An information fusion method integrating Rayleigh channel perception and communication

Abstract

The present invention discloses an information fusion method for integrated Rayleigh channel perception and communication. A Rayleigh channel integration model is established in an OFDM system, and information is processed through a perception stage, a communication stage, and a fusion stage. The Rayleigh channel integration model includes a perception channel model and a communication channel model; in the perception stage, a posterior distribution of channel coefficients is output; in the communication stage, a fusion result is output; in the fusion stage, a posterior distribution of perception communication is output; and based on the mutual information principle, the perception communication mutual information is output through the posterior distribution of perception communication. The information fusion method for integrated Rayleigh channel perception and communication of the present invention has the characteristics of high precision and low information loss.

Description

Technical Field

[0001] The present invention relates to the technical field of information transmission and processing, and in particular to an information fusion method integrating Rayleigh channel perception and communication. Background Art

[0002] In a recently released framework recommendation, the International Telecommunication Union (ITU) identified communication-sensing integration as one of the typical application scenarios of 6G. Communication-sensing integration can provide wide-area multi-dimensional perception and obtain spatial information about unknown objects, connected devices, and the surrounding environment. This scenario will support innovative applications such as automation, safe driving, and digital twins, and combined with artificial intelligence, further enhance the perception of the physical environment.

[0003] Least squares (LS) and minimum mean square error (MMSE) are traditional channel estimation algorithms that are widely used in channel estimation. The basic idea of ​​these traditional channel estimation algorithms is that the transmitter transmits a known signal, parses the estimated channel coefficient based on the received known signal, and then uses the estimated channel coefficient for transmission communication. This process is actually a decision fusion idea, which is similar to a "hard decision", so there is bound to be information loss.

[0004] For Rayleigh fading channels, Proakis studied the channel capacity when the receiver knows the channel state information, but did not involve the perception of the channel coefficients. In 1997, Taricco and Elia proposed a method for calculating the channel capacity of Rayleigh fading channels when the channel state is unknown, but failed to consider the impact of channel perception performance. Currently, there is little research on the integration of Rayleigh channel perception and communication. Even if the channel coefficients and channel perception performance are integrated, high-precision effects cannot be achieved. Summary of the invention

[0005] Purpose of the invention: The present invention provides an information fusion method that integrates Rayleigh channel perception and communication with high precision and low information loss.

[0006] Technical solution: The information fusion method for integrating Rayleigh channel perception and communication of the present invention establishes a Rayleigh channel integration model in an OFDM system, processes information through a perception stage, a communication stage, and a fusion stage, and the information includes perception data and communication data; the OFDM system includes a sub-channel and a mother channel, and the method includes the following steps:

[0007] 1) Obtaining a priori distribution of channel coefficients, perception data, communication data, and a priori distribution of communication signals;

[0008] 2) Establishing a Rayleigh channel integration model, the Rayleigh channel integration model includes a perception channel model and a communication channel model; inputting the channel coefficient prior distribution, perception data, communication data, and communication signal prior distribution into the Rayleigh channel integration model; the Rayleigh channel integration model is:

[0009]

[0010] In the formula, y s is the perception data received in the perception stage; c Communication data received during the communication phase; x s The sent signal is known in the perception stage; x c is the signal sent in the communication phase, which is an unknown signal; h is the channel scattering coefficient; w s is the complex Gaussian white noise in the perception stage; w c is the complex Gaussian white noise in the communication stage;

[0011] 3) In the perception phase of the Rayleigh channel integrated model, the channel coefficient prior distribution, the perception data and the perception channel model are fused to output the channel coefficient posterior distribution; in the communication phase of the Rayleigh channel integrated model, the communication data and the communication channel model are fused to output the fusion result;

[0012] 4) In the fusion stage of the Rayleigh channel integrated model, the fusion result is fused with the posterior distribution of the channel coefficient to output the perceptual communication equivalent channel model; the communication signal prior distribution is fused with the perceptual communication equivalent channel model to output the perceptual communication posterior distribution;

[0013] 5) Based on the mutual information principle, the prior entropy and the posterior entropy are output through the posterior distribution of the perceived communication, and the prior entropy and the posterior entropy are calculated to output the mutual information of the perceived communication.

