CN111181880B

CN111181880B - Phase noise compensation method and system of wireless communication system based on integrated clustering

Info

Publication number: CN111181880B
Application number: CN202010006669.7A
Authority: CN
Inventors: 谢宁; 胡天星
Original assignee: Shenzhen University
Current assignee: Shenzhen University
Priority date: 2020-01-03
Filing date: 2020-01-03
Publication date: 2022-09-02
Anticipated expiration: 2040-01-03
Also published as: CN111181880A

Abstract

The present disclosure relates to a phase noise compensation method of a wireless communication system based on integrated clustering, which includes: the transmitting end transmits a carrier signal to a wireless channel based on channel coding, baseband modulation and radio frequency modulation and generates a receiving signal to be received by the receiving end, the receiving end obtains a baseband signal from the transmitting end based on radio frequency demodulation and a phase-locked loop circuit and obtains a gain baseband signal based on automatic gain control, further, a plurality of clusters corresponding to each standard constellation point and a plurality of received sample points and a plurality of corresponding cluster central points are obtained based on the cluster model, the norm distance between any cluster central point and each standard constellation point is obtained based on the distance calculation model, further determining a target constellation point, further replacing the coordinates of the sample point corresponding to each cluster to the coordinates of the target constellation point corresponding to the cluster based on a cluster mapping model to realize phase noise compensation, and then obtaining a target receiving signal, wherein the receiving end obtains the target signal based on baseband demodulation, channel decoding and the target receiving signal.

Description

Phase noise compensation method and system of wireless communication system based on integrated clustering

Technical Field

The present disclosure relates to the field of wireless communication technologies, and in particular, to a phase noise compensation method and system for a wireless communication system based on integrated clustering.

Background

In modern wireless communication, theoretical analysis of wireless communication systems assumes perfect phase reference estimation, but in actual wireless communication, the phase reference estimation is often noisy (i.e., phase noise) due to imperfect phase-locked loop circuit or imperfect channel estimation, and the phase noise can greatly reduce demodulation performance of the system.

Existing methods for suppressing phase noise focus on improving the accuracy of phase reference estimation, however, the accuracy of phase reference estimation is often not high.

Disclosure of Invention

The present disclosure has been made in view of the above circumstances, and an object thereof is to provide a phase noise compensation method and system for a wireless communication system based on ensemble clustering, which can be easily integrated with an existing wireless communication system and can reduce the adverse effect of phase noise.

To this end, a first aspect of the present disclosure provides a phase noise compensation method for a wireless communication system based on integrated clustering, the phase noise compensation method for a wireless communication system having a transmitting end and a receiving end, the method comprising: the transmitting terminal transmits a carrier signal to a wireless channel based on channel coding, baseband modulation and radio frequency modulation, and the carrier signal obtains a received signal through the wireless channel; the receiving end receives the received signal, obtains a baseband signal from the received signal based on radio frequency demodulation and a phase-locked loop circuit, obtains a gain baseband signal based on the baseband signal and automatic gain control, obtains a plurality of standard constellation points, a plurality of clusters corresponding to a plurality of sample points corresponding to the gain baseband signal and a plurality of cluster center points corresponding to the clusters one by one based on a cluster model and the gain baseband signal, further obtains norm distances between any cluster center point and each standard constellation point based on a distance calculation model, further selects a target constellation point having the minimum norm distance from the cluster center points from the plurality of standard constellation points, and replaces the coordinates of the sample points corresponding to each cluster to the coordinates of the target constellation point corresponding to the cluster based on a cluster mapping model to realize phase noise compensation, and then obtaining a target receiving signal, and obtaining the target signal based on baseband demodulation, channel decoding and the target receiving signal, wherein the clustering model is a weighted integration clustering algorithm, clustering results corresponding to each clustering algorithm are obtained based on a plurality of different clustering algorithms and the gain baseband signal, a copolymerization indication matrix corresponding to each clustering result is obtained based on each clustering result, a set matrix is obtained based on the clustering results and the copolymerization indication matrix, and then the weighted integration clustering algorithm is obtained.

In the disclosure, an emitting end emits a carrier signal to a wireless channel based on channel coding, baseband modulation and radio frequency modulation, the carrier signal obtains a received signal through the wireless channel, a receiving end receives the received signal and obtains a baseband signal therefrom, a gain baseband signal is obtained based on the baseband signal and automatic gain control, then a plurality of clusters corresponding to a plurality of sample points corresponding to the gain baseband signal and a plurality of cluster center points corresponding to the clusters one to one are obtained based on a cluster model, a norm distance between any one cluster center point and each standard constellation point is obtained based on a distance calculation model, a standard constellation point corresponding to a minimum norm distance of the cluster center point is marked as a target constellation point, and then coordinates of the sample point corresponding to each cluster are replaced to coordinates of the target constellation point corresponding to the cluster based on a cluster mapping model to realize phase noise compensation, and further obtaining a target receiving signal, wherein the receiving end obtains the target signal based on baseband demodulation, channel decoding and the target receiving signal. Thus, the negative effects of phase noise can be reduced, providing a higher accuracy phase reference estimate.

In the phase noise compensation method according to the first aspect of the present disclosure, optionally, a modulation order of the wireless communication system is known by the receiving end, and the number of the plurality of clusters is the same as the modulation order. Thereby, the number of clusters can be determined.

In the phase noise compensation method according to the first aspect of the present disclosure, optionally, the target received signal is obtained after coordinates corresponding to each cluster center point are all converted into coordinates of a corresponding target constellation point. Thereby, a target reception signal can be obtained.

In the phase noise compensation method according to the first aspect of the present disclosure, optionally, a norm distance between the ith cluster center point and the jth standard constellation point satisfies: d _ij ＝||C _i -S _j || ² 1, M, j, M, wherein C is C _i Is the ith cluster center point, S _j And M is a modulation order of the multi-system frequency shift keying system. Thus, the norm distance between the cluster center point and the standard constellation point can be obtained.

In the phase noise compensation method according to the first aspect of the present disclosure, optionally, the plurality of different clustering algorithms include a K-means clustering algorithm, a K-center clustering algorithm, and a coacervation hierarchical clustering algorithm. Thereby, a weighted integrated clustering algorithm can be obtained based on the plurality of clustering algorithms.

A second aspect of the present disclosure provides a phase noise compensation system of a wireless communication system having a transmitting apparatus and a receiving apparatus, the phase noise compensation system including: the transmitting device transmits a carrier signal to a wireless channel based on channel coding, baseband modulation and radio frequency modulator modulation, wherein the carrier signal obtains a receiving signal through the wireless channel; the receiving device receives the received signal, obtains a baseband signal from the received signal based on radio frequency demodulation and a phase-locked loop circuit, obtains a gain baseband signal based on the baseband signal and automatic gain control, obtains a plurality of standard constellation points, a plurality of clusters corresponding to a plurality of sample points corresponding to the gain baseband signal and a plurality of cluster center points corresponding to the clusters one by one based on a cluster model and the gain baseband signal, further obtains norm distances between any cluster center point and each standard constellation point based on a distance calculation model, further selects a target constellation point having a minimum norm distance from the cluster center points from the plurality of standard constellation points, and replaces the coordinates of the sample point corresponding to each cluster to the coordinates of the target constellation point corresponding to the cluster based on a cluster mapping model to realize phase noise compensation, and then obtaining a target receiving signal, and obtaining the target signal based on baseband demodulation, channel decoding and the target receiving signal, wherein the clustering model is a weighted integration clustering algorithm, clustering results corresponding to each clustering algorithm are obtained based on a plurality of different clustering algorithms and the gain baseband signal, a co-clustering indication matrix corresponding to each clustering result is obtained based on each clustering result, and a set matrix is obtained based on the clustering results and the co-clustering indication matrix, so that the weighted integration clustering algorithm is obtained.

