Mixed interference suppression method based on forward continuous mean removal spectrum sensing

Abstract

The invention discloses a mixed interference suppression method based on FCME spectrum sensing. Spectrum sensing based on FCME; a mixed interference suppression step; and a balanced demodulation decoding step. A hybrid interference suppression module group is constructed on the basis of spectrum sensing, and different interference suppression algorithms are selected for different modules and combined. The method provided by the invention has strong spectrum sensing capability under complex mixed interference, and is more accurate than other methods in interference suppression, and the decoding module further enhances the anti-interference robustness.

Description

Technical Field

[0001] The invention relates to hybrid interference suppression based on wireless broadband communication, belongs to the technical field of communication signal processing, and specifically relates to a hybrid interference suppression method based on forward continuous mean excision spectrum perception. Background Art

[0002] With the rapid popularization of wireless broadband devices, wireless services have penetrated into all aspects of people's lives and become increasingly important as an important part of the telecommunications infrastructure in society. However, security threats have become a big problem for the confidentiality, integrity and availability of wireless communications. Compared with other security threats such as eavesdropping and data forgery, wireless networks are particularly vulnerable to radio interference attacks.

[0003] With the advancement of software-defined radio, people can easily program a small 100 yuan USB device as a jammer, covering 20MHz bandwidth below 6GHz and up to 100mW of transmission power. The convenience of launching interference attacks makes it urgent to protect wireless networks from intentional and unintentional interference threats. Especially in military communication scenarios, complex battlefield environments mean more complex electromagnetic environments, and military communication systems will inevitably be subject to complex electromagnetic interference. At the same time, the enemy's intentional electromagnetic interference will lead to a decline in military communication performance. Therefore, the ability to resist complex electromagnetic interference is an enabling technology that both civilian and military wireless communications need to have.

[0004] For wireless broadband communication systems, interference can be divided into single-tone interference and partial-band interference according to the frequency domain characteristics of the signal. Common narrowband interference and broadband interference can be classified as partial-band interference, and multi-tone interference can be considered as the superposition of several single-tone interferences. Mixed interference is the superposition of several interferences.

[0005] At present, there are few literatures on hybrid interference suppression technology. Current research is mainly based on GNSS systems, and the main shortcomings are

[0006] (1) Mixed interference is relatively simple, including the mixture of single-tone interference and pulse interference, and the mixture of pulse interference and narrowband interference.

[0007] (2) Lack of spectrum perception means that only specific interference can be detected and identified. Summary of the invention

[0008] In view of the shortcomings of the existing mixed interference suppression methods, the present invention proposes a mixed interference suppression method based on forward consecutive mean excision (FCME) spectrum sensing. A mixed interference suppression method based on forward consecutive mean excision spectrum sensing specifically includes the following steps:

[0009] Step 1: Spectrum sensing based on forward continuous mean excision FCME;

[0010] Assume that the receiver receives a baseband signal r, which is composed of the desired signal d, Gaussian white noise w and interference j, that is,

[0011] r=d+w+j (1)

[0012] The received signal is represented by r = [r(1)...r(N)] T , where N represents the length of a received signal frame, r(1)...r(N) represent the 1st to Nth symbols in a frame signal respectively; the received signal is transformed into the frequency domain by FFT, which is expressed as

[0013]

[0014] Among them, each value of R(k) is called a frequency bin, and k represents the index of the frequency bin;

[0015] Calculate the amplitude spectrum of the received signal,

[0016]

[0017] Where Re and Im represent the real and imaginary operations respectively; the amplitude spectrum Ψ = {Ψ(k)|k∈I1} is sent to the FCME spectrum sensing as the output block; Assume I m and J m They represent the index of the frequency bin where no interference is detected at the mth iteration and the set of indexes of the frequency bin where interference is detected; then, the index set I1 is the set of indexes of the frequency bin where no interference is detected at the first iteration, that is, I1 contains all indexes 1,...,N; in FCME spectrum sensing, the index of the frequency bin where interference is detected is from the first index set I m Move to the second index set J m , this operation is completed once in each iteration; the initial assumption is that no frequency bin is disturbed, that is, J1 is an empty set; when the maximum number of iterations is reached or no new disturbed frequency bin is found, the iteration is stopped;