[0014] Furthermore, the communication signal prior distribution π(x c ) Send signal x in the communication phase c is the independent variable; the independent variable h of the channel coefficient prior distribution π(h) is the channel scattering coefficient, and the π(h) formula is as follows:

[0015]

[0016] Where h satisfies the complex Gaussian distribution with mean 0 and variance 1; π is a constant.

[0017] Furthermore, the channel scattering coefficient h is a constant in the sensing phase and the communication phase.

[0018] Furthermore, the communication stage sends a signal x c It is a constant value during the communication phase.

[0019] Furthermore, in step 3), during the fusion process, the communication channel model adopts the statistical form of the communication channel model p(y c |x c ,h) and the perceptual channel model adopts the statistical form of the perceptual channel model p(y s |h); the details are as follows:

[0020]

[0021]

[0022] In the formula, the variance of the complex Gaussian white noise in the perception stage is the same as the variance of the complex Gaussian white noise in the communication stage, both of which are N0; s is the perception data received in the perception stage; c Communication data received during the communication phase; x s The signal is known for the perception stage.

[0023] Furthermore, in step 3), the channel coefficient posterior distribution p(h|y s )for:

[0024]

[0025] In the formula, is the perceived signal-to-noise ratio; P s is the power allocated to the perception stage and satisfies |x s | 2 =P s ; N0 is the variance of complex Gaussian white noise in the perception stage.

[0026] Furthermore, in step 3), the perceptual communication equivalent channel model p(y c |x c ,y s )for:

[0027]

[0028] Further, in step 4), the perception communication posterior distribution p(x c |y c ,y s )for:

[0029]

[0030] In the formula, For complex variables x c The integral of π(x c ) represents x c Obey the prior distribution of communication signals.

[0031] Furthermore, in step 5), the prior entropy is x c At a given y s The prior entropy h(x c |y s ), the posterior entropy is about x c The posterior entropy h(x c |y c ,y s ), the x c The posterior entropy h(x c |y c ,y s )The formula is:

[0032]

[0033] Furthermore, in step 5), the perceived communication mutual information I(x c ;y c |y s ) is x c At a given y s The prior entropy under the condition and about x c The difference in the posterior entropy of :

[0034] I(x c ;y c |y s )=h(x c |y s )-h(x c |y c ,y s )

[0035] In the formula, x c With y s Independent of each other, h(x c |y s )=h(x c ).

[0036] Beneficial effects: The present invention has the following significant effects: 1. Low information loss is achieved; the present invention establishes an integrated model of Rayleigh channels, constructs a perception channel model and a communication channel model in the integrated model, and fuses the results of the two models processed in the perception stage and the communication stage in the fusion stage, thereby reducing information loss. 2. High accuracy of captured information; the present invention establishes a connection between the two situations where CSI is completely known and completely unknown, and defines the mutual information obtained after fusion as perception communication mutual information. The information captured is highly accurate and provides theoretical support for the perception communication integrated ISAC system and OFDM subsystem. BRIEF DESCRIPTION OF THE DRAWINGS

[0037] Figure 1 is a flow chart of the information fusion method of the present invention;

[0038] Figure 2 This is a simulation result diagram of the posterior distribution of perceptual communication;

[0039] Figure 3 Figure 2 is the simulation result of sensing communication mutual information. DETAILED DESCRIPTION

[0040] The present invention is further explained below in conjunction with the accompanying drawings and specific embodiments. It should be understood that the following specific embodiments are only used to illustrate the present invention and are not used to limit the scope of the present invention. After reading the present invention, various equivalent forms of modifications to the present invention by those skilled in the art all fall within the scope defined by the claims attached to this application.

[0041] See also Figure 1 As shown, the present invention discloses an information fusion method for integrating Rayleigh channel perception and communication, establishes a Rayleigh channel integration model in an OFDM system, processes information through a perception stage, a communication stage, and a fusion stage, and the information includes perception data and communication data. The OFDM system includes a subchannel and a mother channel. The information fusion method for integrating Rayleigh channel perception and communication includes the following steps:

[0042] 1) Obtain the prior distribution of channel coefficients, perception data, communication data and communication signal prior distribution.