In the disclosure, a transmitting device transmits a carrier signal to a wireless channel based on channel coding, baseband modulation and radio frequency modulation, the carrier signal obtains a received signal through the wireless channel, a receiving device receives the received signal and obtains a baseband signal therefrom, obtains a gain baseband signal based on the baseband signal and automatic gain control, further obtains a plurality of clusters corresponding to each of standard constellation points and a plurality of sample points corresponding to the gain baseband signal based on a clustering model, and a plurality of clustering center points corresponding to the clusters one to one, obtains a norm distance between any one of the clustering center points and each of the standard constellation points based on a distance calculation model, and marks a standard constellation point corresponding to a minimum norm distance of the clustering center point as a target constellation point, further replaces coordinates of the sample points corresponding to each of the clusters to coordinates of the target constellation point corresponding to the clustering based on the clustering mapping model to implement phase noise compensation, and further, a target received signal is obtained, and the receiving device obtains the target signal based on the baseband demodulation, the channel decoding and the target received signal.

In the phase noise compensation system according to the second aspect of the present disclosure, optionally, a modulation order of the wireless communication system is known by the receiving apparatus, and the number of the plurality of clusters is the same as the modulation order. Thereby, the number of clusters can be determined.

In the phase noise compensation system according to the second aspect of the present disclosure, optionally, the target received signal is obtained by converting coordinates corresponding to each cluster center point into coordinates of a corresponding target constellation point. Thereby, a target reception signal can be obtained.

In the phase noise compensation system according to the second aspect of the present disclosure, optionally, a norm distance between the ith cluster center point and the jth standard constellation point satisfies: d _ij ＝||C _i -S _j || ² I 1, a, M, j 1, a, M, wherein C _i Is the ith cluster center point, S _j And M is a modulation order of the multi-system frequency shift keying system. Thus, the norm distance between the cluster center point and the standard constellation point can be obtained.

In the phase noise compensation system according to the second aspect of the present disclosure, optionally, the plurality of different clustering algorithms include a K-means clustering algorithm, a K-center clustering algorithm, and a coacervation hierarchical clustering algorithm. Thereby, a weighted integrated clustering algorithm can be obtained based on the plurality of clustering algorithms.

According to the present disclosure, it is possible to provide a phase noise compensation method and system for a wireless communication system based on integrated clustering, which is easily integrated with an existing wireless communication system and can reduce the negative effects of phase noise.

Drawings

Fig. 1 is a block diagram illustrating a classical wireless communication system to which examples of the present disclosure relate.

Fig. 2 is a block diagram illustrating a method of phase noise compensation for an integrated cluster-based wireless communication system in accordance with an example of the present disclosure.

Fig. 3 is a flow diagram illustrating a method of phase noise compensation for an integrated cluster based wireless communication system in accordance with an example of the present disclosure.

Fig. 4 is a diagram illustrating a constellation diagram in a multilevel keying system to which examples of the present disclosure relate.

Fig. 5 is a constellation diagram illustrating determining a target constellation point to which examples of the present disclosure relate.

Fig. 6 shows a constellation diagram of a quadrature phase shift keying modulation system under fixed phase noise to which examples of the present disclosure relate.

Fig. 7 shows a constellation diagram for a quadrature amplitude modulation system under fixed phase noise to which examples of the present disclosure relate.

Fig. 8 is a waveform diagram illustrating average normalized mutual information and correct rate as a function of received signal-to-noise ratio under fixed phase noise according to examples of the present disclosure.

Fig. 9 is a waveform diagram illustrating average bit error rate with received signal-to-noise ratio under fixed phase noise in accordance with an example of the present disclosure.

Fig. 10 is a waveform diagram illustrating average normalized mutual information and correct rate as a function of phase shift under fixed phase noise according to an example of the present disclosure.

Fig. 11 is a waveform diagram illustrating average bit error rate with phase shift under fixed phase noise according to an example of the present disclosure.

Fig. 12 is a waveform diagram illustrating average normalized mutual information with received signal-to-noise ratio under random phase noise according to an example of the present disclosure.

Fig. 13 is a waveform diagram illustrating average bit error rate with received signal-to-noise ratio under random phase noise according to an example of the present disclosure.

Fig. 14 is a block diagram illustrating a phase noise compensation system for an integrated cluster-based wireless communication system in accordance with an example of the present disclosure.

Detailed Description

Hereinafter, preferred embodiments of the present disclosure will be described in detail with reference to the accompanying drawings. In the following description, the same components are denoted by the same reference numerals, and redundant description thereof is omitted. The drawings are schematic and the ratio of the dimensions of the components and the shapes of the components may be different from the actual ones.

The present disclosure provides a phase noise compensation method and system (also referred to as "phase noise compensation method and phase noise compensation system") for a wireless communication system based on integrated clustering. In the present disclosure, a phase noise compensation method and system for a wireless communication system based on integrated clustering (i.e., "integrated clustering algorithm") can be widely applied to an existing wireless communication system, can be more easily integrated with the existing wireless communication system, can significantly reduce the influence of phase noise on phase reference estimation, and can improve the demodulation performance of the wireless communication system, thereby improving the communication quality. The present disclosure is described in detail below with reference to the attached drawings.

Fig. 1 is a block diagram illustrating a classical wireless communication system to which an example of the present disclosure relates. Fig. 2 is a block diagram illustrating a method of phase noise compensation for an integrated cluster-based wireless communication system in accordance with an example of the present disclosure. As shown in fig. 1 and 2, the phase noise compensation method of the present disclosure may be applied to a classical wireless communication system, but examples of the present disclosure are not limited thereto and may also be applied to other wireless communication systems. The phase noise compensation method and system of the present disclosure can be more easily integrated with existing wireless communication systems.

In some examples, the phase noise compensation methods of the present disclosure may operate only on baseband circuitry, thereby enabling cost and complexity reduction. In some examples, the phase noise compensation method of the present disclosure may add a preprocessing process (described later) before the baseband demodulation 370, and may not change other parts, thereby enabling the phase noise compensation method of the present disclosure to be more easily integrated with an existing wireless communication system.

In the present disclosure, as shown in fig. 1 and 2, the phase noise compensation method has a phase noise compensation method of a wireless communication system of a transmitting end 10 and a receiving end 30, wherein the transmitting end 10 can transmit a signal to the receiving end 30 and be received by the receiving end 30.

In the present disclosure, a transmitting end 10 (e.g., an access point) may refer to a device in an access network that communicates over the air-interface, through one or more sectors, with wireless terminals. The transmitting end 10 may be configured to interconvert received air frames and IP frames as a router between the wireless terminal and the rest of the access network, which may include an Internet Protocol (IP) network. The transmitting end 10 may also coordinate the management of attributes for the air interface. For example, the transmitting end 10 may be a Base Transceiver Station (BTS) in GSM or CDMA, a Base Station (NodeB) in WCDMA, and an evolved Node B (NodeB or eNB or e-NodeB) in LTE.

In the present disclosure, the receiving end 30 may be a user. The user may include, but is not limited to, a user device. The user Device may include, but is not limited to, various electronic devices such as a smart Phone, a notebook Computer, a Personal Computer (PC), a Personal Digital Assistant (PDA), a Mobile Internet Device (MID), a wearable Device (e.g., a smart watch, a smart bracelet, and smart glasses), wherein an operating system of the user Device may include, but is not limited to, an Android operating system, an IOS operating system, a Symbian operating system, a blackberry operating system, a Windows Phone8 operating system, and so on.

Fig. 3 is a flow diagram illustrating a method of phase noise compensation for an integrated cluster-based wireless communication system in accordance with an example of the present disclosure.