[0018] Step 2: Mixed interference suppression step;

[0019] According to the spectrum sensing results, the hybrid interference suppression module group is determined; the interference suppression algorithm includes a time domain notch filter and a frequency domain threshold removal filter, and the time domain notch filter is hereinafter referred to as a "notch filter"; for single-tone interference, a notch filter is suitable for interference suppression; when using a second-order notch filter, the received signal {r(n)|n=1,...,N} is sent to the notch filter as input, and assuming that the output signal is x(n), the system function H(z) of the notch filter is expressed as

[0020]

[0021] Among them, X(z) and R(z) are the Z-transforms of x(n) and r(n) respectively; δ and μ are the two gain coefficients of the second-order notch filter. If the absolute values ​​of δ and μ are both less than 1, the notch filter is stable; these two coefficients determine the -3dB bandwidth B and the notch frequency ω respectively. N , expressed as

[0022]

[0023] μ=cos(ω N ),ω N ∈[0,π] (6)

[0024] Among them, bandwidth B and notch frequency ω N Determined based on spectrum sensing results;

[0025] For partial frequency band interference, the frequency domain threshold removal algorithm is used to suppress interference, as follows;

[0026] Assume that the sum of the expected signal and Gaussian white noise is v = d + w = ​​{v(n) | n = 1, ..., N}, and the interference is j = {j(n) | n = 1, ..., N}, where v(n) and j(n) represent their nth symbols respectively, and n is 1 to N. When there is interference, the frequency domain amplitude value of the received signal is

[0027]

[0028] Where J(k) and V(k) represent the Fourier transform FFT of j(n) and v(n) respectively; when there is no interference, the frequency domain amplitude value of the received signal is

[0029]

[0030] Define two hypotheses H0 and H1, respectively, that there is no interference and that there is interference, and their corresponding observation values ​​are

[0031]

[0032] Both observations obey Rayleigh distribution; divide the observation space according to the maximum posterior probability: define the posterior probabilities of the two hypotheses as P(H1|M(k)) and P(H0|M(k)), where P(·) represents the probability of the event, and P(H1|M(k)) and P(H0|M(k)) represent the probability of H1 or H0 occurring under the condition of observing M(k); compare the sizes of the two posterior probabilities and determine that the hypothesis with the larger posterior probability is valid, that is,

[0033]

[0034] The above formula indicates that if the greater than sign is established, then hypothesis H1 is established, otherwise H0 is established; according to the Bayesian formula,

[0035] P(H1|M(k))=P(M(k)|H1)P(H1) / P(M(k)) (12)

[0036] P(H0|M(k))=P(M(k)|H0)P(H0) / P(M(k)) (13)

[0037] Substituting equations (18) and (19) into (17), we obtain

[0038]

[0039] Since J(k)+V(k) and V(k) both obey complex Gaussian distribution,

[0040]

[0041] In the formula, and are the variances of J(k) and V(k), respectively; Substituting (21) into (20), we get

[0042]

[0043] When H1 is true, it means that interference exists, and the value of R(k) for this frequency bin is 0; when H0 is true, it means that interference does not exist, and no processing is done on this frequency bin;

[0044] Step 3: Equalization demodulation and decoding step;

[0045] Assuming ideal synchronization, the frequency domain R0(k) of the received signal after interference suppression satisfies

[0046] R0(k)=S(k)H(k)+W(k)+I0(k) (17)

[0047] Where S(k), H(k) and W(k) represent the FFT of the transmitted signal s(n), the channel impulse response and the noise respectively, and I0(k) represents the residual interference; the output result of frequency domain equalization is satisfy

[0048]

[0049] right Do iFFT transformation to get the balanced time domain for,

[0050]

[0051] in, is used for demodulation;

[0052] Error correction coding is used to further reduce the bit error rate.

