in the CR system as an example. FIG. 3 is a schematic diagram of a method for detecting a signal in the first embodiment of the present invention by using a uniform quantizer to quantize the local detection statistic. The specific implementation method of the minimum error probability criterion is also specifically given in this embodiment.

Step 301: Each local node quantizes the local detection statistic according to a predetermined quantization threshold.

In this embodiment, a uniform quantizer is used. At the initial moment, each local node detects the target frequency band by itself and quantizes according to the quantization threshold determined by the formula (10) to obtain the quantized detection statistic.

_M !(0) = /, l=l,2,---,q

Step 302: The central node quantizes the detection statistic _{M of} each local node at the initial moment _; (0) = /, Z = 1, 2, -, ^ and the set initial prior probability Λ. (0) and the initial likelihood ratio A _a (0), the initial value of the global decision result is obtained.

In order to guarantee the convergence of the iterative algorithm, the initial value should be designed to be close to the true value. In the absence of a priori information, this embodiment will be embarrassing. (0) Set to 1, set Λ, ₇ (0) to increment according to the increase of 1. Using the formula (15), the data of each detection statistic "_; (0) = /, Ζ = 1, 2, ..., is fused, and the initial value of the global judgment result is calculated." (0).

Step 303: Estimate the prior probability and likelihood ratio of the local node at the current moment according to the global decision result of the previous moment and the quantization result of the local node. The following is specifically to estimate the prior probability k at time k. ( ) and the initial likelihood ratio are taken as an example for explanation. Step 303 can be specifically divided into the following steps:

303-1: According to the global judgment result at the k-th moment. ( ) and the quantized detection statistic of each local node ") = /, l = i, 2 ", q, determine the value of the decision state ( ) ( i = \ -, n ) of each local node at the moment.

Considering that the decision result after the data fusion has high accuracy, the embodiment replaces the actual signal information of whether the signal to be detected is a valid signal by the global judgment result, and an estimated value equal to or close to the real situation can be obtained. Assume D. Indicates H. For the global judgment space, the global judgment space expressed as true, when the global judgment result has a high degree of accuracy, the following formula is satisfied: corpse ( , j = ^ ( 16)

P( _M ,=;|D.)*P( _M; =/|H.), ; = 0,1 ( ₁₇ ) So /^Hj, P(H ₀ ), P( _Ui =/ ) and /^ _W; =/|H. The calculation of the probability value can be converted to /^Dj, P(D ₀ ), corpse (^ =/| ) and /^ _; =/|£). ) calculation. At this time, (t) indicates the judgment state value of the third local node at the time & time:

Among them, it indicates the possible value space of the decision state. j' denotes a global decision result, j=0 denotes that the signal to be detected is not a valid signal, and j=l denotes that the signal to be detected is a valid signal. / Indicates the quantized local detection statistic.

303-2: Update the accumulated value of the decision state according to the decision status value (i = \, -, n ) at the first moment.

In this embodiment, the cumulative value is defined as: (^) = ∑ 5 _; (Ρ) Where N _s (fe) represents the number of times the state occurred at the time. Equation (19) can be rewritten as:

J _i (k) = J _i (kl) + S _i (k) (20) In this embodiment, a limited memory method can be used to avoid the influence of historical data, specifically: It can be determined according to a preset fixed window length. The cumulative value of the decision status value before the update is superimposed on the judgment status value of the previous time to the accumulated value of the decision status before the update. That is, when statistically determining the state value, a fixed-length window is used, and each time a decision state value of a new time is counted, a decision state value at the earliest time is removed. Let the previous moment be k, N is the preset fixed window length, and & (t) is the judgment state value of the i-th local node at the kth time, then the formula (20) can be adjusted to obtain the formula (21). ):

J) = ∑ S _; (P) (21) This embodiment can also use the forgetting factor method to update the cumulative value of the decision state, specifically: first multiply the cumulative value of the decision state before the update by a value less than one. Then, the cumulative value of the decision state at the previous time is added to obtain the accumulated value of the updated decision state. Equation (20) can be adjusted to:

Wherein, the forgetting factor is a number greater than 0.9 and less than 1. By the iteration of (22), the earlier the decision result state value corresponding to the weight is smaller, and the current result of the decision result state value corresponds to 1, thereby gradually eliminating the influence of the historical data.