In the present embodiment, as shown in fig. 3, the phase noise compensation method may include the steps of: the transmitting terminal 10 transmits a carrier signal to the wireless channel 20 based on the channel coding 100, the baseband modulation 110 and the radio frequency modulation 120, and the carrier signal passes through the wireless channel 20 to obtain a received signal (step S10); the receiving end 30 receives the received signal, obtains a baseband signal from the received signal based on the rf demodulation 310 and the pll circuit 320, and obtains a gain baseband signal based on the baseband signal and the agc 330 (step S20); the receiving end 30 obtains, based on the cluster model 340 and the gain baseband signal, a plurality of clusters corresponding to each standard constellation point, a plurality of sample points (also referred to as "sample constellation points") corresponding to the gain baseband signal, and a plurality of cluster center points corresponding to each cluster one to one (step S30); obtaining a norm distance between any cluster center point and each standard constellation point based on the distance calculation model 350, and further selecting a target constellation point having a minimum norm distance from the cluster center point from the plurality of standard constellation points (step S40); the coordinates of the sample point corresponding to each cluster are replaced with the coordinates of the target constellation point corresponding to the cluster based on the cluster mapping model 360 to implement phase noise compensation, so as to obtain a target received signal, and the target signal is obtained based on the baseband demodulation 370, the channel decoding 380, and the target received signal (step S50).

In the present disclosure, the transmitting terminal 10 may transmit a carrier signal to the wireless channel 20 based on the channel coding 100, the baseband modulation 110 and the radio frequency modulation 120, the carrier signal may obtain a received signal through the wireless channel 20, the receiving terminal 30 may receive the received signal and obtain a baseband signal therefrom, obtain a gain baseband signal based on the baseband signal and the automatic gain control, further obtain a plurality of clusters corresponding to a plurality of sample points in a constellation corresponding to the gain baseband signal and a plurality of cluster center points corresponding to the clusters based on the cluster model 340, obtain a norm distance between any cluster center point and each standard constellation point based on the distance calculation model 350, mark a standard constellation point corresponding to a minimum norm distance of the cluster center point as a target constellation point, and further replace coordinates of the sample point corresponding to each cluster to coordinates of the target constellation point corresponding to the cluster based on the cluster mapping model 360 to implement phase mapping between the standard constellation points and the clusters The noise compensation is performed to obtain the target received signal, and the receiving end 30 may obtain the target signal based on the baseband demodulation 370, the channel decoding 380 and the target received signal.

In step S10, the transmitting end 10 transmits a carrier signal to the wireless channel 20 based on the channel coding 100, the baseband modulation 110 and the radio frequency modulation 120, and the carrier signal passes through the wireless channel 20 to obtain a received signal.

In some examples, as shown in fig. 1, a data source may first undergo channel coding 100 and baseband modulation 110 to obtain a baseband modulation signal in a transmitting end 10, where a modulation order of the baseband modulation 110 is M, pilot symbols may then be periodically embedded in the baseband modulation symbols, a carrier signal may then be obtained through radio frequency modulation 120, and the transmitting end 10 may transmit the carrier signal onto a wireless channel 20, where a transmission power of the carrier signal may be represented as P _s 。

In some examples, the carrier signal is received by receiving end 30 via wireless channel 20 to obtain a received signal. In some examples, the received signal received by the receiving end 30 may satisfy:

where h (t) is expressed as the channel response (i.e., the actual channel fading estimate) of the fading amplitude η (t) and the fading phase θ (t) and satisfies: h (t) ═ η (t) exp (j θ (t)), s (t) is expressed as a baseband modulation signal, ω (t) — ω ₀ Expressed as carrier frequency, [ phi ] (t) expressed as received carrier random phase, [ n ] (t) expressed as received noise and as complex white gaussian noise and satisfying:

wherein the content of the first and second substances,

is the variance.

In some examples, wireless channel 20 may be a flat fading channel, where each frame of data may experience an independent channel fade, and the channel fade may remain constant for its duration, but may change over different frames of data. Wherein, the frame length can be L, the fading amplitude and the fading phase can be respectively [ - π, π]Modeled as Nakagami-m distribution and uniform distribution, the probability density function of Nakagami-m distribution can satisfy:

eta is greater than or equal to 0, wherein m epsilon [1/2, ∞), and Gamma (·) is a Gamma function. In some examples, the instantaneous received signal-to-noise ratio may satisfy:

the average received signal-to-noise ratio (also called "received signal-to-noise ratio") is expressed as

Where s is expressed as the baseband modulation signal and η is expressed as the fading amplitude.

In step S20, the receiving end 30 receives the received signal, obtains a baseband signal from the received signal based on the rf demodulation 310 and the pll circuit 320, and obtains a gain baseband signal based on the baseband signal and the agc 330.

In some examples, as shown in fig. 1, the receiving end 30 may receive a received signal. The received signal in the receiving end 30 can be processed by rf demodulation 310 and phase-locked loop circuit 320 to obtain a baseband signal. In some examples, the phase-locked loop circuit 320 may be undesirable, and thus may result in a first phase error that may satisfy:

wherein the content of the first and second substances,

denoted as the first phase obtained by the phase-locked loop circuit 320 and phi (t) as the actual first phase. In some examples, the first phase error may be modeled as Tikhonov, whereby a probability density function of the first phase error (first phase noise) may be obtained satisfying:

where α is expressed as the normalized circulating signal-to-noise ratio, I, of the phase-locked loop circuit 320 ₀ Expressed as the zero order correctionThe Seal function. In some examples, the partial transmission power reserved for pilot symbols may satisfy: p _c ＝χP _s Wherein χ is P _s Is fixed coefficient of (a). The approximation of α satisfies:

wherein, B _L Expressed as loop bandwidth, T _b Represented as a bit interval.

In some examples, rf demodulation 310 may utilize the suppressed intersymbol interference of rf modulation 120 of transmitting end 10, and through the pilot symbols and the pilot observations, may enable receiving end 30 to obtain the channel fading estimation and satisfy:

wherein the content of the first and second substances,

denoted as the fading amplitude obtained by the receiving end 30 based on the channel estimation. In some examples, there is a second phase error that satisfies:

wherein the content of the first and second substances,

denoted as the fading phase obtained by the receiving end 30 based on the channel estimation, and θ (t) is denoted as the actual fading phase. In some examples, the probability density function of the second phase error (second phase noise) may satisfy:

wherein ρ is a correlation coefficient and satisfies:

in some examples, ρ may be set to a constant.

In some examples, as shown in fig. 1, the baseband signal may be subjected to an automatic gain control 330 to obtain a gain baseband signal.Wherein, the gain baseband signal can satisfy:

in some examples, the baseband signal is subjected to automatic gain control 330 to obtain a gain baseband signal, which may be divided by the channel fading estimate

The quotient of (1) compensates channel fading to obtain a gain baseband signal after channel fading compensation, and the following conditions are met:

wherein the content of the first and second substances,

expressed as total residual phase noise (i.e., total phase error) and satisfies:

denoted as residual received noise, which may be obtained after the received noise is affected by the channel estimation and satisfies:

in some examples, as shown in fig. 1, in a classical wireless communication system, a gain baseband signal may obtain information (i.e., a data source) transmitted by the transmitting end 10 through baseband demodulation 370, but due to the presence of phase noise, the phase noise may cause constellation points to deviate from an original position, which may greatly reduce demodulation performance of the receiving end 30, and thus the receiving end 30 may not obtain accurate information.

In some examples, the abscissa I of fig. 4 and 5 is the phase amplitude and the ordinate Q is the quadrature amplitude.

Fig. 4 is a diagram illustrating a constellation diagram in a multilevel keying system to which an example of the present disclosure relates. Fig. 4(a) is a constellation diagram in the absence of phase noise, and fig. 4(b) is a constellation diagram in the presence of phase noise.

In some examples, as in FIG. 4 (C:)a) And fig. 4(b), in a multilevel frequency shift keying system, phase noise may cause constellation points to deviate from the original positions. In fig. 4(a), the positions of the constellation points are not affected by the phase noise, where the distances from the center point corresponding to any constellation point region to the two nearest decision region edges are equal, i.e. d ₁ ＝d ₂ . In fig. 4(b), the position of the constellation point is shifted by the phase noise, and the distances from the shifted constellation point region to the two nearest decision region edges are different, i.e. d ₃ ≠d ₄ . In some examples, the demodulation error probability may be determined by the smaller of the distances from the constellation point to the edges of the two decision regions, i.e., in fig. 4(b), the demodulation error probability is determined by d ₄ It is decided that in this case, the phase noise may degrade the demodulation performance.