[0053] In one embodiment of the present invention, in step 1, the iterative processing flow is as follows:

[0054] Step 1: Initialization, set m = 1, and J m ,I m , S m and N m Assign initial value, where J m is an empty set, I m Contains 1~N, S m is the sum of the amplitude spectrum, N m It is a set I m The number of elements in ;

[0055] Step 2: Iterative processing, including two layers of loops. The first layer of loop first calculates S in the current loop. m and N m Then, enter the second loop, the second loop traverses the current I m The frequency bin corresponding to the index contained, if the amplitude spectrum Ψ(k) is greater than the threshold T*S m / N m , T represents the threshold parameter, then the index k corresponding to this frequency bin is increased from I m Move to J m ;

[0056] The threshold parameter T used in the iterative process is determined theoretically based on the statistics of the non-interfering received signal; the complex Gaussian noise and the desired signal still obey the complex Gaussian distribution after FFT transformation, so the amplitude spectrum of the received signal is approximately a Rayleigh distribution with two degrees of freedom; using these assumptions, the threshold parameter T can be determined theoretically in the following way; the first-order moment of the two-degree-of-freedom Rayleigh distribution random variable Ψ is

[0057]

[0058] In the formula, σ 2 is the variance of a zero-mean statistically independent Gaussian random variable; the Γ function is defined as

[0059]

[0060] Where e is a natural constant; the cumulative density function of the Rayleigh distribution random variable is

[0061]

[0062] Where ψ is the independent variable, and solving equation (6) for ψ yields

[0063]

[0064] In the formula, ln is the natural logarithm; in FCME spectrum sensing, all values ​​ψ in the current amplitude spectrum are calculated according to the threshold T*S m / N m =T*E(Ψ) is divided into two groups, one for the frequency bins with interference and the other for the frequency bins without interference; let the target threshold ψ target =T*E(Ψ), then

[0065]

[0066] It can be seen that the threshold parameter T has nothing to do with the noise variance; the parameter F(ψ target ) is the target cumulative density function value, which indicates the relative number of observations contained in the expected set;

[0067] Set the interference indicator variable a(k) and express it in binary; assign a(k) a value according to whether there is interference in the frequency bin, that is,

[0068]

[0069] After iterative processing, a(k) is the spectrum sensing result.

[0070] In another embodiment of the present invention, the threshold parameter T used in the iterative process is based on the mean or variance.

[0071] In a specific embodiment of the present invention, the error correction coding adopts RS coding or low-density parity check LDPC coding.

[0072] This method builds a hybrid interference suppression module group based on spectrum sensing, selects different interference suppression algorithms for different modules and combines them. The method of the invention has strong spectrum sensing ability under complex mixed interference, and is more accurate in interference suppression than other methods. The decoding module further enhances the anti-interference robustness.

[0073] The method of the present invention can provide support for mixed interference suppression in complex electromagnetic environments, and effectively solve the problem of poor performance in suppressing mixed interference in civil and military scenarios. BRIEF DESCRIPTION OF THE DRAWINGS

[0074] Figure 1 A flow chart showing a hybrid interference suppression method based on FCME spectrum sensing of the present invention is shown;

[0075] Figure 2 shows a sample mixed interference spectrum diagram;

[0076] Figure 3 The structure of the hybrid interference suppression module group designed by the present invention is shown;

[0077] Figure 4 The output result of hybrid interference suppression performed by the present invention on an actual wireless broadband signal is shown;

[0078] Figure 5 The equalization demodulation and decoding process designed by the present invention is shown. DETAILED DESCRIPTION

[0079] The flow chart of the hybrid interference suppression method based on FCME spectrum sensing of the present invention is as follows: Figure 1 As shown, the four steps include spectrum sensing based on FCME, mixed interference suppression, equalization demodulation, and decoding, which are specifically as follows.

[0080] Step 1: Spectrum sensing based on FCME;

[0081] The spectrum sensing process based on FCME is as follows: Figure 1 As shown. Assume that the receiver receives a baseband signal r. It is composed of the desired signal d, Gaussian white noise w and interference j, that is,

[0082] r=d+w+j (26)

[0083] The received signal can be expressed as r = [r(1)...r(N)] T , where N represents the length of a received signal frame, and r(1)...r(N) represent the 1st to Nth symbols in a frame of signal. First, the received signal is transformed into the frequency domain through FFT, which can be expressed as

[0084]

[0085] Each value of R(k) is called a frequency bin, and k represents the index of the frequency bin. Then, the amplitude spectrum of the received signal is calculated.

[0086]

[0087] Where Re and Im represent the real and imaginary operations respectively. The amplitude spectrum Ψ = {Ψ(k)|k∈I1} is fed into the FCME spectrum sensing as the output block. Assume I m and J m The indexes of the frequency bins where no interference was detected at the mth iteration and the indexes of the frequency bins where interference was detected are respectively represented. Then, the index set I1 is the set of indexes of the frequency bins where no interference was detected at the first iteration, that is, I1 contains all indexes 1,...,N. In FCME spectrum sensing, the indexes of the frequency bins where interference was detected are added from the first index set I m Move to the second index set J mThis operation is completed once in each iteration. The initial assumption is that no frequency bin is disturbed, that is, J1 is an empty set. When the maximum number of iterations is reached or no new disturbed frequency bin is found, the iteration stops. The iterative processing flow is as follows:

[0088] Step 1: Initialization, set m = 1, and J m ,I m , S m and N m Assign initial value, where J m is an empty set, I m Contains 1~N, S m is the sum of the amplitude spectrum, N m It is a set I m The number of elements.

[0089] Step 2: Iterative processing, including two layers of loops. The first layer of loop first calculates S in the current loop. m and N m Then, the second loop is entered, and the second loop traverses the current I m The frequency bin corresponding to the index contained, if the amplitude spectrum Ψ(k) is greater than the threshold T*S m / N m , T represents the threshold parameter, then the index k corresponding to this frequency bin is changed from I m Move to J m .

[0090] The threshold parameter T used in the iterative process can be determined theoretically based on the statistics of the non-interfering received signal, such as the mean, variance, etc. The complex Gaussian noise and the desired signal still obey the complex Gaussian distribution after FFT transformation, so the amplitude spectrum of the received signal is approximately a Rayleigh distribution with two degrees of freedom. Using these assumptions, the threshold parameter T can be determined theoretically in the following way. The first-order moment of the two-degree-of-freedom Rayleigh distribution random variable Ψ is

[0091]

[0092] In the formula, σ 2 is the variance of a zero-mean statistically independent Gaussian random variable. The Γ function is defined as

[0093]

[0094] Where e is a natural constant. The cumulative density function of a Rayleigh distribution random variable is

[0095]

[0096] Where ψ is the independent variable, equation (6) is solved for ψ to obtain

[0097]

[0098] In the formula, ln is the natural logarithm. In FCME spectrum sensing, all values ​​ψ in the current amplitude spectrum are calculated according to the threshold T*S m / N m =T*E(Ψ) is divided into two groups, one for the frequency bins with interference and the other for the frequency bins without interference. Let the target threshold ψ target =T*E(Ψ), then

[0099]

[0100] It can be seen from formula (7) that the threshold parameter T has nothing to do with the noise variance. target ) is the target cumulative density function value, which indicates the relative number of observations contained in the expected set. target ) is set to 0.99, which means that on average 1% of the frequency bins are falsely detected as being interfered with.

[0101] Set the interference indicator variable a(k) in binary. Assign a(k) a value based on whether there is interference in the frequency bin, i.e.

[0102]

[0103] After iterative processing, a(k) is the spectrum sensing result. Figure 2 Shows examples of spectrum sensing results, Figure 2 (a) is an example of the amplitude spectrum of the received signal when mixed interference exists. Figure 2 (b) shows the corresponding spectrum sensing results.

[0104] Step 2: Mixed interference suppression step;

[0105] The steps of mixed interference suppression are as follows Figure 3 As shown. In this step, the hybrid interference suppression module group is determined according to the spectrum sensing results. The interference suppression algorithm includes a time domain notch filter (hereinafter referred to as "notch filter"), a frequency domain threshold cutoff filter, etc. For single-tone interference, it is suitable to use a notch filter for interference suppression. The second-order notch filter used in the invention is described below. The received signal {r(n)|n=1,...,N} is sent as input to the notch filter. Assuming that the output signal is x(n), the system function H(z) of the notch filter is expressed as

[0106]

[0107] Where X(z) and R(z) are the Z-transforms of x(n) and r(n) respectively. δ and μ are the two gain coefficients of the second-order notch filter. If the absolute values ​​of δ and μ are both less than 1, the notch filter is stable. These two coefficients determine the -3dB bandwidth B and the notch frequency ω respectively. N , which can be expressed as

[0108]

[0109] μ=cos(ω N ),ω N ∈[0,π] (37)

[0110] Among them, bandwidth B and notch frequency ω N The determination is made based on spectrum sensing results.

[0111] For partial frequency band interference, a frequency domain threshold excision algorithm is used to suppress interference. The key to frequency domain threshold excision is to detect interference by setting a threshold, and then remove the frequency bin where the interference exists to complete interference suppression. When performing interference detection, the present invention uses a signal detection technology based on hypothesis testing to perform accurate interference detection. According to the spectrum sensing results, it is only necessary to perform detection within the partial frequency band interference band. The specific algorithm is described as follows.

[0112] Assume that the sum of the expected signal and Gaussian white noise is v = d + w = ​​{v(n) | n = 1, ..., N}, and the interference is j = {j(n) | n = 1, ..., N}, where v(n) and j(n) represent their nth symbols respectively, and n is 1 to N. When there is interference, the frequency domain amplitude value of the received signal is

[0113]

[0114] Where J(k) and V(k) represent the Fourier transform (FFT) of j(n) and v(n) respectively. When there is no interference, the frequency domain amplitude value of the received signal

[0115]

[0116] Define two hypotheses H0 and H1, respectively, that there is no interference and that there is interference, and their corresponding observation values ​​are

[0117]

[0118] Both observations obey Rayleigh distribution. Divide the observation space according to the maximum posterior probability: First, define the posterior probabilities of the two hypotheses as P(H1|M(k)) and P(H0|M(k)), where P(·) represents the probability of the event, and P(H1|M(k)) and P(H0|M(k)) represent the probability of H1 or H0 occurring under the condition of observing M(k). Compare the sizes of the two posterior probabilities and determine that the hypothesis with the larger posterior probability is valid, that is,

[0119]

[0120] The above formula indicates that if the greater than sign is true, then hypothesis H1 is true, otherwise H0 is true. According to the Bayesian formula,

[0121] P(H1|M(k))=P(M(k)|H1)P(H1) / P(M(k)) (43)

[0122] P(H0|M(k))=P(M(k)|H0)P(H0) / P(M(k)) (44)

[0123] Substituting equations (18) and (19) into (17), we obtain

[0124]

[0125] Since J(k)+V(k) and V(k) both obey complex Gaussian distribution,

[0126]

[0127] In the formula, and are the variances of J(k) and V(k), respectively. Substituting (21) into (20), we get

[0128]

[0129] When H1 is true, it means that interference exists, and the value of R(k) for this frequency bin is 0; when H0 is true, it means that interference does not exist, and this frequency bin is not processed. Figure 4 is the output result of hybrid interference suppression for actual wireless broadband signals, where Figure 4 (a) represents the amplitude spectrum before interference suppression, Figure 4 (b) shows the amplitude spectrum after interference suppression.

[0130] Step 3: Equalization demodulation and decoding step;

[0131] The flow chart of equalization demodulation and decoding steps is as follows Figure 5As shown. The equalization demodulation and decoding steps include FFT transformation, frequency domain equalization, demodulation, iFFT transformation and error correction decoding. Assuming ideal synchronization, the frequency domain R0(k) of the received signal after interference suppression satisfies

[0132] R0(k)=S(k)H(k)+W(k)+I0(k) (48)

[0133] Where S(k), H(k) and W(k) represent the FFT of the transmitted signal s(n), the channel impulse response and the noise respectively, and I0(k) represents the residual interference. This term exists because interference suppression cannot eliminate all interference, and residual interference is inevitable. Then the output result of frequency domain equalization is satisfy

[0134]

[0135] right Perform iFFT transformation to obtain the equalized time domain result for

[0136]

[0137] in is used for demodulation.