[0043] 2) Establishing a Rayleigh channel integration model, which includes a perception channel model and a communication channel model. Inputting the channel coefficient prior distribution, perception data, communication data, and communication signal prior distribution into the Rayleigh channel integration model. The Rayleigh channel integration model is:

[0044]

[0045] In the formula, y s is the perception data received in the perception stage; c Communication data received during the communication phase; x s The sent signal is known in the perception stage; x c is the signal sent in the communication phase, which is an unknown signal; h is the channel scattering coefficient; w s is the complex Gaussian white noise in the perception stage; w c is the complex Gaussian white noise in the communication stage; where x s Satisfy|x s | 2 =P s , P s The power allocated to the perception stage, x c Satisfy E[|x c |2 ]=P c , P c The power allocated for the communication phase; w s and w c The mean is 0 and the variance is N0.

[0046] 3) In the perception stage of the Rayleigh channel integrated model, the channel coefficient prior distribution, the perception data and the perception channel model are integrated to output the channel coefficient posterior distribution. Among them, the perception channel model (1) adopts the statistical form of the common perception channel model, which is p(y s |h); the details are as follows:

[0047]

[0048] Where, the variance of complex Gaussian white noise in the perception stage is N0; y s is the perception data received in the perception stage; x s The signal is known for the perception stage.

[0049] The posterior distribution of the channel coefficient p(h|y s )for:

[0050]

[0051] In the formula, is the perceived signal-to-noise ratio; N0 is the variance of the complex Gaussian white noise in the perception stage; π is a constant, which is taken as 3.14.

[0052] In the communication phase of the Rayleigh channel integrated model, the communication data and the communication channel model are fused to output the fusion result. Among them, the communication channel model (2) adopts the statistical form of the communication channel model, which is p(y c |x c ,h); the details are as follows:

[0053]

[0054] Where, the variance of complex Gaussian white noise in the communication stage is N0; y c The communication data received during the communication phase.

[0055] 4) In the fusion stage of the Rayleigh channel integrated model, the fusion results of the communication data and the communication channel model are fused with the posterior distribution of the channel coefficients to output the perceptual communication equivalent channel model. c |x c ,y s )for:

[0056]

[0057] The communication signal prior distribution is combined with the perceptual communication equivalent channel model to output the perceptual communication posterior distribution; the perceptual communication posterior distribution p(x c |y c ,y s )for:

[0058]

[0059] In the formula, For complex variables x c The integral of π(x c ) represents x c Obey the prior distribution of communication signals.

[0060] 5) Based on the mutual information principle, the prior entropy and posterior entropy outputted by the posterior distribution of the perceptual communication are calculated and the prior entropy and posterior entropy are used to output the perceptual communication mutual information. Perceptual communication mutual information I(x c ;y c |y s ) specifically:

[0061] I(x c ;y c |y s )=h(x c |y s )-h(x c |y c ,y s ) (8)

[0062] In the formula, h(x c |y s ) is x c At a given y s The prior entropy under the condition, due to x c With y s are independent of each other, so h(x c |y s )=h(x c );h(x c |y c ,y s ) is about x c The posterior entropy of .

[0063] Among them, the communication posterior distribution model p(y c |x c ,h), channel coefficient posterior distribution p(h|y s ), perceptual communication equivalent channel model p(y c |x c ,y s ), the posterior distribution of perceptual communication p(x c |y c,y s ) are all in the form of probability density functions. In addition, the information fusion method processes information through the perception stage, communication stage, and fusion stage in sequence, and the process can be performed multiple times. The channel scattering coefficient h has different values ​​in the multiple processing processes, but is a constant value in the perception stage and the communication stage of the same processing process. The communication stage sends a signal x c The value of is different in the multiple processing processes, but is a constant value in the communication phase of the same processing process. The information fusion method is specifically explained below.

[0064] In step 1), the communication signal prior distribution π(x c ) Send signal x in the communication phase c is the independent variable, and the signal x is sent in the communication phase c It is a constant in the communication stage. The independent variable h of the channel coefficient prior distribution π(h) is the channel scattering coefficient, and the channel scattering coefficient h is a constant in the perception stage and the communication stage. The formula (9) of π(h) is as follows:

[0065]

[0066] Where h satisfies the complex Gaussian distribution with mean 0 and variance 1.

[0067] In step 3), the posterior distribution of the channel coefficient p(h|y s ) The acquisition process is:

[0068] Because of w s It obeys the complex Gaussian distribution. When h is given, the statistical form of the perceptual channel model can be obtained from formula (1):

[0069]

[0070] x s It is known that, then equation (3) is adjusted to equation (9):

[0071] y s '=h+w s '(9)

[0072] In the formula, y s '=y s / x s ;w s '=w s / x s ;x s For a known sent signal, satisfying |x s | 2 =P s , P s The power allocated to the perception stage, w sThe mean is 0 and the variance is N0, so w s 'Satisfy the mean value is 0 and the variance is ρ s -2 =(P s / N0) -1 The complex Gaussian distribution, ρ s 2 is the signal-to-noise ratio in the perception stage.

[0073] Since h is a complex parameter, and h R is the corresponding real part of h; h I is the corresponding imaginary part of h. Therefore, substituting the real part into equation (9) and converting it into the real part, we get equation (10):

[0074] y sR '=h R +w sR ' (10)

[0075] In the formula, y sR ' is y s ' real part; w sR ' is w s ' real part; w sR 'Satisfy the mean value is 0 and the variance is ρ s -2 / 2 Gaussian distribution.

[0076] For formula (10), when h is given, we have formula (11):

[0077]

[0078] From (9), we can get h R The prior distribution (12) of is as follows:

[0079]

[0080] According to the properties of the conjugate prior distribution, we get equation (13), and the real part of equation (3) is now complete:

[0081]

[0082] Similarly, converting (3) into the imaginary part yields (14):

[0083]

[0084] In the formula, y' sI for y' s The imaginary part of .

[0085] Combining equations (13) and (14) yields equation (15), which is then rearranged to yield equation (4), which is the posterior distribution of the channel coefficient p(h|ys ),as follows:

[0086]

[0087]

[0088] In step 3), w c Obeying the complex Gaussian distribution, given h and x c Under the condition of , formula (2) can be used to obtain formula (5), which is the statistical form of the communication channel model p(y c |x c ,h) is:

[0089]

[0090] In step 4), the perceptual communication equivalent channel model p(y c |x c ,y s )The solution process is:

[0091] In the fusion stage of the Rayleigh channel integrated model, the fusion results of the communication data and the communication channel model are fused with the posterior distribution of the channel coefficients. Formula (6) is the perceptual communication equivalent channel model p(y c |x c ,y s ):

[0092]

[0093] In step 4), the process of solving the posterior distribution of perceptual communication is as follows: based on the Bayesian formula, the communication signal prior distribution is integrated with the perceptual communication equivalent channel model (6), and then the perceptual communication posterior distribution (7) is output. c ) Send signal x in the communication phase c is the independent variable, and the signal x is sent in the communication phase c is an unknown constant. The obtained perceptual communication posterior distribution p(x c |y c ,y s )for:

[0094]

[0095] In the formula, For complex variables x c The integral of π(x c ) represents x c Obey the prior distribution of communication signals.

[0096] In step 5), the prior entropy and the posterior entropy are output by sensing the posterior distribution of communication based on the mutual information principle.c At a given y s The prior entropy h(x c |y s ), due to x c With y s are independent of each other, so h(x c |y s )=h(x c ), h(x c ) is about x c The prior entropy of :

[0097]

[0098] The posterior entropy is about x c The posterior entropy h(x c |y c ,y s ), x c The posterior entropy h(x c |y c ,y s ) formula is

[0099]

[0100] The prior entropy and the posterior entropy are calculated to output the perceptual communication mutual information. The perceptual communication mutual information is the real mutual information, and the information unit is bit, which is used to quantify the trade-off. Perceptual communication mutual information I(x c ;y c |y s ) is x c At a given y s The prior entropy under the condition and about x c The difference in the posterior entropy is:

[0101] I(x c ;y c |y s )=h(x c |y s )-h(x c |y c ,y s )(8)

[0102] In the formula, h(x c |y s )=h(x c ), h(x c ) is about x c The prior entropy of .

[0103] In order to verify the effectiveness of the information fusion method for integrating Rayleigh channel perception and communication of the present invention, a simulation test was carried out on the method of the present invention.

[0104] See also Figure 2 As shown, Figure 2 Under the condition of different perception signal-to-noise ratio and communication signal-to-noise ratio (perception signal-to-noise ratio and communication signal-to-noise ratio are both 15dB), the perception communication fusion method of the present invention is applied to obtain the perception communication posterior distribution simulation results in the form of a three-dimensional scatter plot. When the perception signal-to-noise ratio and the communication signal-to-noise ratio are both very high, the perception communication posterior distribution is distributed at the real position of the target with a very high probability, so for x c The estimation error is small.

[0105] See also Figure 3 As shown, Figure 3 The simulation results of the mutual information of Rayleigh channel perception and communication under different conditions of perception signal-to-noise ratio and communication signal-to-noise ratio. When the communication signal-to-noise ratio is fixed, the larger the perception signal-to-noise ratio, the greater the perception communication mutual information I(x c ;y c |y s ). At the same time, as the perceived signal-to-noise ratio increases, the perceived communication mutual information continues to approach the channel capacity when the channel state (CSI) is known. This is because, as the perceived signal-to-noise ratio continues to increase, the perceived channel coefficient will be infinitely close to the true value, that is, close to the situation where the channel is completely known.

[0106] In summary, the information fusion method for integrating Rayleigh channel perception and communication of the present invention has the characteristics of high precision and low information loss.

Claims

1. A Rayleigh channel sensing and communication integrated information fusion method, characterized in that: An integrated Rayleigh channel model is established in an OFDM system, and information is processed through a perception stage, a communication stage, and a fusion stage, wherein the information includes perception data and communication data; the OFDM system includes a subchannel and a mother channel, and the method includes the following steps: 1) Obtaining a priori distribution of channel coefficients, perception data, communication data, and a priori distribution of communication signals; 2) Establishing a Rayleigh channel integration model, the Rayleigh channel integration model includes a perception channel model and a communication channel model; inputting the channel coefficient prior distribution, perception data, communication data, and communication signal prior distribution into the Rayleigh channel integration model; the Rayleigh channel integration model is: In the formula, y s is the perception data received in the perception stage; c is the communication data received during the communication phase; x s The sent signal is known in the perception stage; x c is the signal sent in the communication phase, which is an unknown signal; h is the channel scattering coefficient; w s is the complex Gaussian white noise in the perception stage; w c is the complex Gaussian white noise in the communication stage; 3) In the perception phase of the Rayleigh channel integrated model, the channel coefficient prior distribution, the perception data and the perception channel model are fused to output the channel coefficient posterior distribution; in the communication phase of the Rayleigh channel integrated model, the communication data and the communication channel model are fused to output the fusion result; 4) In the fusion stage of the Rayleigh channel integrated model, the fusion result is fused with the posterior distribution of the channel coefficient to output the perceptual communication equivalent channel model; the communication signal prior distribution is fused with the perceptual communication equivalent channel model to output the perceptual communication posterior distribution; 5) Based on the mutual information principle, the prior entropy and the posterior entropy are output through the posterior distribution of the perceived communication, and the prior entropy and the posterior entropy are calculated to output the mutual information of the perceived communication; The communication signal prior distribution π(x c ) Send signal x in the communication phase c is the independent variable; the independent variable h of the channel coefficient prior distribution π(h) is the channel scattering coefficient, and the π(h) formula is as follows: Where h satisfies the complex Gaussian distribution with mean 0 and variance 1; π is a constant; The channel scattering coefficient h is a constant value in the sensing phase and the communication phase; The communication phase sends a signal x c A constant value in one of the communication phases; In step 3), during the fusion process, the communication channel model adopts the statistical form of the communication channel model p(y c |x c ,h) and the perceptual channel model adopts the statistical form of the perceptual channel model p(y s |h); the details are as follows: In the formula, the variance of the complex Gaussian white noise in the perception stage is the same as the variance of the complex Gaussian white noise in the communication stage, both of which are N0; In step 3), the channel coefficient posterior distribution p(h|y s )for: In the formula, is the perceived signal-to-noise ratio; P s is the power allocated to the perception stage and satisfies |x s | 2 =P s ; In step 4), the perceptual communication equivalent channel model p(y c |x c ,y s )for: In step 4), the posterior distribution p(x) of the perceptual communication is obtained based on the Bayesian formula. c |y c ,y s )for: In the formula, For complex variables x c The integral of π(x c ) represents x c Obey the prior distribution of communication signals; In step 5), the prior entropy is x c At a given y s The prior entropy h(x c |y s ), the posterior entropy is about x c The posterior entropy h(x c |y c ,y s ), the x c The posterior entropy h(x c |y c ,y s )The formula is: In step 5), the perceived communication mutual information I(x c ;y c |y s ) is x c At a given y s The prior entropy under the condition and about x c The difference in the posterior entropy of : I(x c ;y c |y s )=h(x c |y s )-h(x c |y c ,y s ); In the formula, x c With y s Independent of each other, h(x c |y s )=h(x c ).

Images