The present disclosure provides a phase noise compensation method of a wireless communication system capable of reducing an influence of phase noise on phase reference estimation, which may preprocess a gain baseband signal to obtain a target reception signal.

In step S30, the receiving end 30 may obtain, based on the cluster model 340, each standard constellation point and a plurality of clusters corresponding to a plurality of sample points in a constellation diagram corresponding to the gain baseband signal (here, the gain baseband signal may be a "gain baseband signal after compensating for channel fading") and a plurality of cluster center points corresponding to each cluster one by one. The clustering model 340 may be a weighted-ensemble clustering algorithm (also referred to as an "ensemble clustering algorithm"), and obtains clustering results corresponding to each clustering algorithm based on a plurality of different clustering algorithms and gain baseband signals, obtains a co-cluster indication matrix corresponding to each clustering result based on each clustering result, obtains a set matrix based on the clustering results and the co-cluster indication matrix, and further obtains a weighted-ensemble clustering algorithm (described in detail later).

In some examples, as shown in fig. 5, the gain baseband signal may be subject to a clustering model 340 to obtain a plurality of constellations corresponding to the gain baseband signalThe sample points, that is, the cluster model 340 may receive a plurality of sample points (e.g., the sample point 401, the sample point 402, the sample point 403, etc.) from the gain baseband signal, the plurality of sample points may correspond to the constellation points in the constellation diagram, an integrated cluster result (described later in detail) may be obtained based on a weighted integrated clustering algorithm, and the integrated cluster result includes a plurality of clusters corresponding to the plurality of sample points (e.g., the sample point 401, the sample point 402, and the sample point 403 may correspond to the cluster 400) (cluster analogy may be the above-mentioned constellation point region) and cluster center points corresponding to the respective clusters, e.g., the cluster 400 and the corresponding cluster center point C ₁ Center point C corresponding to cluster 410 ₂ Center point C corresponding to cluster 420 ₃ Center point C corresponding to cluster 430 ₄ . In some examples, the receiving end 30 may obtain each standard constellation point in the constellation diagram corresponding to the gain baseband signal, for example, the standard constellation point S ₁ Standard star point S ₂ Standard star point S ₃ And standard constellation point S ₄ . The number of standard constellation points may be the same as the number of clusters.

In some examples, the modulation order M may be known by the receiving end 30, and the number of clusters may be the same as the modulation order M. Thereby, the number of clusters can be determined. In some examples, the clustering model 340 may be a weighted ensemble clustering algorithm (described later) that may be determined based on a plurality of different clustering algorithms and gain baseband signals. In some examples, the plurality of different clustering algorithms may include different types of clustering algorithms, such as a K-means clustering algorithm, a K-center clustering algorithm, and a coacervate hierarchy clustering algorithm. Thereby, a weighted ensemble clustering algorithm can be obtained based on the plurality of clustering algorithms.

In step S40, the receiving end 30 may obtain a norm distance between any one of the cluster center points and each of the standard constellation points based on the distance calculation model 350, and further select a target constellation point having the smallest norm distance from the cluster center point from the plurality of standard constellation points.

In some examples, the cluster center point and the standard constellation point obtained by the receiving end 30 based on the distance calculation model 350 and the clustering model 340 may beAnd obtaining the norm distance between any one cluster central point and each standard constellation point. In some examples, a norm distance between any one cluster center point and each standard constellation point may satisfy: d _ij ＝||C _i -S _j || ² 1, M, j, M (5), wherein C is a radical of formula i _i Is the ith central point, S _j Is the jth standard constellation point, and M is the modulation order. Therefore, the norm distance between the clustering center point and the standard constellation point can be obtained. In some examples, the receiving end 30 is based on

The standard constellation point with the minimum norm distance from the ith cluster center point can be obtained, and the standard constellation point is marked as the target constellation point of the ith cluster center point. Therefore, the target constellation point corresponding to each cluster center point can be determined, namely the target constellation point corresponding to each cluster is determined. For example, as shown in FIG. 5, cluster center point C for cluster 400 ₁ (i.e., the 1 st cluster center point), the cluster center point C can be obtained based on the formula (5) ₁ With each standard constellation point (e.g. standard constellation point S) ₁ Standard star point S ₂ Standard star point S ₃ And standard constellation point S ₄ ) The norm distance between the two and the cluster center point C can be obtained by the formula (6) ₁ Standard constellation point (e.g. standard constellation point S) with the smallest norm distance ₂ ) Therefore, the standard star point S can be obtained ₂ Marking as a Cluster center Point C ₁ Is the target constellation point, i.e. the standard constellation point S ₂ May be the target constellation point corresponding to the cluster 400, thereby obtaining the target constellation point corresponding to each cluster, e.g. the standard constellation point S ₃ May be the target constellation point, standard constellation point S, corresponding to cluster 410 ₄ May be the target constellation point, standard constellation point S, corresponding to cluster 420 ₁ May be the target constellation point to which cluster 430 corresponds.

In step S50, the receiving end 30 may replace the coordinates of the sample point corresponding to each cluster with the coordinates of the target constellation point corresponding to the cluster based on the cluster mapping model 360 to implement phase noise compensation, thereby obtaining a target received signal, and obtain the target signal based on the baseband demodulation 370, the channel decoding 380, and the target received signal.

In some examples, the receiving end 30 may replace the coordinates of the sample point corresponding to each cluster to the coordinates of the target constellation point corresponding to the cluster based on the target constellation points corresponding to each cluster obtained by the cluster mapping model 360 and the distance calculation model 350, that is, each cluster region is moved such that the coordinates of the cluster center point corresponding to each cluster region are moved to the coordinates of the target constellation point corresponding to the cluster center point. For example, as shown in FIG. 5, sample point 401, sample point 402, and sample point 403 in cluster 400 will all move to standard constellation point S ₂ Corresponding movement will occur for sample points in other clusters in fig. 5 (e.g., cluster 410, cluster 420, cluster 430). In this case, the coordinates of the sample point corresponding to each cluster may be replaced with the coordinates of the target constellation point corresponding to the cluster to implement phase noise compensation, so that the target constellation can be obtained, and the target received signal can be obtained.

In some examples, the gain baseband signal may be pre-processed (i.e., the gain baseband signal passes through the clustering model 340, the distance calculation model 350, and the clustering mapping model 360 in sequence) to obtain the target received signal. In some examples, the target received signal may be subjected to baseband demodulation 370 and channel decoding 380 to obtain the target signal, so that the receiving end 30 can more accurately obtain the information sent by the transmitting end 10.

In some examples, the clustering model 340 may be a weighted-ensemble clustering algorithm, and may obtain clustering results corresponding to each clustering algorithm based on a plurality of different clustering algorithms and gain baseband signals, obtain a co-cluster indication matrix corresponding to each clustering result based on each clustering result, obtain a set matrix based on the clustering results and the co-cluster indication matrix, and further obtain a weighted-ensemble clustering algorithm.

In some examples, obtaining the real and imaginary parts of the plurality of sample points from the gain baseband signal, equation (4), may be used as a two-dimensional input signal for the selected clustering algorithm. In some examples, two types of clustering algorithms based on partitions and based on hierarchies can be selected for cluster division, for example, the clustering algorithm based on partitions includes a K-means clustering algorithm, a K-center clustering algorithm, and the like, all clusters can be quickly determined, all sample points can be cluster-divided, and each cluster can contain at least one sample point; the hierarchy-based clustering algorithm may include a agglomerative hierarchy clustering algorithm, and the like. In this embodiment, a weighted ensemble clustering algorithm may be employed, thereby enabling higher quality clustering results to be obtained. In some examples, the clustering results corresponding to the respective clustering algorithms may be obtained based on a plurality of different clustering algorithms and gain baseband signals, for example, a K-means clustering algorithm, a K-means (absolute value distance) clustering algorithm, a K-means (included angle cosine) clustering algorithm, a K-center point (absolute value distance) clustering algorithm, a K-center point (included angle cosine) clustering algorithm, a coacervation hierarchical clustering algorithm (average), and a coacervation hierarchical clustering algorithm (weighted) may be adopted to perform clustering division on a plurality of sample points obtained from the gain baseband signals respectively to obtain 9 clustering results (also referred to as "base clustering results"), and the 9 base clustering results may be integrated to obtain an integrated clustering result.