[0138] Although there are interference suppression steps, good performance cannot be achieved after interference suppression in many complex scenarios. At this time, it is necessary to rely on error correction coding ("error correction coding" is well known to those skilled in the art) to further reduce the bit error rate and improve the performance of the system. The effective design of error correction coding has been proven to be extremely important for wireless broadband systems to achieve acceptable bit error rates. RS coding and low-density parity check LDPC coding have good error correction performance. Therefore, the present invention adopts these two error correction codes. LDPC coding is close to Shannon capacity on additive white Gaussian noise (AWGN) and many other channel types. Depending on the actual situation, you can choose to use a single code, or a cascade of two error correction codes.

Claims

1. A hybrid interference suppression method based on forward continuous mean excision spectrum sensing, characterized in that: Specifically include the following steps: Step 1: Spectrum sensing based on forward continuous mean excision FCME; Assume that the receiver receives a baseband signal r, which is composed of the desired signal d, Gaussian white noise w and interference j, that is, r=d+w+j (1) The received signal is represented by r = [r(1)...r(N)] T , where N represents the length of a received signal frame, r(1)...r(N) represent the 1st to Nth symbols in a frame signal respectively; the received signal is transformed into the frequency domain by FFT, which is expressed as Among them, each value of R(k) is called a frequency bin, and k represents the index of the frequency bin; Calculate the amplitude spectrum of the received signal, Where Re and Im represent the real and imaginary operations respectively; the amplitude spectrum Ψ = {Ψ(k)|k∈I1} is sent to the FCME spectrum sensing as the output block; Assume I m and J m They represent the index of the frequency bin where no interference is detected at the mth iteration and the set of indexes of the frequency bin where interference is detected; then, the index set I1 is the set of indexes of the frequency bin where no interference is detected at the first iteration, that is, I1 contains all indexes 1,...,N; in FCME spectrum sensing, the index of the frequency bin where interference is detected is from the first index set I m Move to the second index set J m , this operation is completed once in each iteration; the initial assumption is that no frequency bin is disturbed, that is, J1 is an empty set; when the maximum number of iterations is reached or no new disturbed frequency bin is found, the iteration is stopped; Step 2: Mixed interference suppression step; According to the spectrum sensing results, the hybrid interference suppression module group is determined; the interference suppression algorithm includes a time domain notch filter and a frequency domain threshold removal filter, and the time domain notch filter is hereinafter referred to as a "notch filter"; for single-tone interference, a notch filter is suitable for interference suppression; when using a second-order notch filter, the received signal {r(n)|n=1,...,N} is sent to the notch filter as input, and assuming that the output signal is x(n), the system function H(z) of the notch filter is expressed as Among them, X(z) and R(z) are the Z-transforms of x(n) and r(n) respectively; δ and μ are the two gain coefficients of the second-order notch filter. If the absolute values ​​of δ and μ are both less than 1, the notch filter is stable; these two coefficients determine the -3dB bandwidth B and the notch frequency ω respectively. N , expressed as μ=cos(ω) N ),oh N ∈[0,π] (6) Among them, bandwidth B and notch frequency ω N Determined based on spectrum sensing results; For partial frequency band interference, the frequency domain threshold removal algorithm is used to suppress interference, as follows; Assume that the sum of the expected signal and Gaussian white noise is v = d + w = ​​{v(n) | n = 1, ..., N}, and the interference is j = {j(n) | n = 1, ..., N}, where v(n) and j(n) represent their nth symbols respectively, and n is 1 to N. When there is interference, the frequency domain amplitude value of the received signal is Where J(k) and V(k) represent the Fourier transform FFT of j(n) and v(n) respectively; when there is no interference, the frequency domain amplitude value of the received signal is Define two hypotheses H0 and H1, respectively, that there is no interference and that there is interference, and their corresponding observation values ​​are Both observations obey Rayleigh distribution; divide the observation space according to the maximum posterior probability: define the posterior probabilities of the two hypotheses as P(H1|M(k)) and P(H0|M(k)), where P(·) represents the probability of the event, and P(H1|M(k)) and P(H0|M(k)) represent the probability of H1 or H0 occurring under the condition of observing M(k); compare the sizes of the two posterior probabilities and determine that the hypothesis with the larger posterior probability is valid, that is, The above formula indicates that if the greater than sign is established, then hypothesis H1 is established, otherwise H0 is established; according to the Bayesian formula, P(H1|M(k))=P(M(k)|H1)P(H1) / P(M(k)) (12) P(H0|M(k))=P(M(k)|H0)P(H0) / P(M(k)) (13) Substituting equations (18) and (19) into (17), we obtain Since J(k)+V(k) and V(k) both obey complex Gaussian distribution, In the formula, and are the variances of J(k) and V(k), respectively; Substituting (21) into (20), we get When H1 is true, it means that interference exists, and the value of R(k) for this frequency bin is 0; when H0 is true, it means that interference does not exist, and no processing is done on this frequency bin; Step 3: Equalization demodulation and decoding step; Assuming ideal synchronization, the frequency domain R0(k) of the received signal after interference suppression satisfies R0(k)=S(k)H(k)+W(k)+I0(k) (17) Where S(k), H(k) and W(k) represent the FFT of the transmitted signal s(n), the channel impulse response and the noise respectively, and I0(k) represents the residual interference; the output result of frequency domain equalization is satisfy right Perform iFFT transformation to obtain the equalized time domain result for, in, is used for demodulation; Error correction coding is used to further reduce the bit error rate.

2. The hybrid interference suppression method based on FCME spectrum sensing according to claim 1, characterized in that: In step 1, the iterative processing flow is as follows: Step 1: Initialization, set m = 1, and J m ,I m , S m and N m Assign initial value, where J m is an empty set, I m Contains 1~N, S m is the sum of the amplitude spectrum, N m It is a set I m The number of elements in ; Step 2: Iterative processing, including two layers of loops. The first layer of loop first calculates S in the current loop. m and N m Then, enter the second loop, the second loop traverses the current I m The frequency bin corresponding to the index contained, if the amplitude spectrum Ψ(k) is greater than the threshold T*S m / N m , T represents the threshold parameter, then the index k corresponding to this frequency bin is increased from I m Move to J m ; The threshold parameter T used in the iterative process is determined theoretically based on the statistics of the non-interfering received signal; the complex Gaussian noise and the desired signal still obey the complex Gaussian distribution after FFT transformation, so the amplitude spectrum of the received signal is approximately a Rayleigh distribution with two degrees of freedom; using these assumptions, the threshold parameter T can be determined theoretically in the following way; the first-order moment of the two-degree-of-freedom Rayleigh distribution random variable Ψ is In the formula, σ 2 is the variance of a zero-mean statistically independent Gaussian random variable; the Γ function is defined as Where e is a natural constant; the cumulative density function of the Rayleigh distribution random variable is Where ψ is the independent variable, and solving equation (6) for ψ yields In the formula, ln is the natural logarithm; in FCME spectrum sensing, all values ​​ψ in the current amplitude spectrum are calculated according to the threshold T*S m / N m =T*E(Ψ) is divided into two groups, one for the frequency bins with interference and the other for the frequency bins without interference; let the target threshold ψ target =T*E(Ψ), then It can be seen that the threshold parameter T has nothing to do with the noise variance; the parameter F(ψ target ) is the target cumulative density function value, which indicates the relative number of observations contained in the expected set; Set the interference indicator variable a(k) and express it in binary; assign a(k) a value according to whether there is interference in the frequency bin, that is, After iterative processing, a(k) is the spectrum sensing result.

3. The hybrid interference suppression method based on FCME spectrum sensing as claimed in claim 2, characterized in that: The threshold parameter T used in the iterative process is based on the mean or variance.

4. The hybrid interference suppression method based on FCME spectrum sensing as claimed in claim 1, characterized in that: The error correction coding adopts RS coding or low-density parity check LDPC coding.

Images