A method for estimating noise power by oversampling

Abstract

The present invention discloses a method for estimating noise power through oversampling, which oversamples the signal received by a cognitive user in a time slot; then obtains frequency domain signals at various frequency points after discrete Fourier transform of all samples obtained by oversampling, wherein the range of the frequency points is determined by the range of the noise frequency band determined by the number of oversampling times and the oversampling rate and the bandwidth of the intervals taken on both sides of the noise frequency band; and then estimates the noise power according to the power spectral density of the frequency domain signals at various frequency points; the advantage is that the method can accurately estimate the noise power when the main user signal exists.

Description

Technical Field

[0001] The present invention relates to a cognitive radio technology, and in particular to a method for estimating noise power through oversampling, which estimates the size of noise power by oversampling a signal received by a cognitive user. Background Art

[0002] The application and popularization of the fifth generation of wireless communication technology has brought convenience to people's lives, but it has also intensified the demand for spectrum resources. At present, in my country, due to the poor propagation characteristics of millimeter waves, the need for large-scale deployment and high costs, the preferred frequency bands for 5G are still the medium and low frequency bands below 6GHz. However, studies have shown that the spectrum utilization rate of medium and low frequency bands is very low. Therefore, in order to alleviate the spectrum pressure, it is necessary to improve the spectrum utilization rate of medium and low frequency bands. A very effective method is to use cognitive radio technology. Cognitive radio technology allows cognitive users to use frequency bands that are not currently used by primary users without affecting the communication of primary users, thereby improving the utilization rate of spectrum. In cognitive radio, a very key technology is spectrum sensing. The role of spectrum sensing is to help cognitive users detect frequency bands that are not currently used by primary users. When cognitive users detect whether the primary user signal exists, some prior knowledge is required, and noise power is one of them.

[0003] Most of the existing noise power estimation methods estimate the noise power by calculating the eigenvalues ​​in the covariance matrix of the signal received by the cognitive user, such as the minimum eigenvalue or the average of several eigenvalues ​​except the maximum eigenvalue. The main problem of these methods is that when the primary user signal exists, the estimated noise power will be too large, which will affect the spectrum detection result. Summary of the invention

[0004] The technical problem to be solved by the present invention is to provide a method for estimating noise power by oversampling, which can accurately estimate the noise power when a primary user signal exists.

[0005] The technical solution adopted by the present invention to solve the above technical problems is: a method for estimating noise power by oversampling, characterized in that its processing process is: oversampling the signal received by the cognitive user in a time slot; then obtaining all the samples obtained by oversampling and performing discrete Fourier transform to obtain frequency domain signals at each frequency point, the range of the frequency point is determined by the range of the noise frequency band determined by the number of oversampling times and the oversampling rate and the bandwidth of the intervals taken on both sides of the noise frequency band; and then estimating the noise power according to the power spectral density of the frequency domain signal at each frequency point.

[0006] The method specifically comprises the following steps:

[0007] Step 1: In the cognitive radio system, randomly select a time slot and perform L oversampling on the signal received by the cognitive user in the time slot to obtain L samples. The i-th sample is recorded as x(i); where L is a positive integer, L = 2 n , n is a positive integer, n≥7, the oversampling rate is K, K is a positive integer, K≥2, i is a positive integer, 1≤i≤L;

[0008] Step 2: According to the oversampling times L and the oversampling rate K, determine the range of the noise band: And take a part of the band with bandwidth U on both sides of the noise band as the interval; then according to the range of the noise band The bandwidth U of the interval is used to obtain the frequency domain signal at frequency point m after L samples are discrete Fourier transformed, which is recorded as X(m). Among them, the value range of U is 20Hz~50Hz, and the unit of m is Hz. e is the natural cardinality, j is the imaginary unit;

[0009] Step 3: Estimate the noise power based on the power spectral density of the frequency domain signal at all frequency points, denoted as The symbol “||” is the modulo operator, |X(m)| 2 represents the power spectral density of X(m).

[0010] Compared with the prior art, the advantages of the present invention are:

[0011] 1) After the method of the present invention oversamples the signal received by the cognitive user, only the power spectral density of the noise signal remains in the middle, and the power spectral density of the primary user signal superimposed on the noise signal is distributed on both sides. In this way, the noise power can be estimated based on the power spectral density of the noise signal remaining in the middle, so that the method of the present invention can also accurately estimate the noise power when the primary user signal exists.

[0012] 2) A portion of the frequency band is taken on both sides of the noise band as an interval, that is, the interval on both sides forms a protection band, which can avoid the influence of the main user signal leaking into the noise band on the noise power estimation, thereby making the noise power estimation more accurate.

[0013] 3) Compared with the estimation method based on eigenvalues, the method of the present invention does not require eigenvalue decomposition and has low computational complexity. BRIEF DESCRIPTION OF THE DRAWINGS

[0014] Figure 1 It is a flowchart of the method of the present invention;

[0015] Figure 2 is a power spectrum density diagram obtained after oversampling the signal received by the cognitive user;

[0016] Figure 3 The figure is a performance curve comparison diagram of the noise power estimation value and the actual value obtained by using the method of the present invention and the existing eigenvalue-based estimation method respectively as the system signal-to-noise ratio changes. DETAILED DESCRIPTION

[0017] The present invention is further described in detail below with reference to the accompanying drawings.