In some examples, clustering the plurality of sample points using different clustering algorithms may obtain a plurality of basis clustering results. In some examples, the performance may be different between multiple base clustering results in a given wireless channel 20, and thus a weighted integration method may be employed to assign a weight to each base clustering result. Thereby, an integrated clustering result can be obtained.

In some examples, obtaining an integrated clustering algorithm may include two phases: a generation phase and a constellation detection phase. In some examples, the generating stage includes: based on each clustering result, a corresponding co-clustering indication matrix of each clustering result can be obtained, that is, a plurality of obtained base clustering results can be respectively converted into respective corresponding co-clustering indication matrices. For example, N can be obtained _b The individual base clusters the result, and can be N _b The individual base clustering results are respectively converted into the corresponding copolymerization classesIndication matrix D _i (i＝1,2,...,N _b ). Wherein, in the t-th base clustering result, if the ith and jth sample points belong to the same cluster, D _i (i, j) ═ 1; otherwise D _i (i,j)＝0。

In some examples, a set matrix may be obtained based on the clustering results and the co-cluster indication matrix, which may in turn obtain a weighted ensemble clustering algorithm. In some examples, there may be multiple (i.e., N) according to a weighted combination rule _b One) the basis clustering results can be summarized as the set matrix E satisfying:

wherein, ω is _i Is

D

_i 1,2, N _b S.t. is expressed as satisfying the constraint. W is a vector of weights satisfying:

in some examples, to suppress overfitting of the parameter W to one of the base cluster results, a regularization term may be introduced

Which can be expressed as the sum of the negative entropies of the weights of the individual base clustering results.

In some examples, the constellation detection stage includes: it can be assumed that each item E in the set matrix E _i,j The evidence provided by the base clustering result is represented, that is, the ith and jth sample points belong to the same constellation point (e.g., a standard constellation point), that is, the ith and jth sample points belong to the same cluster. In some examples, assuming that the demodulation system of the present disclosure includes M constellation points, for the ith sample point, a preset parameter β may be introduced _i,k For indicating the intensity of the ith sample point from the kth constellation point, i.e. beta _i,k A higher value of (a) indicates a higher probability that the ith sample point is from the kth constellation point, where k is 1, 2. In some examples, the number of clusters may be equal to the modulation order, i.e. may satisfy: and K is M.

In some examples, B may be represented asThe symbol-constellation point tendency matrix satisfies:

wherein the content of the first and second substances,

expressed as a generic set representation symbol, M is expressed as the number of constellation points, N _L Expressed as the number of received sample points,

the value of (d) may be expressed as the likelihood that the ith and jth sample points belong to the same constellation point. In some examples, E _i,j The value of (d) may indicate the likelihood that the ith and jth sample points are located in the same cluster. The bi-directional correlation in the set matrix E may be affected by an unobserved non-negative parameter Q, which may satisfy: q ═ BB ^T Wherein each element is beta _i,k β _j,k May represent the kth constellation point pair Q _i,j Wherein each term Q of the non-negative parameter Q _i,j Can satisfy the following conditions:

Q _i,j can be expressed as (BB) ^T ) _i,j . In some examples, assume a single element E _i,j (i.e., each entry E in the set matrix E _i,j ) Can be determined by

Giving a solution to the problem that, among others,

and may be a poisson probability density function, where Γ (·) is a Gamma function. As described above, the integrated demodulation system probability can be expressed as

In some examples, the value of B may be estimated by maximizing the conditional probability defined in equation (9). Can be obtained by taking the negative logarithm and the decreasing constant of equation (9)The objective function of the clustering algorithm is integrated in a weighted mode, and since Γ (1) ═ Γ (2) ═ 1, E _i,j Values between 0 and 1, Γ (E) may be assumed _i,j +1) ═ 1, the objective function can therefore satisfy:

in some examples, the regularization term R defined in equation (8) may be added and E replaced with equation (7), e.g., E _i,j Can be expressed as ω _t (D _t ) _i,j Wherein, t is 1,2 _p ，ω _t Can be equivalent to omega _i ，D _t Can be equivalent to D _i And subsequently D _m Thereby, the objective function, i.e., equation (10), can be converted into

Wherein, the first and the second end of the pipe are connected with each other,

D _t expressed as a co-clustering indication matrix corresponding to the t-th base clustering result, lambda is a trade-off parameter which satisfies: λ ≧ 0, which can control the balance between the objective function defined in equation (10) and the regularization term R.

In some examples, given W, equation (11) may be degenerated as:

s.t. denotes s.t.beta _i,k ≥0 (12)

Which is satisfied with the above. J (B) may be minimized according to B. In some examples, to solve the constraint optimization problem in equation (12), a multiplication update rule may be employed, assuming φ _i,k Is a Lagrangian multiplier, which can be constrained to β _i,k Not less than 0 and phi ═ phi [ phi _i,k ]Then the lagrangian function can satisfy:

in some examples, may be according to β _i,k Obtaining a Lagrangian function

Satisfies the following gradient:

β _i,k the estimated value of (c) may satisfy:

thereby can obtain

In some examples, according to an order optimization condition, φ _i,k β _i,k 0, one can obtain:

whereby beta can be obtained _i,k The update rule of (2):

in some examples, referring to the actual case, it can be obtained based on equation (17):

in some examples, a multiplicative update rule may be used, if used to initialize β for non-negative values _i,k Then β _i,k The value of (c) may remain non-negative. After updating B and fixing, equation (11) can be degenerated as:

can use

The lagrange multiplier solves the unconstrained minimization problem of:

wherein χ is restricted to

Lagrange multiplier. In some examples, to obtain

Can assume that the gradients of all variables disappear, so that:

and

it can be obtained from formula (21):

in some examples, formula (23) may be substituted for formula (22) to obtain ω _m Satisfies the following update rule:

wherein, ω is _m ω equivalent to ω in the foregoing _t 。

In some examples, a weighted ensemble clustering algorithm may be obtained based on the plurality of basis clustering results, equation (11), equation (18), and equation (24). In some examples, multiple basis clustering results may be obtained based on different clustering algorithms, and a trade-off parameter λ is predefined, and if λ ∞ is set, the weights of all element systems may be forced to be equal, and if λ ∞ is set, the regularization term is discarded.

In some examples, λ may be determined by equation (24) to depend on

The value of (c). The trade-off parameter λ may satisfy:

in some examples, to determine a suitable λ value, one may vary λ ₀ To evaluate the corresponding performance. If λ ₀ Is sufficiently large, the weights assigned to the respective basis clustering results can be made approximately equal, unless some of the basis clustering results are poor and the corresponding performance will not follow λ ₀ Increased and changed significantly. In some examples, the weighted ensemble clustering algorithm includes: obtaining a set matrix E based on a plurality of base clustering results and based on weighing parameters lambda and beta _i,k Update rule formula (18) and ω _m Iteratively updating B and W until B and W meet the criterion (24). And the value of the objective function can be obtained based on equation (11). For example, stop when the number of iterations reaches 150. In some examples, to suppress the occurrence of the local minimum, the above algorithm may be repeated a plurality of times with a random initial condition, for example, the algorithm may be repeated 10 times with a random initial condition, and B and W corresponding to the minimum value of the objective function are selected therefrom. Thereby, an integrated clustering result obtained by integrating a plurality of base clustering results can be obtained.