303-3: Extract the weighted value of the decision state value according to the accumulated value of the updated decision state, and estimate the prior probability Λ used at time k+1. ( ) and initial likelihood ratio A _a (t):

303-4: Determine the prior probability A used at time k+1. (fe) and initial likelihood ratio) converge, if 杲 Yes, go to step 304, otherwise, let t = t + l, execute 303-1.

In this embodiment, Λ is judged. The method of whether (it) and Λ _ί7 (fc) ( = 1, 2, · · ·, ) converge can be determined by the judgment commonly used in the prior art |A. (yt)-A. T-1)1 and |Λ)-Λ, -1)1 is less than the set value, and if so, it is considered to converge.

Step 304: According to the estimated first risk rate Λ. (k) and the initial likelihood ratio A _a and the local detection statistics quantized at the current time, a relatively accurate global judgment result is calculated.

In this embodiment, after the local detection statistic is quantized by using the homo-quantizer, and the quantized discrete local detection statistic can be data fusion through the minimum error criterion, the correct global is estimated. As a result of the decision, the finite memory method is also used to eliminate the influence of historical data, which can increase the adaptability of the estimation algorithm to the time-varying detection environment.

FIG. 4 is a schematic structural diagram of a local detecting apparatus according to an embodiment of the present invention. As shown in FIG. 4, the local detecting device includes: a detector 410, which detects a detection statistic corresponding to a signal to be detected in the local node. The quantizer 420 is configured to quantize the detection statistic obtained by the detector according to a quantization threshold corresponding to the local detecting device.

The local detecting device may further include:

The quantization threshold determiner 430 determines the quantization level according to the transmission overhead requirement between the local detection device and the central node; and determines the quantization threshold corresponding to the local detection device according to the statistical rule of the detection statistics obtained by the detector 410.

The quantization threshold determination module 430 includes:

The statistics obtaining unit 431 acquires a distribution function, a mean value, or a variance of the detection statistic. The first quantization threshold determination unit 432 determines that the detection statistic is at H based on the distribution function, the mean value, or the variance of the detection statistic obtained by the statistical acquisition unit 431. And the quantization interval with the largest deviation between the two, and the quantization threshold corresponding to the local detecting device is determined according to the quantization interval, where H is. The signal to be detected is a valid signal, indicating that the signal to be detected is not a valid signal.

Of course, in this embodiment, if the quantization threshold is determined by using a dynamic range method, the quantization threshold determination module includes:

a second quantization threshold determining unit, wherein a dynamic range of the detection statistic obtained by the dynamic range determining unit is divided by the number of the quantized levels, and a quantization interval of the local detecting device is obtained, according to the quantization interval and detection of the local detecting device The minimum value of the statistic determines the quantization threshold corresponding to the local detecting device.

FIG. 5 is a schematic structural diagram of a center detecting device according to an embodiment of the present invention. As shown in Figure 5, the center detecting device includes:

The obtaining module 510 is configured to obtain a quantization result of at least one local detecting device, where the quantization result is a quantized result obtained by quantizing the detection statistic corresponding to the detection signal.

The executing module 520 obtains, according to the quantization result obtained by the obtaining module 510, a global decision result indicating whether the signal to be detected is a valid signal.

Execution module 520 includes:

The estimating unit 521 estimates the prior probability and the likelihood ratio of the local node at the current moment according to the global decision result of the previous moment and the quantization result obtained by the obtaining module 510.

The data fusion unit 522 obtains the global decision result based on the prior probability and the likelihood ratio of the local node at the current time estimated by the estimation unit 521. The estimating unit 521 includes:

The accumulation sub-unit 523 accumulates the decision status values of the previous time, where the decision status value is generated by comparing the global decision result according to the previous time with the quantization result obtained by the acquisition module.

The selecting sub-unit 524 selects the weighting value of each decision state value from the accumulated result of the accumulating sub-unit 523, and calculates the prior probability and the likelihood ratio estimation of the local node by using the weighting values of the respective decision state values. value.

In order to better reflect the performance of the algorithm, it is assumed that the CR system is in a difficult detection environment. If the algorithm can achieve good performance under extremely difficult conditions, it will naturally achieve satisfactory results in a better test environment. The difficulty of detecting the environment is reflected in two aspects: First, the performance of the local detection algorithm is limited, assuming that the local node uses energy detection, and the detection time is only 10,000 samples of duration, when the primary user is a DVB-T digital television signal, The corresponding time is about lms. Second, the signals at all local nodes are in deep fading, and their SNR is distributed between -21dB and -19dB. This SNR range is about the minimum signal-to-noise ratio required by the system.

The performance of the algorithm is measured by the error probability P _{e of the} global decision, which can be expressed as: where /^! ^ and /^!!. ) indicates that the primary user's pending detection signal is a valid signal and the prior probability in the absence condition, we make them both 0.5 in the simulation. The Ρ _σΜ and P _GF distributions represent the global missed detection probability and the global false alarm probability.

The simulation is run under the above assumptions, and the relationship between the error probability and the number of local detection nodes is shown in Fig. 6 and Fig. 7. The "LOQ-x Bit" in Fig. 6 indicates the case where the local optimum quantizer is used, and the "LOQ-x Bit" based on the dynamic range in Fig. 7 quantizes the local detected data into X bits. "Raw Data" means that all the detected data is transmitted to the central node, and the central node uses the minimum error criterion for data fusion. Use "or" and "and" for 1-bit detection statistics The criterion is compared with the data fusion method used in the second embodiment. By comparing the simulations in FIG. 6 and FIG. 7, it can be seen that the cooperative sensing method used in the second embodiment of the present invention is adopted regardless of the quantization method. It is 4 times better than the "or" and "and" criteria. Even if only the local detection result is quantized to 1 bit, only less than 50 local detection nodes are needed, and the global error probability can be controlled to a lower level, which is about 0.1. The cooperative sensing method adopting the "or" and "and" fusion criteria, although the number of nodes increases, its error probability decreases, but the change is very slow, even when the number of nodes increases to 1000, the error probability is still At a high level, about 0.15. This shows that when the detection environment is very poor, because the information of the detection statistics is limited, the synergy gain that the "or" and "and" criteria can provide is also limited, and even if the detection nodes are greatly increased, the detection performance cannot be significantly improved. In the cooperative sensing method in the embodiment of the present invention, since more detection statistic information can be obtained from the local node, the system can meet the detection performance requirement by appropriately increasing the number of local detection nodes.

Those skilled in the art will appreciate that the drawings are only a schematic representation of a preferred embodiment, and that the modules or steps in the drawings are not necessarily required to practice the invention.

The above-mentioned embodiment "steps" does not mean that the process flow of the embodiment has a sequence, and the serial number of the embodiment of the present invention is merely for the description, and does not represent the advantages and disadvantages of the embodiment.

The solution described in the claims is also the scope of protection of the embodiments of the present invention.

One of ordinary skill in the art will appreciate that all or part of the processing of the above-described embodiments can be accomplished by a program that instructs related hardware, which can be stored in a computer readable storage medium.

The above is only the preferred embodiment of the present invention and is not intended to limit the scope of the present invention. Any modifications, equivalents, improvements, etc. made within the spirit and scope of the present invention are intended to be included within the scope of the present invention.

Claims

Rights request

A method for locally detecting a signal, the method comprising:

The method according to claim 1, wherein the determining the quantization threshold corresponding to the local node comprises:

Determining a quantization level according to a transmission overhead requirement between the local node and the central node; determining a quantization threshold corresponding to the local node according to the number of quantization levels of the local node.

The method according to claim 2, wherein the determining that the signal to be detected is a valid signal indicates that the signal to be detected is not a valid signal, and determining the quantization threshold corresponding to the local node includes:

Determining, according to a distribution function, a mean value, or a variance of the detection statistic, a quantization interval that maximizes a deviation between the detection statistic and H ₀ ;

Determining, according to the quantization interval, a quantization threshold corresponding to the local node.

The method according to claim 2, wherein the determining the quantization threshold corresponding to the local node comprises:

Determining, according to the maximum value and the minimum value of the detection statistic, a dynamic range of the detection statistic; dividing the dynamic range by the number of the quantization levels to obtain a quantization interval of the local node; The quantization interval of the local node and the minimum value of the detection statistic determine a quantization threshold corresponding to the local node.

5. A method of detecting a signal at a center, the method comprising:

Obtaining the quantized result obtained by quantifying the detection statistic by the local node; Based on the quantized result, a global decision result indicating whether the signal to be detected is a valid signal is obtained.

The method according to claim 5, wherein the determining, according to the quantized result, a global decision result indicating whether the signal to be detected is a valid signal comprises:

According to the minimum error criterion, data fusion is performed on the quantized result to obtain an estimated value of the global decision result.

The method according to claim 6, wherein the performing data fusion on the quantized result comprises:

Estimating the prior probability and the likelihood ratio of the local node at the current moment according to the global decision result of the previous moment and the quantization result of the local node;

A global decision result is obtained based on the estimated prior probability and likelihood ratio of the local node at the current time.

8. The method according to claim 7, wherein the estimating the prior probability and the likelihood ratio of the local node at the current moment comprises:

And accumulating the decision state values of the previous time, the decision state values being obtained by comparing the quantized result of the local node with the global decision result;

The weighted values of the respective decision state values are selected from the accumulated results, and the estimated values of the prior nodes and the likelihood ratios of the local nodes are calculated by using the weighted values of the respective decision state values.

9. The method according to claim 8, wherein the accumulating the decision status values of the previous time comprises:

Determining the decision status value of the local node at the previous moment by comparing the quantized result and the global decision result at the previous moment;

And accumulating the cumulative value of the decision state of the local node according to the decision state value of the local node at the previous moment, and the cumulative value of the decision state is an accumulation function of the previous state decision state value.

The method according to any one of claims 5-9, wherein the local node is one or more.

11. A local detecting device, wherein the local detecting device comprises:

a detector, configured to detect a detection statistic corresponding to the signal to be detected in the local node, and a quantizer, configured to quantize the detection statistic obtained by the detector according to a quantization threshold corresponding to the local detection device.

The local detecting device according to claim 11, wherein the local detecting device further comprises:

a quantization threshold determiner, configured to determine a quantization level according to a transmission overhead requirement between the local detection device and the central node; and determine a quantization corresponding to the local detection device according to a statistical rule of the detection statistics obtained by the detector Threshold.

13. The local detecting device according to claim 12, wherein H is set. Representing that the to-be-detected signal is a valid signal, indicating that the to-be-detected signal is not a valid signal, and the quantization threshold determiner includes:

a statistical acquisition unit, configured to obtain a distribution function, a mean value, or a variance of the detection statistic; the first quantization threshold determining unit is configured to determine, according to a distribution function, a mean value, or a variance of the detection statistic obtained by the statistical acquisition unit, The detection statistic is at H. And a quantization interval where the deviation is the largest, and the quantization threshold corresponding to the local detecting device is determined according to the quantization interval.

The local detecting device according to claim 12, wherein the quantization threshold determinator comprises:

a dynamic range determining unit, configured to determine a dynamic range of the detection statistic according to a maximum value and a minimum value of the detection statistic;

a second quantization threshold determining unit, configured to use the dynamic range obtained by the dynamic range determining unit The quantization level is divided to obtain a quantization interval of the local detecting device, and the quantization threshold corresponding to the local detecting device is determined according to the local detecting device quantization interval and the minimum value of the detection statistic.

15. A center detecting device, wherein the center detecting device comprises:

An obtaining module, configured to obtain a quantized result of the local detecting device, where the quantized result is a quantized result obtained by quantifying a detection statistic corresponding to the signal to be detected;

The center detecting device according to claim 15, wherein the execution module comprises:

An estimating unit, configured to estimate a prior probability and a likelihood ratio of the local node at the current moment according to the global decision result of the previous moment and the quantization result obtained by the acquiring module;

And a data fusion unit, configured to obtain a global decision result according to a prior probability and a likelihood ratio of the local node at the current moment estimated by the estimating unit.

The center detecting device according to claim 16, wherein the estimating unit comprises:

An accumulation subunit, configured to accumulate a decision state value of a previous time, the decision state value being generated by comparing a global decision balance of a previous time and a quantization result obtained by the obtaining module;

a subunit, configured to select, from the accumulated result of the accumulating subunit, a weighting value of each of the judgment state values, and calculate, by using the weighting value of each of the judgment state values, a prior probability and a similarity of the local node The estimated value of the ratio.

18. A system for detecting signals, comprising:

a local node, configured to determine a quantization threshold corresponding to the local node, to be checked according to the quantization threshold The detection statistic corresponding to the measurement signal is quantized to obtain a quantization result, and the local node is at least one; the central node is configured to obtain a quantization result obtained by the local node to quantize the detection statistic, and according to the quantization result, A global decision result characterizing whether the signal to be detected is a valid signal.