[0018] The present invention proposes a method for estimating noise power by oversampling, and its flow chart is as follows: Figure 1 As shown in the figure, the processing process is as follows: oversample the signal received by the cognitive user in a time slot; then obtain the frequency domain signals at each frequency point after discrete Fourier transform of all the samples obtained by oversampling, and the range of the frequency point is determined by the range of the noise band determined by the oversampling times and the oversampling rate and the bandwidth of the intervals on both sides of the noise band; then estimate the noise power according to the power spectral density of the frequency domain signal at each frequency point.

[0019] The method specifically comprises the following steps:

[0020] Step 1: In the cognitive radio system, randomly select a time slot and perform L oversampling on the signal received by the cognitive user in the time slot to obtain L samples. The i-th sample is recorded as x(i); where L is a positive integer, L = 2 n , n is a positive integer, n≥7, in this embodiment, L=2 10 =1024, the oversampling rate is K, K is a positive integer, K≥2, K is generally a positive integer greater than or equal to 2. In real scenarios, due to the limitations of hardware complexity and cost, the value of K will not be too large. In this embodiment, K=4, K is much smaller than L, i is a positive integer, 1≤i≤L.

[0021] Step 2: According to the oversampling times L and the oversampling rate K, determine the range of the noise band: And take a part of the band with bandwidth U on both sides of the noise band as the interval; then according to the range of the noise band The bandwidth U of the interval is used to obtain L samples (discrete time domain signals) and the frequency domain signal at frequency point m is obtained after discrete Fourier transform, which is recorded as X(m). The value range of U is 20 Hz to 50 Hz. In this embodiment, U is 50 Hz. The unit of m is Hz. e is the natural cardinality, e=2.71…, j is the imaginary unit, π=3.14….

[0022] Step 3: Estimate the noise power based on the power spectral density of the frequency domain signal at all frequency points, denoted as Among them, 2U is actually the bandwidth of the protection band. In order to avoid the influence of the main user signal leaking into the noise band on the noise power estimation, the symbol “||” is the modulo operator, |X(m) 2 represents the power spectral density of X(m).

[0023] If the signal received by the cognitive user is not oversampled, the frequency domain signals of the noise signal and the primary user signal are mixed together, and at each frequency point there is the primary user signal plus the noise signal. After the signal received by the cognitive user is oversampled, only the power spectral density of the noise signal remains in the middle, and the power spectral density of the primary user signal superimposed on the noise signal is distributed on both sides, such as Figure 2 The power spectral density diagram shown is obtained after oversampling the signal received by the cognitive user, so that the noise power can be estimated based on the power spectral density of the noise signal remaining in the middle.

[0024] The feasibility and effectiveness of the method of the present invention are further illustrated by the following simulation.

[0025] In the simulation, the number of oversampling times L = 1024, the oversampling rate K = 4, and the sampling frequency f s Set to 1000 Hz, the system signal-to-noise ratio (the ratio of the power of the main user signal to the power of the noise signal) varies from -10 dB to 20 dB. Figure 3 A performance curve comparison diagram of the noise power estimation value and the actual value obtained by the method of the present invention and the existing eigenvalue-based estimation method as the system signal-to-noise ratio changes is given. Figure 3 It can be seen that as the system signal-to-noise ratio increases, the difference between the noise power estimation value obtained by the existing eigenvalue-based estimation method and the actual value gradually decreases and tends to a constant, but it is not accurate enough, while the noise power estimation value obtained by the method of the present invention is almost equal to the actual value. Therefore, the noise power estimation performance of the method of the present invention is better than the existing eigenvalue-based estimation method. The existing eigenvalue-based estimation method is cited from Y. Zeng and YC Liang, Eigenvalue-based spectrum sensing algorithms for cognitive radio [J]. IEEE Transactions on Communications, 2009, 57 (6): 1784–1793. (Eigenvalue-based spectrum sensing algorithms for cognitive radio).

Claims

1. A method for estimating noise power by oversampling, characterized in that Its processing process is as follows: oversample the signal received by the cognitive user in a time slot; then obtain the frequency domain signals at each frequency point after discrete Fourier transform of all the samples obtained by oversampling. The range of the frequency points is determined by the range of the noise band determined by the number of oversampling times and the oversampling rate and the bandwidth of the intervals on both sides of the noise band; then estimate the noise power based on the power spectral density of the frequency domain signals at each frequency point; The method specifically comprises the following steps: Step 1: In the cognitive radio system, randomly select a time slot and perform L oversampling on the signal received by the cognitive user in the time slot to obtain L samples. The i-th sample is recorded as x(i); where L is a positive integer, L = 2 n , n is a positive integer, n≥7, the oversampling rate is K, K is a positive integer, K≥2, i is a positive integer, 1≤i≤L; Step 2: According to the oversampling times L and the oversampling rate K, determine the range of the noise band: And take a part of the band with bandwidth U on both sides of the noise band as the interval; then according to the range of the noise band The bandwidth U of the interval is used to obtain the frequency domain signal at frequency point m after L samples are discrete Fourier transformed, which is recorded as X(m). Among them, the value range of U is 20Hz~50Hz, and the unit of m is Hz. e is the natural cardinality, j is the imaginary unit; Step 3: Estimate the noise power based on the power spectral density of the frequency domain signal at all frequency points, denoted as The symbol "||" is the modulus operator, |X(m)| 2 represents the power spectral density of X(m).

Images