Fig. 6 illustrates a constellation diagram for a quadrature phase shift keying modulation system under fixed phase noise to which examples of the present disclosure relate. Fig. 7 shows a constellation diagram for a quadrature amplitude modulation system under fixed phase noise to which examples of the present disclosure relate. As shown in fig. 6 and 7, fig. 6(a) and 7(a) are constellation diagrams of standard constellation points in respective corresponding systems, that is, constellation diagrams without phase noise, fig. 6(b) and 7(b) are a plurality of sample points received without using the present embodiment in respective corresponding systems, fig. 6(c) and 7(c) are constellation diagrams of cluster center points in respective corresponding systems corresponding to using a K-means clustering algorithm, and fig. 6(d) and 7(d) are constellation diagrams of cluster center points in respective corresponding systems corresponding to using a hierarchical clustering algorithm. In fig. 6 and 7, L is 100,

m＝1.5。

in some examples, as shown in fig. 6 and 7, by comparing fig. 6(b) and 6(a) or 7(b) and figure7(a) it can be found that phase noise is caused

There is such that all constellation points are rotated by 0.1 pi, and a plurality of sample points are rotated by the same angle along the corresponding constellation point. As shown in fig. 6(c) and 6(d), and fig. 7(c) and 7(d), since the distance between two constellation points is large enough, both the K-means clustering algorithm and the agglomerative hierarchical clustering algorithm can find the correct constellation center corresponding to the rotated constellation point, that is, each clustering center point can obtain the correct target constellation point, so that the receiving end 30 can obtain more accurate information, which is a precondition that the clustering mapping model 360 works well. With the increase of the modulation order, the distance between two constellation points becomes smaller, so that different clustering algorithms generate different clustering centers, which may result in poor effect of some clustering algorithms. For example, as shown in fig. 6(c), the K-means clustering algorithm generates some clearly erroneous cluster center points. However, as shown in FIG. 6(d), the agglomerative hierarchical clustering algorithm works well. Therefore, a method for adaptively selecting a proper clustering algorithm under different conditions (namely, a weighted integration clustering algorithm) can be found.

In some examples, the plurality of sample points are cluster partitioned based on the cluster model 340 and the gain baseband signal, and since the number of sample points corresponding to each cluster is unknown, different clusters may receive different numbers of sample points. Wherein the average normalized mutual information may detect the similarity between the cluster sample set (i.e. the divided clusters) and the reference constellation point set, C, as described above _i Is expressed as the cluster center point corresponding to the ith cluster, let C _i In particular the corresponding ith cluster, order C _i Denoted as the ith cluster sample set, G _j Represented as the sample set (i.e., the set of reference constellation points) of the jth ground truth cluster. Then C is _i Relative to G _j The average normalized mutual information of (a) may satisfy:

where K denotes the number of clusters, N denotes the number of multiple sample points, N _C,i Expressed as the number of sample points corresponding to the ith cluster sample set, N _G,j Expressed as the number of sample points corresponding to the jth set of reference constellation points, N _ij Expressed as the number of received sample points shared by the ith cluster sample set and the jth reference constellation point set.

In some examples, the value of the average normalized mutual information may be equal to 1 in an ideal case, i.e., all sample points are correctly identified, and the value of the average normalized mutual information may be reduced in an undesirable case, such as a low received signal-to-noise ratio. In some examples, since the average normalized mutual information only considers the clustering performance, the average normalized mutual information may be used as an intermediate indicator in the present embodiment, and the demodulation error probability may be used as a final performance indicator of the detection clustering algorithm.

In some examples, as shown in fig. 8 to 13, the average normalized mutual information or the demodulation error probability is analyzed under different system conditions, where a is a case of using a weighted ensemble clustering algorithm, B is a case of using a K-means clustering algorithm, C is a case of using a K-means (absolute value distance) clustering algorithm, D is a case of using a K-means (included angle cosine) clustering algorithm, E is a case of using a K-center-point clustering algorithm, F is a case of using a K-center-point (absolute value distance) clustering algorithm, G is a case of using a K-center-point (included angle cosine) clustering algorithm, H is a case of using a coacervate clustering algorithm, I is a case of using a coacervate-level clustering algorithm (average), and J is a case of using a coacervate-level clustering algorithm (weighting), in fig. 9, 11, and 13. K is the case for the normal scheme.

Fig. 8 is a waveform diagram illustrating average normalized mutual information and correct rate as a function of received signal-to-noise ratio under fixed phase noise according to examples of the present disclosure. Fig. 9 is a waveform diagram illustrating average bit error rate with received signal-to-noise ratio under fixed phase noise in accordance with an example of the present disclosure. As shown in fig. 8 and fig. 9, a quadrature phase shift keying modulation system is adopted, L is 100, m is 1.5,

in some examples, as shown in fig. 8, fig. 8(a) shows a waveform of the average normalized mutual information as a function of the received signal-to-noise ratio, and fig. 8(b) shows a waveform of the correct rate as a function of the received signal-to-noise ratio. As can be seen from fig. 8, as the received signal-to-noise ratio increases, the values of the average normalized mutual information and the accuracy rate are significantly improved, that is, the clustering performance is improved. Fig. 8 shows a waveform diagram of the variation of the average normalized mutual information with the received signal-to-noise ratio of 10 clustering algorithms, wherein the K-means (absolute value distance) clustering algorithm has the worst clustering performance for the high received signal-to-noise ratio region, and the coacervate hierarchical clustering algorithm (weighting) has the worst clustering performance for the low received signal-to-noise ratio region. As can be seen from fig. 8, although the clustering performance of other base clustering algorithms is similar, the weighted ensemble clustering algorithm is the best choice among them. Wherein, the base clustering algorithm is a clustering algorithm corresponding to a curve B to a curve J.

In some examples, as shown in fig. 9, fig. 9(a) shows a waveform of an average bit error rate of demodulation as a function of a received signal-to-noise ratio, and fig. 9(b) shows a waveform of an average bit error rate of demodulation and decoding as a function of a received signal-to-noise ratio. As can be seen from fig. 9, as the received snr increases, the average ber values in both cases significantly decrease, i.e., the corresponding final performance is improved (i.e., the demodulation and decoding performance of the receiving end 30 is improved). As can be seen from fig. 8 and 9, the difference between the average bit error rates of any two clustering algorithms is significantly greater than the difference between the average normalized mutual information, and it can be seen that the change of the system condition has a greater influence on the average bit error rate. As can be seen from fig. 9(a), the demodulation performance corresponding to the weighted-ensemble clustering algorithm is the best, and the demodulation performance of the K-means (absolute distance) clustering algorithm is poor. And at higher received signal-to-noise ratios the final performance using the K-means (absolute distance) clustering algorithm is worse than that of the normal scheme (i.e. normally processing the signal, e.g. using the classical wireless communication system disclosed in fig. 1), since some excessive mapping occurs in the K-means (absolute distance) clustering algorithm. As can be seen from FIG. 9(b), the average of demodulation and decoding when the weighted ensemble clustering algorithm is usedThe average error rate is respectively superior to the common scheme, for example, the average error rate with the common scheme is 10 ^-3 Compared with the weighted integrated clustering algorithm, the demodulation performance is improved by about 10dB, and the decoding performance is improved by about 8 dB.

Fig. 10 is a waveform diagram illustrating average normalized mutual information and correct rate as a function of phase shift under fixed phase noise according to an example of the present disclosure. Fig. 11 is a waveform diagram illustrating average bit error rate with phase shift under fixed phase noise according to an example of the present disclosure. As shown in fig. 10 and 11, quadrature phase shift keying modulation systems are adopted, L is 100, m is 1.5, and the received signal-to-noise ratio is 20dB, that is, the received signal-to-noise ratio is 20dB

In some examples, as shown in fig. 10, fig. 10(a) shows a waveform diagram of average normalized mutual information as a function of phase shift, and fig. 10(b) shows a waveform diagram of correct rate as a function of phase shift. As can be seen from fig. 10, the performance corresponding to the weighted ensemble clustering algorithm is the best, while the performance corresponding to the K-means (absolute value distance) clustering algorithm is the worst, so that the instability of the single clustering algorithm (e.g., using a base clustering algorithm) can be seen. As can be seen from fig. 10, in addition to the K-means (absolute value distance) clustering algorithm, the values of the average normalized mutual information and the accuracy corresponding to other clustering algorithms are almost independent of the phase shift, so that the robustness of the weighted ensemble clustering algorithm of the present disclosure can be highlighted.

In some examples, as shown in fig. 11, fig. 11(a) shows a waveform of an average bit error rate of demodulation as a function of phase shift, and fig. 11(b) shows a waveform of an average bit error rate of demodulation and decoding as a function of phase shift. Fig. 11 shows a waveform of the average bit error rate of 10 clustering algorithms as the received signal-to-noise ratio changes, and it can be seen from fig. 11 that as the phase noise increases, the average bit error rate of all clustering algorithms, including the average bit error rate of the common scheme, increases. For larger phase shifts, the average bit error rate values of the respective clustering algorithms and general schemes converge to a maximum value, e.g. when

The maximum average error rate value is 0.5.

In some examples, as can be seen in FIG. 11, for example, when

In time, some base clustering algorithms are not as good as the common scheme, and the weighted integration clustering algorithm can still keep better performance. For a large phase noise, all the base clustering algorithms and weighted ensemble clustering algorithms are inferior to the normal scheme because the rotated constellation points are closer to the edge of the decision region (as shown in fig. 4) and some excessive mapping occurs. However, large phase noise rarely occurs in practice because it is easily captured by the calibration circuit of the receiving terminal 30 and can be suppressed to a small phase noise. Comparing fig. 10(a) and (b), it can be seen that the average bit error rate of the weighted-integration clustering algorithm is better than that of the conventional scheme in demodulation and decoding respectively under the condition of smaller phase shift. Further, for some phase shifts, e.g.

The decoding performance of the weighted ensemble clustering algorithm is better than that of the common scheme.

In some examples, the influence of the frame length L, Nakagami channel parameter m on the average normalized mutual information, the correct rate and the average error rate is analyzed, as shown in table 1 and table 2, respectively. Wherein, a new metric is provided for the demodulation error probability to illustrate the relative improvement ratio provided by the weighted integration clustering algorithm of the embodiment, and the following is satisfied:

wherein the PDE _Normal Expressed as demodulation error probability, PDE, of the conventional scheme _Ensemble Expressed as the demodulation error probability of the weighted ensemble clustering algorithm. Wherein, both the tables 1 and 2 adopt the quadrature phase shift keying modulation system, the receiving signal to noise ratio is 10dB, and the phase noise is

TABLE 1

In some examples, as shown in table 1, table 1 shows the effect of the frame length L on the average normalized mutual information, the correct rate and the average bit error rate, where m is 1.5, and the relative improvement ratio corresponding to the demodulated average bit error rate and the decoded average bit error rate can be obtained by analogy with equation (25). As can be seen from table 1, different frame lengths L are not sensitive to the influence of the average normalized mutual information and the accuracy, but fluctuate slightly. As can be seen from table 1, as the frame length L increases, the average error rate of demodulation and the average error rate of decoding decrease, i.e. their corresponding performances improve, but the speed of improving the performances corresponding to the average error rate gradually decreases or even decreases, because the larger the number of received sample points, the worse the clustering performance (i.e. the lower the clustering accuracy). For example, when the value of the frame length L is changed from 75 to 150, the relative improvement ratios of the demodulated average bit error rate and the decoded average bit error rate are increased from 65.60% to 72.81% and from 45.41% to 59.44%, respectively. By comparing the performance improvement corresponding to the demodulated average bit error rate and the decoded average bit error rate, when the frame length L is too small, the rate of improvement of the decoding performance by the weighted integration clustering algorithm of the present disclosure becomes very small because the demodulation error exceeds the error correction capability of the employed channel coding 100.

TABLE 2

In some examples, as shown in table 2, table 2 shows the effect of the Nakagami channel parameter m on the average normalized mutual information, the correct rate and the average bit error rate, where L ═ 100. As can be seen from table 2, as the value of m increases, both the average normalized mutual information and the accuracy rate increase, indicating that the wireless channel 20 with less channel fading significantly improves the clustering performance. In addition, as the value of m increases, the performance corresponding to the average bit error rate of demodulation and the average bit error rate of decoding is improved, which indicates that channel fading also affects the final performance.

As described above, as can be seen from tables 1 and 2, the weighted ensemble clustering algorithm of the present embodiment can be optimally selected.

Fig. 12 is a waveform diagram illustrating average normalized mutual information with received signal-to-noise ratio under random phase noise according to an example of the present disclosure. Fig. 13 is a waveform diagram illustrating average bit error rate with received signal-to-noise ratio under random phase noise according to an example of the present disclosure. As shown in fig. 12 and 13, quadrature phase shift keying modulation systems are adopted, χ is 0.1, and B is _L T _b ＝2，ρ＝1-10 ^-3 ，L＝300，m＝1.5。

In some examples, as shown in fig. 12, fig. 12(a) shows random phase noise caused by an imperfect phase-locked loop circuit, and its corresponding probability density function can be obtained by equation (2), and fig. 12(b) shows random phase noise caused by an imperfect channel estimation, and its corresponding probability density function can be obtained by equation (3). As shown in fig. 12 and 8, the average normalized mutual information of fig. 12 is almost the same as the variation of fig. 8. Since the value of the average normalized mutual information is almost independent of the fixed phase noise but dependent on the received signal-to-noise ratio.

In some examples, as shown in fig. 13, fig. 13(a) shows random phase noise caused by an imperfect phase-locked loop circuit, and its corresponding probability density function can be obtained by equation (2), and fig. 13(b) shows random phase noise caused by an imperfect channel estimation, and its corresponding probability density function can be obtained by equation (3). As shown in fig. 13, the average bit error rate is obtained by averaging the mixture of the large phase noise and the small phase noise, and the proposed phase noise compensation method still shows its superiority when the small phase noise is dominant, as shown in fig. 13 (a). As can be seen from comparing the results of fig. 13(a) and 9(b), since the phase noise is random rather than fixed, the gap of improvement between the general schemes of the phase noise compensation method of the present disclosure becomes small, and the demodulation performance approaches the decoding performance of the general schemes. However, the phase noise compensation method and the general scheme of the present disclosure may have a lower limit of bit error when large phase noise is dominant.

Fig. 14 is a system block diagram illustrating phase noise compensation for a weighted ensemble clustering based wireless communication system in accordance with an example of the present disclosure.

The present disclosure relates to a phase noise compensation system 1 for a wireless communication system based on weighted ensemble clustering. The phase noise compensation system 1 includes a transmitting apparatus 50 and a receiving apparatus 60. In the present disclosure, the transmitting device 50 in the phase noise compensation system 1 may be similar to the transmitting terminal 10 in the phase noise compensation method, and the receiving device 60 may be similar to the receiving terminal 30 in the phase noise compensation method.

In some examples, as shown in fig. 14, the phase noise compensation system 1 may include a transmitting apparatus 50 and a receiving apparatus 60. In some examples, the transmitting device 50 may transmit a signal to the receiving device 60 and be received by the receiving device 60.

In some examples, transmitting apparatus 50 may transmit a carrier signal to a wireless channel based on channel coding, baseband modulation, and radio frequency modulator modulation, the carrier signal obtaining a received signal over the wireless channel. The transmitting device 50 may send information to the receiving device 60. The specific process can be seen in step S10 in the phase noise compensation method described above.

In some examples, the receiving device 60 may receive a received signal, obtain a baseband signal from the received signal based on rf demodulation and phase-locked loop circuitry, and obtain a gain baseband signal based on the baseband signal and automatic gain control. The receiving device 60 can receive and process the received signal. The specific process can be seen in step S20 in the phase noise compensation method described above.

In some examples, the receiving apparatus 60 may obtain, based on the cluster model and the gain baseband signal, each standard constellation point and a plurality of clusters corresponding to a plurality of sample points in a constellation corresponding to the gain baseband signal and a plurality of cluster center points in one-to-one correspondence with each cluster. The clustering model may be a weighted ensemble clustering algorithm, which may be determined based on a plurality of different clustering algorithms and gain baseband signals. Thereby, a plurality of clusters corresponding to a plurality of sample points in the constellation corresponding to the gain baseband signal, cluster center points corresponding to the respective clusters, and respective standard constellation points can be obtained. The specific process can be seen in step S30 in the phase noise compensation method described above.

In some examples, the receiving device 60 may obtain a norm distance between any cluster center point and each standard constellation point based on the distance calculation model, and further select a target constellation point having a minimum norm distance from the cluster center point from the plurality of standard constellation points. Thus, the target constellation point corresponding to each cluster center point can be obtained, and the target constellation point corresponding to each cluster can be obtained. The specific process can be seen in step S40 in the phase noise compensation method described above.

In some examples, the receiving apparatus 60 may replace the coordinates of the sample point corresponding to each cluster to the coordinates of the target constellation point corresponding to the cluster based on the cluster mapping model to implement phase noise compensation, thereby obtaining the target received signal, which is obtained based on baseband demodulation, channel decoding, and the target received signal. The receiving device 60 can thereby obtain the information transmitted by the transmitting device 50 relatively accurately. The specific process can be seen in step S50 in the phase noise compensation method.

As described above, in the present disclosure, the transmitting device 50 transmits a carrier signal to a wireless channel based on channel coding, baseband modulation, and radio frequency modulation, the carrier signal obtains a received signal through the wireless channel, the receiving device 60 receives the received signal and obtains a baseband signal therefrom, obtains a gain baseband signal based on the baseband signal and automatic gain control, further obtains a plurality of clusters corresponding to a plurality of sample points in a constellation corresponding to the gain baseband signal and a plurality of cluster center points corresponding to the respective clusters based on a cluster model, obtains a norm distance between any one cluster center point and each standard constellation point based on a distance calculation model, marks the standard constellation point corresponding to the minimum norm distance of the cluster center points as a target constellation point, and further replaces coordinates of the sample point corresponding to each cluster to coordinates of the target point corresponding to the cluster based on the cluster mapping model to implement phase noise compensation, and thus a target received signal is obtained, and the receiving apparatus 60 obtains the target signal based on the baseband demodulation, the channel decoding, and the target received signal. Thus, the receiving apparatus 60 can compensate for the generated phase noise, and can obtain the information transmitted by the transmitting apparatus 50 relatively accurately.

While the present disclosure has been described in detail in connection with the drawings and the examples, it should be understood that the above description is not intended to limit the present disclosure in any way. Those skilled in the art can make modifications and variations to the present disclosure as needed without departing from the true spirit and scope of the disclosure, which fall within the scope of the disclosure.

Claims

1. A phase noise compensation method of a wireless communication system based on integrated clustering is a phase noise compensation method of a wireless communication system with a transmitting end and a receiving end,

the method comprises the following steps:

the transmitting terminal transmits a carrier signal to a wireless channel based on channel coding, baseband modulation and radio frequency modulation, and the carrier signal obtains a received signal through the wireless channel;

the receiving end receives the received signal, obtains a baseband signal from the received signal based on radio frequency demodulation and a phase-locked loop circuit, obtains a gain baseband signal based on the baseband signal and automatic gain control, obtains a plurality of standard constellation points, a plurality of clusters corresponding to a plurality of sample points corresponding to the gain baseband signal and a plurality of cluster center points corresponding to the clusters one by one based on a cluster model and the gain baseband signal, further obtains norm distances between any cluster center point and each standard constellation point based on a distance calculation model, further selects a target constellation point having the minimum norm distance from the cluster center points from the plurality of standard constellation points, and replaces the coordinates of the sample points corresponding to each cluster to the coordinates of the target constellation point corresponding to the cluster based on a cluster mapping model to realize phase noise compensation, thereby obtaining a target reception signal, obtaining the target signal based on baseband demodulation, channel decoding, and the target reception signal,

wherein the clustering model is a weighted integration clustering algorithm for adaptively selecting a proper clustering algorithm under different conditions, a clustering result corresponding to each clustering algorithm is obtained based on a plurality of different clustering algorithms and the gain baseband signal, a co-clustering indication matrix corresponding to each clustering result is obtained based on each clustering result, a set matrix is obtained based on the clustering results and the co-clustering indication matrix, and the weighted integration clustering algorithm is obtained based on the set matrix and preset parameters,

the norm distance between the ith clustering center point and the jth standard constellation point meets the following requirements: d _ij ＝||C _i -S _j || ² 1, M, j, M, wherein C is C _i Is the ith cluster center point, S _j Is the jth standard constellation point, M is the modulation order of the multi-system frequency shift keying system, and the receiving end is based on

Obtaining a standard constellation point with the minimum norm distance with the ith clustering center point, marking the standard constellation point as a target constellation point of the ith clustering center point,

in the replacement, each clustering region is moved to move the coordinate of the clustering center point corresponding to each clustering region to the coordinate of the target constellation point corresponding to the clustering center point.

2. The phase noise compensation method according to claim 1, characterized in that:

the modulation order of the wireless communication system is known by the receiving end, and the number of the plurality of clusters is the same as the modulation order.

3. The phase noise compensation method according to claim 1, characterized in that:

and the target receiving signal is obtained after the coordinates corresponding to the clustering central points are converted into the coordinates of the corresponding target constellation points.

4. The phase noise compensation method according to claim 1, characterized in that:

the plurality of different clustering algorithms include a K-means clustering algorithm, a K-center clustering algorithm, and a coacervation hierarchy clustering algorithm.

5. A phase noise compensation system of a wireless communication system based on integrated clustering is a phase noise compensation system of a wireless communication system with a transmitting device and a receiving device,

the method comprises the following steps:

the transmitting device transmits a carrier signal to a wireless channel based on channel coding, baseband modulation and radio frequency modulator modulation, wherein the carrier signal obtains a receiving signal through the wireless channel;

the receiving device receives the received signal, obtains a baseband signal from the received signal based on radio frequency demodulation and a phase-locked loop circuit, obtains a gain baseband signal based on the baseband signal and automatic gain control, obtains a plurality of standard constellation points, a plurality of clusters corresponding to a plurality of sample points corresponding to the gain baseband signal and a plurality of cluster center points corresponding to the clusters one by one based on a cluster model and the gain baseband signal, further obtains norm distances between any cluster center point and each standard constellation point based on a distance calculation model, further selects a target constellation point having a minimum norm distance from the cluster center points from the plurality of standard constellation points, and replaces the coordinates of the sample point corresponding to each cluster to the coordinates of the target constellation point corresponding to the cluster based on a cluster mapping model to realize phase noise compensation, thereby obtaining a target reception signal, obtaining the target signal based on baseband demodulation, channel decoding, and the target reception signal,

wherein the clustering model is a weighted integration clustering algorithm for adaptively selecting a proper clustering algorithm under different conditions, a clustering result corresponding to each clustering algorithm is obtained based on a plurality of different clustering algorithms and the gain baseband signal, a copolymerization indication matrix corresponding to each clustering result is obtained based on each clustering result, a set matrix is obtained based on the clustering results and the copolymerization indication matrix, and the weighted integration clustering algorithm is further obtained,

the norm distance between the ith clustering center point and the jth standard constellation point meets the following requirements: d is a radical of _ij ＝||C _i -S _j || ² 1, M, j, M, wherein C is C _i Is the ith cluster center point, S _j Is the j standard constellation point, M is the modulation order of the multi-system frequency shift keying system, and the receiving device is based on

6. The phase noise compensation system of claim 5, wherein:

a modulation order of the wireless communication system is known by the receiving device, and the number of the plurality of clusters is the same as the modulation order.

7. The phase noise compensation system of claim 5, wherein:

8. The phase noise compensation system of claim 5, wherein: