Intelligent reflector-assisted symbiotic sense semantic bit secure transmission method

Abstract

The invention provides an intelligent reflector-assisted symbiotic sense semantic bit secure transmission method, and relates to the technical field of wireless communication. Due to the fact that an eavesdropper only has a semantic-oriented communication structure, the intelligent reflecting surface can modulate a bit signal generated by the intelligent reflecting surface into an incident semantic signal through reflected beam forming, and therefore the intelligibility of semantic data is destroyed. The superposed signals can construct a mutualism relationship for expected legal users, and form a parasitic interference relationship for eavesdroppers at the same time. Based on the constructed system model, a semantic security rate maximization problem is constructed, and the problem needs to meet the lowest service quality requirements of bit data decoding and target perception. According to the method, an original problem is decomposed into a plurality of sub-problems, the sub-problems can be solved by deriving a closed-form solution or converting the sub-problems into an equivalent convex form, and the method is remarkably superior to a traditional reference scheme in the aspect of semantic security.

Description

TECHNICAL FIELD

[0001] The application belongs to the technical field of wireless communication, and particularly relates to a smart reflecting surface assisted symbiotic perception semantic bit secure transmission method. BACKGROUND

[0002] In the current acceleration of digitalization and intelligentization, the sixth generation mobile communication (6G) has become the key driving force for the future development of the information society. In this context, the concept of integrated sensing and communication (ISAC) breaks the functional boundaries of traditional communication systems and can simultaneously achieve communication and environmental sensing tasks, laying a solid foundation for various applications such as large-scale unmanned aerial vehicle management and control, intelligent logistics networks, and collaborative autonomous driving systems. Through efficient resource management, integrated sensing and communication can achieve multi-dimensional coordination of spectrum, hardware, and computing power. To balance the sensing and communication capabilities, the research on pulse-based integrated sensing and communication systems mainly focuses on signal waveform design and beamforming strategies and other technical challenges. However, the redundancy of sensing data feedback puts additional burden on the channel, and the growing demand for data and the problem of spectrum scarcity lead to the gradual approach to the Shannon limit of transmission performance. Therefore, the paper "Beyond transmitting bits: Context, semantics, and task-oriented communications" integrates the innovative paradigm of semantic communication with the efficient framework of integrated sensing and communication, proposing an integrated sensing and communication semantic (ISASC) system. In particular, the flow and transmission of semantic and sensing information at the physical layer in the integrated sensing and communication semantic system make the system security inherently vulnerable. That is, when an eavesdropper can obtain the complete semantic knowledge base and background information, the redundant semantic representation in the data compression process may inadvertently leak sensitive information, thereby posing serious security risks.

[0003] Although the researches "A physical layer security framework for integrated sensing and semantic communication systems" and "Secure resource allocation for integrated sensing and semantic communication system" explore the physical layer security of the integrated communication-sensing-semantic system, there are still deficiencies in resisting external eavesdropping through the cooperative jamming mechanism based on symbiotic radio. At the same time, the literature "Semantic communication-assisted physical layer security over fading wiretap channels" does not study the encryption of semantic stream information by symbiotic radio using bit stream information. In addition, the literature "Integrated sensing and communication with symbiotic radio and RIS-assisted backscatter" studies the role of symbiotic radio in improving the transmission performance of bit-oriented systems, but does not consider the eavesdropping problem brought by environmental openness. Although the literature "Improving physical layer security with RIS-assisted symbiotic radio" uses artificial noise generated by symbiotic radio to enhance physical layer security, there is still a lack of research on the trade-off between sensing performance and semantic security. SUMMARY

[0004] In view of the deficiencies of the prior art, the present application provides an intelligent reflecting surface assisted symbiotic communication-sensing-semantic bit security transmission method, which uses symbiotic radio as a cooperative transmission mode for bits and semantics, and considers the information leakage problem in the wireless channel. When the eavesdropper only has a traditional semantic-oriented communication structure, the bit stream generated by the symbiotic radio not only improves the spectral efficiency of the system, but also can be used as an interference signal to reduce the accuracy of the eavesdropper extracting semantic information, aiming to improve the sensing performance and data security at the same time.

[0005] The technical scheme of the present application is:

[0006] The present application provides an intelligent reflecting surface assisted symbiotic communication-sensing-semantic bit security transmission method, comprising the following steps:

[0007] Constructing a communication-sensing-semantic integrated system;

[0008] Modeling a communication and perception semantic integrated system;

[0009] According to the modeling result, a semantic security rate optimization problem is constructed, and an optimal solution is obtained by solving the semantic security rate optimization problem, including a beamforming vector corresponding to communication and perception, an RIS reflection coefficient vector, a receiving beamforming vector, and an average number of semantic symbol signals after encoding each data segment.

[0010] Further, the communication and perception semantic integrated system includes one multifunctional semantic base station, K single-antenna communication users, one single-antenna potential eavesdropper, one single-antenna perception target, and one intelligent reflecting surface with N reflecting units;

[0011] The multifunctional semantic base station works in a full-duplex mode and is equipped with a uniform linear array, wherein The uniform linear array composed of M antennas is used to broadcast a joint transmission signal composed of semantic symbol signals and perception signals, The uniform linear array composed of M antennas is used to receive echo signals reflected by the single-antenna perception target, and , M is the number of antennas; and the semantic symbol signal is a signal carrying semantic information;

[0012] The intelligent reflecting surface with N reflecting units adopts a reflection modulation strategy, that is, modulates a bit signal generated by itself onto the joint transmission signal to generate a reflection signal and reflect it, thereby realizing non-line-of-sight information transmission from the multifunctional semantic base station to the K single-antenna communication users and providing communication services for the K single-antenna communication users;

[0013] The K single-antenna communication users receive the joint transmission signal sent by the multifunctional semantic base station and the reflection signal generated by the intelligent reflecting surface;

[0014] The single-antenna potential eavesdropper eavesdrops by receiving the joint transmission signal sent by the multifunctional semantic base station and the reflection signal generated by the intelligent reflecting surface;

[0015] The single-antenna perception target reflects the joint transmission signal to generate an echo signal and returns it to the multifunctional semantic base station.

[0016] Further, the communication and perception semantic integrated system is modeled, specifically including the following steps:

[0017] A1: modeling the semantic symbol signal generated by the multifunctional semantic base station;

[0018] The input original text is first subjected to semantic encoding to extract semantic representation and generate semantic data wherein represents the i-th semantic data in the semantic data a data segment, the number of data segments and , the total number of data segments at the multi-functional semantic base station; semantic data is compressed into a semantic symbol vector through a semantic compression mapping function , denotes a semantic symbol signal whose power has been normalized, denotes the serial number of the semantic symbol signal, denotes the length of the semantic data, the average number of semantic symbol signals after encoding each data segment; mapped into a transmissible physical signal waveform by a semantic channel encoder and finally sent out by the multi-functional semantic base station;

[0019] A2: modeling the signal received by a single-antenna communication user, the signal-to-interference-and-noise ratio, and the signal-to-noise ratio of bit signal decoding in the communication process;

[0020] The multi-functional semantic base station broadcasts the downlink semantic symbol signal and the sensing signal to all single-antenna communication users, and the joint transmission signal of the multi-functional semantic base station is denoted as wherein denotes the th joint transmission signal, denotes the sensing signal in the th joint transmission signal, and are the beamforming vectors corresponding to communication and sensing, respectively; let , and denote the complex baseband equivalent channels from the multi-functional semantic base station to the th single-antenna communication user, the intelligent reflecting surface, and the single-antenna potential eavesdropper, denotes a complex number set, denotes the number of single-antenna communication users; the complex baseband equivalent channels from the intelligent reflecting surface to the th single-antenna communication user and the single-antenna potential eavesdropper are and ; the phase shift matrix of the intelligent reflecting surface is wherein denotes the RIS reflection coefficient vector, denotes the base number of the natural logarithm, denotes the phase shift of the th reflecting unit, and , denotes the number of reflecting units, denotes the imaginary unit, and Each element satisfies the unit modulus constraint. The smart reflecting surface adopts binary phase shift keying technology to modulate the bit signal generated by itself onto the joint transmission signal, and sets bit signal generated by the smart reflecting surface, then the signal received by the

[0021] (1) ;

[0022] wherein, represents the signal received by the represents the transpose of the matrix, represents the additive white Gaussian noise at the represents the noise variance of the

[0023] The signal-to-interference-noise ratio of the

[0024] (2) ;

[0025] wherein, represents the signal-to-interference-noise ratio of the

[0026] The signal-to-noise ratio of bit signal decoding is represented as:

[0027] (3) ;

[0028] wherein, represents the signal-to-noise ratio of bit signal decoding, represents the multiple of the secondary symbol period compared with the primary symbol period, and is an integer and satisfies ;

[0029] A3: Model the signal and signal-to-interference-noise ratio received by the single-antenna potential eavesdropper;

[0030] For the single-antenna potential eavesdropper, the signal received by the single-antenna potential eavesdropper is represented as:

[0031] (4) ;

[0032] wherein, represents the signal received by the single-antenna potential eavesdropper, is the additive white Gaussian noise at the single-antenna potential eavesdropper, represents the noise variance of the receiving end of the single-antenna potential eavesdropper; ​​​​​​

[0033] The corresponding signal-to-interference-plus-noise ratio at the single-antenna potential eavesdropper is:

[0034] (5);

[0035] wherein, denotes the corresponding signal-to-interference-plus-noise ratio at the single-antenna potential eavesdropper;

[0036] A4: Model the echo signal and the sensing performance in the target sensing process;

[0037] The echo signal is described as:

[0038] (6);

[0039] wherein, denotes the echo signal, denotes the channel from the multi-functional semantic base station to the single-antenna sensing target, , and denote the antenna spacing, the wavelength and the angle of arrival, respectively, is the additive white Gaussian noise at the semantic base station, denotes the noise variance of the single-antenna sensing target, denotes the identity matrix;

[0040] The semantic base station adopts a receive beamforming vector , and the received echo signal is:

[0041] (7);

[0042] wherein, denotes a specific construction form of the channel matrix;

[0043] The sensing performance of the communication-sensing semantic integrated system is measured by the following formula:

[0044] (8);

[0045] wherein, denotes the sensing performance of the communication-sensing semantic integrated system;

[0046] A5: Model the semantic security rate;

[0047] The semantic rate expression is:

[0048] (9);

[0049] wherein, denotes the semantic rate, representing the allocated channel bandwidth, representing the semantic similarity function, representing the signal-to-noise ratio, representing a single-antenna communication user or a single-antenna potential eavesdropper, representing a single-antenna communication user, representing a single-antenna potential eavesdropper;

[0050] The semantic similarity function is represented as:

[0051] (10) ;

[0052] wherein, represents an analytical approximation of the semantic similarity function , , , , ,

[0053] The semantic transmission rate at the nth single-antenna communication user and the single-antenna potential eavesdropper is:

[0054] (11) ;

[0055] wherein, represents the semantic transmission rate at the nth single-antenna communication user and the single-antenna potential eavesdropper;

[0056] The semantic security rate of the nth single-antenna communication user is:

[0057] (12) ;

[0058] wherein, represents the achievable semantic transmission rate at the single-antenna communication user, represents the achievable semantic transmission rate at the single-antenna potential eavesdropper, represents a max operation.

[0059] Further, the semantic security rate optimization problem is:

[0060] (18) ;

[0061] (19) ;

[0062] (20) ;

[0063] ​​​(21);

[0064] (22);

[0065] (23);

[0066] wherein, denotes the maximum transmit power, and are the RIS bit transmission and the target-aware quality of service threshold, respectively, denotes the maximum value; is the equivalent channel matrix, is the equivalent channel matrix, is the equivalent reflection coefficient vector; is the RIS phase offset.

[0067] Further, the specific process for solving the semantic security rate optimization problem is:

[0068] B1: divide the semantic security rate optimization problem into four sub-problems, including the optimization problem of the beamforming vector corresponding to communication and perception, the optimization problem of the RIS reflection coefficient vector, the optimization problem of the receiving beamforming vector, and the optimization problem of the average number of semantic symbol signals after encoding each data segment;

[0069] B2: solve the optimization problem of the beamforming vector corresponding to communication and perception, the optimization problem of the RIS reflection coefficient vector, the optimization problem of the receiving beamforming vector, and the optimization problem of the average number of semantic symbol signals after encoding each data segment, to obtain the optimal solution .

[0070] Further, the construction process of the optimization problem of the beamforming vector corresponding to communication and perception is:

[0071] introduce auxiliary variables , and satisfy:

[0072] (25);

[0073] (26);

[0074] introduce auxiliary variables , and satisfy the following conditions:

[0075] (33);

[0076] (34);

[0077] (35);

[0078] (36);

[0079] Finally, the optimization problem of communicating and sensing corresponding beamforming vectors is represented as:

[0080] (57);

[0081] (19);

[0082] (41);

[0083] (42);

[0084] (43);

[0085] (44);

[0086] (45);

[0087] (48);

[0088] (50);

[0089] (51);

[0090] (53);

[0091] (54);

[0092] (55);

[0093] (56);

[0094] wherein, , ; denotes taking the real part of a complex number, , and denotes the solution obtained in the th iteration; is an auxiliary variable, , , , and is the the solution obtained by the sub-iteration; is the composite baseband equivalent channel from the reflecting elements to the single-antenna potential eavesdropper, , are all equivalent channel matrices.

[0095] Further, the construction process of the optimization problem of the RIS reflection coefficient vector is:

[0096] Introduce a new matrix variable , , , , is a unit matrix of size (M+1)×(M+1), let ; Introduce two groups of auxiliary variables and ;

[0097] Finally, the optimization problem of the RIS reflection coefficient vector is represented as:

[0098] (74);

[0099] (20);

[0100] (63);

[0101] (65);

[0102] (66);

[0103] (67);

[0104] (68);

[0105] (69);

[0106] (70);

[0107] (71);

[0108] (72);

[0109] (73);

[0110] wherein, denotes the sum of the main diagonal elements of the matrix, , , , , , , , , and the solution obtained in the th iteration; and the solution obtained in the th iteration.

[0111] Further, the optimization problem of the receive beamforming vector is constructed as:

[0112] (76);

[0113] The optimization problem of the average number of coded semantic symbol signals of each data segment is constructed as:

[0114] (78);

[0115] wherein, define = , = .

[0116] Further, the optimization problem of the communication and perception corresponding beamforming vector, the optimization problem of the RIS reflection coefficient vector, the optimization problem of the receive beamforming vector and the optimization problem of the average number of coded semantic symbol signals of each data segment are specifically:

[0117] B2.1: the communication and perception corresponding beamforming vector is solved by iteration through the CVX toolbox under the condition that the variable is given;

[0118] B2.2: the optimization problem of the RIS reflection coefficient vector is solved by the CVX toolbox;

[0119] Specifically, by using the Gaussian randomization technology, the suboptimal solution that maximizes the optimization problem of the RIS reflection coefficient vector can be obtained from the optimal solution , and then the suboptimal solution is projected to the unit modulus constraint to obtain , wherein, represents the M+1th component of the vector , represents the phase angle of ;

[0120] The conservative update rule is adopted:

[0121] (81);

[0122] wherein, denotes the RIS reflection coefficient vector of the i-th iteration, denotes the safe rate when the beamforming vector corresponding to the communication and the perception takes the i-th iteration solution , the RIS takes the arbitrary optimized reflection coefficient vector . denotes the safe rate when the beamforming vector corresponding to the communication and the perception takes the i-th iteration solution , the RIS takes the reflection coefficient vector of the i-th iteration solution . denotes the safe rate when the beamforming vector corresponding to the communication and the perception takes the i-th iteration solution , the RIS takes the reflection coefficient vector of the i-th iteration solution .

[0123] B2.3: solving the optimization problem of the receiving beamforming vector;

[0124] define , the optimal solution of the problem is the principal eigenvector of the matrix , and the principal eigenvector is taken as the optimal solution of the perception beamforming vector;

[0125] B2.4: solving the optimization problem of the average number of semantic symbol signals after encoding each data segment;

[0126] Regarding the variable is a monotonically decreasing function, and its optimal value is taken at the left end point of the interval, so is set to the minimum value allowed.

[0127] B2.5: integrating the optimal solution of each sub-problem as the final optimal solution.

[0128] Compared with the prior art, the beneficial effects of the present application are:

[0129] 1) As an alternative to traditional encryption and decryption technology, the present application proposes a semantic-bit coexistence communication and perception semantic integration framework based on intelligent reflecting surface assisted symbiotic radio, which guarantees the secure transmission of semantic information. Specifically, the bit signal generated by the intelligent reflecting surface is modulated into the semantic information. For legitimate users, the superimposed signal can be used as an additional multipath component, while also improving the spectral efficiency of the bit signal. However, since the eavesdropper only focuses on the interception and recovery of the semantic information, the bit signal is equivalent to artificial noise for the eavesdropper, and by introducing invalid or error bits, the eavesdropper can be effectively prevented from reconstructing the semantic information.

[0130] ​2) To further enlarge the difference between the legal user and the eavesdropper in semantic understanding, the application constructs a joint optimization problem considering the bit signal decoding demand and the target perception demand to maximize the semantic security rate. In view of the non-convexity of the problem, a solution method based on block coordinate descent is designed, which decomposes the original optimization problem into four sub-problems about the transmitting beamforming vector, the receiving beamforming, the smart reflecting surface phase shift and the semantic parameter. Subsequently, the non-convexity is processed by using the semi-definite programming (SDP) and the sequential convex approximation method, and then the optimal solution is obtained. BRIEF DESCRIPTION OF DRAWINGS

[0131] Figure 1 A schematic diagram of a communication-perception-semantic integrated system in an embodiment of the application;

[0132] Figure 2 A relationship diagram of SSR and SBS antenna number in an embodiment of the application;

[0133] Figure 3 A diagram of SSR and transmitting power in an embodiment of the application;

[0134] Figure 4 A diagram of SSR and semantic parameter Z in an embodiment of the application;

[0135] Figure 5 A relationship diagram of SSR and BS antenna number in an embodiment of the application;

[0136] Figure 6 A diagram of SSR and transmitting power in an embodiment of the application;

[0137] Figure 7 A diagram of SSR and RIS element number in an embodiment of the application;

[0138] Figure 8 A diagram of beam pattern gain along angle in an embodiment of the application. DETAILED DESCRIPTION

[0139] The application will be described in detail below with reference to the drawings and embodiments.

[0140] The embodiment provides a smart reflecting surface assisted symbiotic communication-perception-semantic integrated system.

[0141] Step 1: Construct a communication-perception-semantic integrated system;

[0142] The communication-perception-semantic integrated system, as shown in Figure 1 includes one multifunctional semantic base station, K single-antenna communication users, one single-antenna potential eavesdropper, one single-antenna perception target and one smart reflecting surface with N reflecting units.

[0143] The multi-functional semantic base station works in a full-duplex mode, is equipped with a uniform linear array (ULA) composed of N antennas, wherein the uniform linear array composed of N antennas is used to broadcast a joint transmission signal composed of a semantic symbol signal and a perception signal, the uniform linear array composed of N antennas is used to receive echo signals reflected by a single-antenna perception target, and N is the number of antennas; the semantic symbol signal is a signal carrying semantic information;

[0144] The smart reflecting surface with N reflecting units adopts a reflection modulation strategy, that is, modulates the bit signal generated by itself onto the joint transmission signal to generate a reflection signal and reflect it, thereby realizing non-line-of-sight (NLoS) information transmission from the multi-functional semantic base station to K single-antenna communication users and providing communication services for the K single-antenna communication users;

[0145] The K single-antenna communication users receive the joint transmission signal sent by the multi-functional semantic base station and the reflection signal generated by the smart reflecting surface;

[0146] The single-antenna potential eavesdropper eavesdrops by receiving the joint transmission signal sent by the multi-functional semantic base station and the reflection signal generated by the smart reflecting surface;

[0147] The single-antenna perception target reflects the joint transmission signal to generate an echo signal and returns it to the multi-functional semantic base station;

[0148] In the communication-perception-semantic integrated system, all legitimate terminals have joint decoding capability, that is, all single-antenna communication users can realize semantic information reconstruction and bit information decoding; and the single-antenna potential eavesdropper only focuses on intercepting semantic information. In addition, it is assumed that by using a low-complexity channel estimation algorithm, the channels of all single-antenna communication users, single-antenna perception targets and single-antenna potential eavesdroppers can be perfectly estimated.

[0149] ​The reflection modulation strategy is adopted in this embodiment because the legitimate user (single antenna communication user) has prior knowledge of intelligent reflecting surface modulation and coding, and the reflected signal is considered as multipath gain. At the same time, the intelligent reflecting surface will backscatter the bit signal generated by itself and transmit it to all users. With the unique location advantage of the intelligent reflecting surface in processing base station signals, this method can enable users to simultaneously obtain real-time environmental monitoring data and broadcast communication information, which not only significantly improves the situation awareness capability, but also creates new possibilities for the landing of various location-based services. However, for a single antenna potential eavesdropper, the reflected signal is equivalent to an artificial noise, which will destroy its understanding of the original semantic information - by mixing invalid or error bits, the eavesdropper may obtain an error but seemingly reasonable semantic result. In summary, intelligent reflecting surface assisted symbiotic radio has two core functions: on the one hand, it can intelligently reflect semantic base station to terminal signals, effectively improving communication quality and system coverage; on the other hand, the intelligent reflecting surface can carry and transmit local data to the terminal by modulating its reflection characteristics. In this way, the intelligent reflecting surface can fully utilize the base station signal to realize real-time broadcasting and bit interference generation, which not only improves the spectrum efficiency, but also provides a natural artificial noise generation scheme for protecting semantic-level information.

[0150] Step 2: Model the communication and perception semantic integrated system;

[0151] Step 2.1: Model the semantic symbol signal generated by the multifunctional semantic base station;

[0152] During semantic streaming, the multifunctional semantic base station carries the DeepSC (Deep Semantic Communication) model in the literature "Power-Efficient Optimization for Coexisting Semantic and Bit-Based Users in NOMA Networks" for text transmission. The DeepSC model can generate clear task information bodies containing deep intentions and contextual backgrounds;

[0153] Specifically, the input original text is first processed by a semantic encoder to extract semantic representations and generate semantic data wherein represents the i-th data segment in the semantic data, is the number of data segments and , is the total number of data segments at the multifunctional semantic base station; and the semantic data is compressed into a semantic symbol vector by a semantic compression mapping function , represents the power-normalized semantic symbol signal, The sequence number representing the semantic symbol signal, L represents the length of the semantic data, The average number of semantic symbol signals after encoding each data segment; in addition, Will be mapped to a transmissible physical signal waveform by the semantic channel encoder and finally transmitted by the multifunctional semantic base station;

[0154] Step 2.2: Model the signal received by the single-antenna communication user during the communication process, the signal-to-interference-and-noise ratio, and the bit signal decoding signal-to-noise ratio.

[0155] The multifunctional semantic base station broadcasts the downlink semantic symbol signal and the perception signal to all single-antenna communication users. The joint transmission signal of the multifunctional semantic base station can be represented as Wherein represents the first Joint transmission signal, represents the perception signal in the first Joint transmission signal, and are the beamforming vectors corresponding to communication and perception, respectively. The present invention sets the communication link and the perception link to be statistically independent and uncorrelated. Let , and represent the complex baseband equivalent channel from the multifunctional semantic base station to the first Single-antenna communication user, intelligent reflecting surface, and single-antenna potential eavesdropper, represents a set of complex numbers, represents the number of single-antenna communication users. Similarly, the complex baseband equivalent channels from the intelligent reflecting surface to the first Single-antenna communication user and single-antenna potential eavesdropper are and . The phase shift matrix of the intelligent reflecting surface is Wherein represents the RIS reflection coefficient vector, represents the base of the natural logarithm, represents the phase shift of the first Reflective element (RIS), and , represents the number of reflective elements, represents the imaginary unit, and Each element in satisfies the unit modulus constraint. The intelligent reflecting surface uses binary phase shift keying (BPSK) technology to modulate the bit signal generated by itself onto the joint transmission signal. Let represent the bit signal in the bit signal generated by the intelligent reflecting surface. The signal received by the first

[0156] (1);

[0157] wherein, denotes the signal received by the th single-antenna communication user, denotes the transpose of a matrix, denotes the additive white Gaussian noise (AWGN) at the th single-antenna communication user, denotes the noise variance of the th single-antenna communication user;

[0158] The legitimate user, i.e., the single-antenna communication user, first decodes and infers the original semantic symbol signal , thus the signal-to-interference-and-noise ratio (SINR) of the th single-antenna communication user can be expressed as:

[0159] (2);

[0160] wherein, denotes the signal-to-interference-and-noise ratio of the th single-antenna communication user;

[0161] After fully inferring the semantic information, the legitimate user will perform serial interference cancellation (SIC) and then decode the bit signal generated by the intelligent reflecting surface. At the same time, the sensing signal will interfere with the decoding of the bit signal generated by the intelligent reflecting surface. Accordingly, the signal-to-noise ratio (SNR) of the bit signal decoding can be expressed as:

[0162] (3);

[0163] wherein, denotes the signal-to-noise ratio of the bit signal decoding, denotes the multiple of the period of the secondary symbol (i.e., the reflecting signal generated by the intelligent reflecting surface) compared to the period of the primary symbol (i.e., the joint transmitted signal of the multifunctional semantic base station), and is an integer and satisfies ;

[0164] Step 2.3: Model the signal received by the single-antenna potential eavesdropper and the signal-to-interference-and-noise ratio;

[0165] For the single-antenna potential eavesdropper, the signal received by the single-antenna potential eavesdropper can be expressed as:

[0166] (4);

[0167] wherein, denotes the signal received by the single-antenna potential eavesdropper, is the additive white Gaussian noise at the single-antenna potential eavesdropper, denotes the noise variance of the single-antenna potential eavesdropper receiver.

[0168] Unlike the legitimate single-antenna communication users, the single-antenna potential eavesdropper is unaware of the specific modulation strategy adopted by the smart reflecting surface, and thus it can only decode the semantic symbol signal through the direct link, that is, when the single-antenna potential eavesdropper attempts to intercept the semantic symbol signal transmitted by the multi-functional semantic base station, the reflected signal reflected by the smart reflecting surface path becomes unpredictable interference. It can be deduced that the corresponding signal-to-interference-and-noise ratio at the single-antenna potential eavesdropper is:

[0169] (5) ;

[0170] wherein, denotes the corresponding signal-to-interference-and-noise ratio at the single-antenna potential eavesdropper;

[0171] Step 2.4: Model the echo signal and sensing performance in the target sensing process;

[0172] In the communication-sensing semantic integrated system, the sensing signal is also used for radar target sensing. Therefore, the multi-functional semantic base station working in the full-duplex mode can obtain environmental sensing information by capturing the echo signal reflected by the single-antenna sensing target, which can be described as:

[0173] (6) ;

[0174] wherein, denotes the echo signal, represents the channel from the multi-functional semantic base station to the single-antenna sensing target, , and denote the antenna spacing, wavelength, and angle of arrival, respectively, denotes the complex baseband equivalent channel from the multi-functional semantic base station (SBS) to the single-antenna sensing target; is the additive white Gaussian noise at the semantic base station, denotes the noise variance of the single-antenna sensing target, denotes the identity matrix;

[0175] To enhance the reception strength of the echo signal, the semantic base station adopts a reception beamforming vector , and accordingly the received echo signal is:

[0176] (7) ;

[0177] wherein, represents a specific construction form of the channel matrix;

[0178] Since the detection probability of a single-antenna sensing target is directly related to its sensing signal-to-noise ratio (SNR), the sensing performance of the communication-sensing semantic integration system can be measured by the following formula:

[0179] (8);

[0180] wherein, represents the sensing performance of the communication-sensing semantic integration system;

[0181] Step 2.5: Model the semantic security rate;

[0182] As a measure of semantic transmission performance, semantic similarity is measured by the semantic loss between the original data and the reconstructed data. To quantify the transmission efficiency of the semantic symbol signal, the semantic rate (unit: suts / s) is defined as the amount of effective semantic information delivered per unit time, and its expression is:

[0183] (9);

[0184] wherein, represents the semantic rate, represents the allocated channel bandwidth, represents the semantic similarity function, represents the signal-to-noise ratio, represents a single-antenna communication user or a single-antenna potential eavesdropper, represents a single-antenna communication user, represents a single-antenna potential eavesdropper;

[0185] However, the semantic similarity function currently lacks a clear expression, so the present application introduces a data-driven regression function through a deep semantic communication tool, and for different values, based on the signal-to-noise ratio / signal-to-interference-and-noise ratio , the semantic similarity function is approximately modeled to provide a closed-form expression for theoretical research. Therefore, the semantic similarity function needs to be standardized as an “S - shape” function, with a value range between the minimum semantic similarity and the maximum semantic similarity , and the semantic similarity function used can be expressed as:

[0186] (10);

[0187] wherein, represents the analytical approximation of the semantic similarity function , , , With are approximation parameters, by setting these approximation parameters, the semantic similarity function in formula (10) can be extended and applied to various types of data sources such as text, voice, image and video. Thus, the semantic transmission rate that can be achieved at the first single-antenna communication user and the single-antenna potential eavesdropper is:

[0188] (11) ;

[0189] wherein, represents the semantic transmission rate that can be achieved at the first single-antenna communication user and the single-antenna potential eavesdropper; Based on the definition of physical layer security, the semantic security rate of the first single-antenna communication user (SCU) is:

[0190]

[0191] (12) ;

[0192] wherein, represents the semantic transmission rate that can be achieved at the single-antenna communication user, represents the semantic transmission rate that can be achieved at the single-antenna potential eavesdropper, represents the maximum value operation, that is,

[0193] Step 3: According to the modeling result, a semantic security rate optimization problem is constructed, and the beamforming vector corresponding to communication and perception, the RIS reflection coefficient vector, the receiving beamforming vector and the average number of semantic symbol signals after encoding each data segment are solved

[0194] Step 3.1: Construct a problem of maximizing the total semantic security rate of the communication-perception semantic integrated system;

[0195] In order to further improve the physical layer security of the semantic-bit coexistence transmission architecture based on RIS assisted symbiotic radio, the present application constructs a semantic security rate maximization problem. The problem is optimized while meeting the quality of service requirements of bit signal decoding and the target perception performance requirements. In view of the non-convexity of the constructed optimization problem, the present application designs an iterative algorithm based on block coordinate descent, sequential convex approximation and semi-definite relaxation technology for solving, which specifically includes the following steps:

[0196] In order to construct the semantic security rate optimization problem, the signal-to-interference-and-noise ratio of the single-antenna communication user is re-expressed as:

[0197] ​​​​​ (13);

[0198] wherein, is an equivalent reflection coefficient vector, is an equivalent channel matrix, in particular, the equivalent channel matrix is composed of two parts: the first row describes the reflection channel of the reflection unit after removing the phase term, and a diagonalization operation is applied (to ensure the correctness of the matrix operation); the N+1th row represents the direct link. The equivalent reflection coefficient vector is constructed by adding the N+1th element (with a value of 1) to the vector . When is multiplied by , the operation results of the first N rows correspond to the reflection link, and the operation result of the N+1th row corresponds to the direct link. Further define as the total noise power at the single-antenna communication user, and its expression is:

[0199] (14);

[0200] and the signal-to-noise ratio for decoding the bit signal can be expressed as:

[0201] (15);

[0202] wherein, is an equivalent channel matrix;

[0203] Similarly, the signal-to-noise ratio at the single-antenna potential eavesdropper can be written as:

[0204] (16);

[0205] wherein, is a complex baseband equivalent channel from the reflection unit to the single-antenna potential eavesdropper, , are equivalent channel matrices;

[0206] define as the total noise power at the single-antenna potential eavesdropper, and its expression is:

[0207] (17);

[0208] The optimization goal of the present application is to maximize the total semantic security rate of the communication-aware semantic integrated system. Substituting equation (10) into equation (12), the problem of maximizing the total semantic security rate of the communication-aware semantic integrated system can be obtained, as shown in equations (18)-(23):

[0209] (18);

[0210] (19);

[0211] (20);

[0212] (21);

[0213] (22);

[0214] (23);

[0215] where denotes the maximum transmit power, and are the quality of service thresholds for RIS bit transmission and target perception, respectively, denotes the maximum value of

[0216] In the above optimization problem (18), constraint (19) limits the total transmit power, (20) specifies the value range of RIS phase shift, (21) restricts the value interval of semantic parameters, (22) ensures the minimum signal-to-noise ratio requirement of target perception, and (23) guarantees the minimum signal-to-noise ratio requirement of RIS data decoding;

[0217] Step 3.2: Divide the problem of maximizing the total semantic security rate of the integrated communication and perception semantic system into four sub-problems, including the optimization problem of the beamforming vectors corresponding to communication and perception, the optimization problem of the RIS reflection coefficient vector, the optimization problem of the receive beamforming vector, and the optimization problem of the average number of semantic symbol signals after encoding each data segment;

[0218] However, the optimization problem (18) has the characteristics of high coupling and non-convexity of variables, and the optimal solution cannot be directly solved , and existing convex optimization methods are also difficult to directly apply. To solve this problem, the present application adopts an optimization architecture based on block coordinate descent (BCD), which divides the optimization variables into , , and four independent modules.

[0219] Without loss of generality, set the semantic parameters to satisfy , and , the objective function of the optimization problem (18) can be further represented as:

[0220] (24);

[0221] To convert all non-convexities into constraints, introduce auxiliary variables , and satisfy

[0222] (25);

[0223] (26);

[0224] By the above transformation, the optimization problem can be reformulated as

[0225] (27);

[0226] (19)-(23) and (25)(26)(28);

[0227] The construction procedure of the optimization problem of the communication and sensing corresponding beamforming vectors is as follows:

[0228] At this time, the optimization problem (27)-(28) for can be expressed as

[0229] (29);

[0230] (19), (22), (23) and (25)(26)(30);

[0231] To further deal with the non-convex constraints (25) and (26), define = , = Substitute equation (13) and equation (16) into and , and the following equation can be obtained:

[0232] (31);

[0233] (32);

[0234] wherein , ;

[0235] Further introduce auxiliary variables in equation (25) and (26), and satisfy the following conditions:

[0236] (33);

[0237] (34);

[0238] (35);

[0239] (36);

[0240] By the above transformation, the non-convex problems (29) and (30) can be converted into the following equivalent forms:

[0241] (37);

[0242] (19), (22), (23) and (33) (34) (35) (36) (38);

[0243] (39);

[0244] (40);

[0245] For the non-convex constraints (22), (23), (39) and (40), we use the first-order Taylor expansion to convert them into convex forms, which can be expressed as:

[0246] (41);

[0247] (42);

[0248] (43);

[0249] (44);

[0250] where denotes taking the real part of a complex number, , and denotes the solution obtained in the th iteration; however, it should be noted that the constraints (33)-(36) still exhibit non-convexity.

[0251] To this end, we introduce auxiliary variables , which can be equivalently converted into:

[0252] (45);

[0253] (46);

[0254] (47);

[0255] (48);

[0256] Similarly, constraints (33) and (36) can be equivalently transformed as:

[0257] (49);

[0258] (50);

[0259] (51);

[0260] (52);

[0261] Among the above transformed constraints, (46), (47), (49) and (52) are still non-convex constraints. Specifically, by using the first-order Taylor expansion approximation, each concave constraint is replaced by its affine function, and the following constraints can be obtained:

[0262] (53);

[0263] (54);

[0264] (55);

[0265] (56);

[0266] where, , , , and are the solutions obtained in the th iteration, respectively;

[0267] Finally, the optimization problem of the communication and sensing corresponding beamforming vector can be expressed as:

[0268] (57);

[0269] (19), (41)-(44), (45), (48), (50), (51) and (53) -(56) (58);

[0270] The construction process of the optimization problem of the RIS reflection coefficient vector is as follows:

[0271] For the optimization problem (29) (30), the optimization problem of the RIS reflection coefficient vector can be expressed as:

[0272] (59);

[0273] (20), (23) and (25)(26)(60);

[0274] for handling the constraint of unit norm vector , the semi-definite relaxation method is adopted to update the variable . Introducing a new matrix variable , the unit norm constraint can be equivalently transformed into matrix form: 、 and , is a (M+1) x (M+1) identity matrix, where the main diagonal is all 1 and the rest is 0, and let . In the semi-definite relaxation step, the rank-one constraint is not considered for the time being, so that the problem is transformed into a semi-definite programming problem which can be solved efficiently. Given the variable , two sets of auxiliary variables and are introduced into the constraint conditions (25) and (26), so that:

[0275] (61);

[0276] (62);

[0277] Subsequently, can be transformed into:

[0278] (63);

[0279] (64);

[0280] (65);

[0281] (66);

[0282] where represents the sum of the main diagonal elements of the matrix, , . Similarly, can be transformed into:

[0283] (67);

[0284] (68);

[0285] (69);

[0286] where, , ;

[0287] The first-order Taylor expansion is used to handle the above non-convex constraints (61), (62) and (64), which are as follows:

[0288] (70);

[0289] (71);

[0290] (72);

[0291] where, , , , , and is the solution obtained in the th iteration;

[0292] For the constraint (23), the first-order Taylor expansion is also used to transform it into:

[0293] (73);

[0294] where, and are the solution obtained in the th iteration;

[0295] Finally, the optimization subproblem of the RIS reflection coefficient vector can be expressed as:

[0296] (74);

[0297] (20), (63), (65), (66), (67)-(69), (70)-(73) (75);

[0298] The construction process of the optimization problem of the receive beamforming vector is as follows:

[0299] When other variables are fixed, the optimization problems (18)-(23) can be simplified as the problem of maximizing the target perceived signal-to-noise ratio, and at this time the optimization problem of the receive beamforming vector can be transformed into a standard Rayleigh quotient problem:

[0300] (76);

[0301] The process for constructing the optimization problem of the average number of semantic symbol signals after encoding each data segment is as follows:

[0302] For semantic parameters (the average number of semantic symbol signals after encoding each data segment). The optimization problem can be simplified to:

[0303] (21)(77);

[0304] Since problem (77) contains only a single optimization variable, a function is constructed to represent it. Objective function:

[0305] (78);

[0306] Define the Sigmoid function , Representing the variable, substituting it into equation (78) yields: , Representing an integral object, according to the chain rule, ,therefore: By judgment first derivative We can analyze its monotonicity by taking the values ​​of . The first derivative is:

[0307] (79);

[0308] in, ;

[0309] Furthermore, we can obtain:

[0310] (80);

[0311] when hour, ,at this time derivative It is a negative number; when hour, ,at this time Therefore, the derivative It is still less than 0. Therefore, it can be seen that... Within the domain, Monotonically decreasing.

[0312] Step 3.3: Solve the optimization problems of beamforming vectors for communication and sensing, RIS reflection coefficient vectors, receiving beamforming vectors, and the average number of semantic symbol signals after encoding each data segment to obtain the optimal solution. ;

[0313] Step 3.3.1: Given the variable , the beamforming vector corresponding to communication and sensing is solved iteratively by CVX toolbox;

[0314] Step 3.3.2: The RIS reflection coefficient vector is solved by CVX toolbox;

[0315] Through the above analysis, problems (74) and (75) have been transformed into convex problems, which belong to the standard semi-definite programming problem and can be solved by CVX toolbox. However, it should be noted that the solution of this optimization problem has removed the rank-one constraint. By using the Gaussian randomization technique, the suboptimal solution that maximizes the objective function value of problems (74) and (75) can be obtained from the optimal solution , and then the suboptimal solution is projected onto the unit modulus constraint to obtain , where represents the M+1th component of the vector , and represents the phase angle of . It should be noted that the semi-definite relaxation step cannot guarantee that the objective function value is improved at each iteration, so in order to ensure that the objective function value does not decrease during the iteration process, a conservative update rule is used:

[0316] (81);

[0317] where represents the RIS reflection coefficient vector at the th iteration, represents the safety rate when the beamforming vector corresponding to communication and sensing is taken as the th iteration solution , and the RIS takes any optimized reflection coefficient vector ; represents the safety rate when the beamforming vector corresponding to communication and sensing is taken as the th iteration solution , and the RIS reflection coefficient vector is the th iteration solution ; that is, when the obtained by the new iteration is consistent with the of the previous iteration, the updated transmission strategy is still used;

[0318] Step 3.3.3: Solve the optimization problem of the receive beamforming vector;

[0319] Define , it is not difficult to find that the optimal solution of this problem is the matrix a main eigenvector of the Hessian matrix of the objective function, and the eigenvector is taken as the optimal solution of the perceptual beamforming vector;

[0320] Step 3.3.4: solving the optimization problem of the average number of semantic symbol signals after encoding each data segment; optimizing the objective function Regarding the variable is a monotonically decreasing function, and the optimal value is taken at the left end point of the interval, so the minimum value allowed for the variable can be set as .

[0321] Step 3.3.5: integrating the optimal solution of each sub-problem as the final optimal solution;

[0322] The algorithm decomposes the original problem into four sub-problems and gradually approaches the optimal solution through alternating optimization. In each iteration process, only one variable block is optimized, while the other variable blocks are fixed, and the selected variable block is solved to meet the numerical accuracy requirement. Therefore, the main computational complexity of the algorithm comes from the transmit beamforming design and RIS phase optimization modules based on the sequential convex approximation (SCA) method. The overall computational complexity is determined by the iterative solving process. In problem (57), the total number of variables is , so the computational complexity of solving the transmit beamforming sub-problem is , where is the number of iterations of this sub-problem, and is the accuracy of the sequential convex approximation method. The computational complexity of the RIS phase offset optimization can be represented as , where is the number of iterations of this sub-problem. The computational complexity of the optimal perceptual receive beamforming vector can be represented as . Therefore, the overall computational complexity of the joint optimization algorithm can be represented as: , where represents the total number of iterations of the outermost alternating optimization framework;

[0323] The performance of the proposed communication-perception-semantic integrated system is verified by simulation. A large number of experiments first verify the convergence of the designed block coordinate descent algorithm. The results show that the intelligent reflecting surface assisted symbiotic radio communication-perception-semantic integrated framework outperforms other benchmark schemes in terms of semantic security rate performance. In addition, the influence of key system parameters such as the number of intelligent reflecting surface reflecting units, the maximum power budget of the semantic base station (SBS), the number of semantic base station antennas, and the minimum quality of service requirement for bit signal decoding is also discussed.

[0324] The performance of the proposed algorithm in the integrated sensing and semantic communication system assisted by RIS is evaluated through numerical simulation. Among them, the base station and RIS are respectively deployed at coordinates (0m, 0m) and (50m, 0m); the coordinates of 3 legitimate users and 1 eavesdropper are (47m, 17m), (47m, -17m), (47m, 17m) and (49m, 9m) respectively; the sensing target is located at -40° direction, and the distance from the base station to the target is 100m. In addition, the simulation parameters are set as follows, , N = 64, B = 5MHz, I = 10; L = 256; . The path loss model adopts two forms: one is , where is the path loss exponent, is the reference distance, is the transmission distance, and the path loss exponent of the direct link from the base station to the target is set to = 3; the other is , where is the path loss exponent, is the transmission distance, the path loss exponent of the RIS assisted link is set to = 2.2, and the path loss exponent of the direct link from the base station to the legitimate user and the eavesdropper is set to = 3.6. The small-scale fading obeys the Rayleigh distribution, the Rayleigh factor of the RIS related channel is 10, and the Rayleigh factor of the direct path is 0; the deterministic line-of-sight component is derived from the product of the direction vectors of the transmitting end and the receiving end, and the non-line-of-sight component obeys the Rayleigh fading.

[0325] To verify the reliability of the proposed scheme, 4 kinds of comparison schemes are set for experiment: Without RIS scheme: the integrated sensing and semantic communication system does not contain RIS assisted transmission, and the base station and the user only communicate through the direct link; Without Sensing scheme: only the communication signal is considered, and all the transmission power is allocated to the secure transmission; Without SR scheme: the RIS is not phase encrypted and modulated, and there is no cooperative interference of artificial noise; Without PLS scheme: the existence of the eavesdropper is ignored, and the optimization target is only to maximize the semantic rate of the user.

[0326] Figure 2 ​​​​​The curves of semantic security rate of the five schemes versus the number of base station antennas are presented. As can be seen from the figure, with the increase of the number of antennas, the semantic security rate gradually increases, which is mainly due to the stronger beamforming gain, so a larger antenna array can bring stable rate improvement; at the same time, all curves tend to be flat when the number of antennas reaches 8-10. When the number of antennas is in the range of 2-10, the semantic security rate of the proposed scheme is significantly higher than that of the schemes without RIS, without symbiotic radio and without physical layer security: when the number of antennas is 6, the semantic security rate gain of the proposed scheme relative to the above three comparison schemes is about 160.7%, 60.5% and 81.1% respectively, fully embodying the advantages of RIS, cooperative artificial noise and physical layer security technology. In addition, the semantic security rate of the non-aware scheme is the highest, because this scheme removes the dedicated sensing signal, and the legitimate user can obtain higher signal quality, but the performance gap between the proposed scheme and the non-aware scheme is only 7.9%.

[0327] Figure 3 The semantic security rate curves of different comparison schemes under different transmit powers are described. As can be seen from the figure, in the low power and medium power interval, the semantic security rate shows a stable upward trend; when the transmit power exceeds 30dBm, the curve appears a clear plateau. The main reason for this phenomenon is that stronger transmit power will increase the received signal strength, thereby improving the security rate, but when the power reaches a certain threshold, the rate improvement space gradually saturates. In the whole power range, the proposed joint design scheme can still achieve a higher semantic security rate, and the curve of the non-aware scheme is always in the highest position, which is consistent with the law of Figure 2 The figure in the figure is a local amplification of the curve of the scheme without physical layer security. As can be seen from the figure, although the semantic security rate of this scheme increases as the power increases from 10dBm to 40dBm, its growth rate is significantly slower than that of other schemes, indicating that optimizing the physical layer security is a key means to improve the semantic security rate.

[0328] Figure 4 The curves of the semantic security rate of each scheme versus the semantic parameter Z are presented. As can be seen from the figure, the semantic security rate of all schemes shows a monotonous downward trend as increases: when is small (about ), the rate decreases relatively steeply; as further increases, the downward trend gradually flattens out, and eventually tends to a lower stable value. The figure in the figure is an amplification of the interval where takes a larger value. As can be seen from the figure, the semantic security rate gain of each scheme in this interval is very limited. The reason for this phenomenon is that when Z increases, the signals received by the user and the eavesdropper will be further compressed, resulting in a decrease in semantic security rate; therefore, as increases, the five curves gradually converge, and the performance gap between the schemes gradually narrows.

[0329] The proposed scheme is compared with three other optimization strategies: Fixed RIS phase (Fixed RIS) scheme: the phase of RIS is not optimized; Fixed semantic parameter scheme: the semantic parameter Z = 3 and Z = 5 are set respectively, without adaptive adjustment; Random beamforming (Random Beamforming) scheme: the beamforming vector of the base station is not optimized.

[0330] Figure 5 The curves of semantic security rate of the five schemes with the number of base station antennas are shown. As can be seen from the figure, the semantic security rate of the random beamforming scheme is the lowest because the base station beamforming vector is not optimized; the curves of the schemes with fixed semantic parameters Z = 3 or Z = 5 are also lower than the proposed scheme because the semantic parameters are not adaptively optimized; all curves increase with the number of antennas and tend to be stable when the number of antennas reaches 8. When the number of base station antennas is 6, the semantic security rate gain of the proposed scheme relative to the fixed RIS phase scheme is about 20.7%, and the gains relative to the random beamforming, Z = 3 and Z = 3 schemes are as high as 103.8%, 197.7% and 394.8% respectively. This result shows that RIS phase optimization, adaptive semantic control and beamforming optimization are the keys to achieve higher semantic security rate.

[0331] Figure 6 The curves of semantic security rate of the five schemes with the maximum transmit power are shown. As can be seen from the figure, the semantic security rate of all schemes increases with the increase of the maximum transmit power , and the growth tends to be flat in the high power interval (30dBm-40dBm). The proposed scheme achieves the highest semantic security rate in the entire power range, which benefits from the joint optimization of beamforming and semantic parameters. Compared with the fixed RIS phase, fixed semantic parameter and random beamforming schemes, RIS phase optimization has the most significant effect on the improvement of semantic security rate; on this basis, the performance gain of semantic parameter optimization is slightly lower than that of base station beamforming optimization.

[0332] Figure 7 The curves of semantic security rate of the five schemes with the number of RIS units are shown. As can be seen from the figure, the semantic security rate of the proposed scheme is the highest; but the performance gain of all schemes slows down significantly when the number of RIS units exceeds 16. This phenomenon shows that the number of RIS units has limitations on performance improvement, and blindly increasing the number of units may cause waste of hardware resources.

[0333] Figure 8The curve of the normalized transmit beam pattern with the change of azimuth angle is shown, and the positions of the target, three communication users and RIS (0° direction) are marked by dashed lines. For the proposed scheme, the main lobe of the transmit beam points to the direction of the RIS, because when the direct link faces a large path loss, the RIS-assisted indirect link is more advantageous; the no-RIS scheme does not use RIS, so the beam will not point to the 0° direction, and the energy is mainly concentrated in the user direction; the no-awareness scheme does not use a dedicated awareness signal, so the beam pattern does not point to the target direction, and its gain to the target direction is significantly lower than that of the proposed scheme.

[0334] Based on the RIS-assisted symbiotic radio technology, the present application proposes a semantic-bit coexistence transmission architecture for guaranteeing the semantic data transmission security of the integrated sensing and semantic communication system. Specifically, the RIS modulates its bit signal onto the incident signal and reflects the superimposed signal to the eavesdropper and all users; since the eavesdropper cannot decode the semantic data from the non-line-of-sight transmission signal, its understanding ability of the original semantic information will be significantly reduced. Based on the proposed architecture, the present application constructs a semantic security rate maximization problem, which optimizes the transmit beamforming, RIS phase offset, sensing receive beamforming and semantic parameters jointly, while ensuring the quality of service requirements of target sensing and bit transmission. In view of the non-convexity of the semantic security rate optimization problem, a block coordinate descent algorithm based on sequential convex approximation and semi-definite relaxation technology is designed. The simulation results show that compared with the traditional integrated sensing and semantic communication system, the proposed scheme achieves significant gain in the semantic security rate, further verifying the practical feasibility of the RIS-assisted integrated sensing and semantic communication system in enhancing the physical layer security of the semantic layer.

Claims

1. An intelligent reflecting surface assisted symbiotic common sense semantic bit secure transmission method, characterized in that, The method comprises the following steps: constructing a communication-aware semantic integration system; modeling the communication-aware semantic integration system; constructing a semantic security rate optimization problem according to the modeling result, solving the semantic security rate optimization problem to obtain an optimal solution, including a beamforming vector corresponding to communication and sensing, an RIS reflection coefficient vector, a receiving beamforming vector, and an average number of semantic symbol signals after encoding each data segment.

2. The smart reflective surface assisted commensal syndetic semantic bit secure transmission method of claim 1, wherein, The communication-aware semantic integration system comprises one multifunctional semantic base station, K single-antenna communication users, one single-antenna potential eavesdropper, one single-antenna sensing target, and one intelligent reflecting surface with N reflecting units; The multi-functional semantic base station works in a full duplex mode, is equipped with a uniform linear array, wherein a uniform linear array composed of N antennas is used to broadcast a joint transmission signal composed of a semantic symbol signal and a perception signal, a uniform linear array composed of N antennas is used to receive echo signals reflected by a single-antenna perception target, and , N is the number of antennas; the semantic symbol signal is a signal carrying semantic information. The intelligent reflecting surface with N reflecting units adopts a reflection modulation strategy, that is, modulates the bit signals generated by itself onto the joint transmission signal to generate a reflection signal and reflect it, thereby realizing non-line-of-sight information transmission from the multifunctional semantic base station to the K single-antenna communication users and providing communication services for the K single-antenna communication users; The K single-antenna communication users receive the joint transmission signal sent by the multifunctional semantic base station and the reflection signal generated by the intelligent reflecting surface; The single-antenna potential eavesdropper eavesdrops by receiving the joint transmission signal sent by the multifunctional semantic base station and the reflection signal generated by the intelligent reflecting surface; The single-antenna sensing target reflects the joint transmission signal to generate a return signal and returns it to the multifunctional semantic base station.

3. The smart reflective surface assisted commensal metacognitive semantic bit secure transmission method of claim 2, wherein, The modeling of the communication-aware semantic integration system specifically comprises the following steps: A1: modeling the semantic symbol signal generated by the multifunctional semantic base station; The input raw text is first extracted by a semantic encoder to obtain a semantic representation, generating semantic data wherein represents the i-th data segment in the semantic data, is the number of data segments and , , is the total number of data segments at the multi-functional semantic base station; and the semantic data is compressed into a semantic symbol vector by a semantic compression mapping function , represents a semantic symbol signal whose power has been normalized, represents the sequence number of the semantic symbol signal, represents the length of the semantic data, is the average number of semantic symbol signals after encoding each data segment; is mapped into a transmittable physical signal waveform by a semantic channel encoder, and is finally transmitted by the multi-functional semantic base station; A2: modeling the signal received by the single-antenna communication user during communication, the signal-to-interference-and-noise ratio, and the signal-to-noise ratio of bit signal decoding; The multi-functional semantic base station broadcasts downlink semantic symbol signals and sensing signals to all single-antenna communication users, and the joint transmission signal of the multi-functional semantic base station is represented as wherein represents the first joint transmission signal, represents the sensing signal in the first joint transmission signal, and are beamforming vectors corresponding to communication and sensing, respectively; it is assumed that , and are complex baseband equivalent channels from the multi-functional semantic base station to the first single-antenna communication user, the intelligent reflecting surface, and the single-antenna potential eavesdropper, represents a complex number set, represents the number of single-antenna communication users; the complex baseband equivalent channels from the intelligent reflecting surface to the first single-antenna communication user and the single-antenna potential eavesdropper are and , respectively; the phase shift matrix of the intelligent reflecting surface is wherein represents an RIS reflection coefficient vector, represents the base number of a natural logarithm, represents the phase shift of the first reflection unit, and , represents the number of reflection units, represents an imaginary unit, and each element in satisfies a unit modulus constraint; the intelligent reflecting surface uses a binary phase shift keying technology to modulate the bit signal generated by itself onto the joint transmission signal, and it is assumed that represents a bit signal in the bit signal generated by the intelligent reflecting surface, and the signal received by the first user is （1）； in, Indicates the first The signal received by a single-antenna communication user To represent the transpose of a matrix, Representing the Additive white Gaussian noise at a single-antenna communication user. Indicates the first Noise variance of a single-antenna communication user; No. The signal-to-interference-plus-noise ratio (SINR) of a single-antenna communication user is expressed as: （2）； wherein, represents the signal-to-interference-and-noise ratio for the first single-antenna communication user; The signal-to-noise ratio of bit signal decoding is expressed as: （3）； wherein, represents a signal-to-noise ratio of bit signal decoding, represents a multiple of a period of a secondary symbol compared to a primary symbol period, and is an integer and satisfies ; A3: modeling the signal received by the single-antenna potential eavesdropper and the signal-to-interference-and-noise ratio; For the single-antenna potential eavesdropper, the signal received is expressed as: （4）； wherein, represents the signal received by a single-antenna potential eavesdropper, is additive white Gaussian noise at the single-antenna potential eavesdropper, represents the noise variance at the receiving end of the single-antenna potential eavesdropper; The corresponding signal-to-interference-and-noise ratio at the single-antenna potential eavesdropper is: （5）； wherein denotes the corresponding signal-to-interference-and-noise ratio at a single-antenna potential eavesdropper; A4: modeling the return signal during target sensing and the sensing performance; The return signal is described as: （6）； wherein denotes the echo signal, denotes the channel from the multi-functional semantic base station to the single-antenna sensing target, , denotes the channel from the multi-functional semantic base station to the single-antenna sensing target, denotes the inter-antenna distance, the wavelength and the angle of arrival, respectively, denotes the additive white Gaussian noise at the semantic base station, denotes the noise variance of the single-antenna sensing target, denotes denotes the identity matrix; Semantic base station employs receive beamforming vectors The received echo signal is given by （7）； wherein a particular form of construction of the representative channel matrix; The sensing performance of the communication-aware semantic integration system is measured by the following formula: （8）； wherein, represents the perception performance of the communication-aware semantic integration system; A5: modeling the semantic security rate; The semantic rate expression is: (9)； wherein denotes a semantic rate, denotes an allocated channel bandwidth, denotes a semantic similarity function, denotes a signal-to-noise ratio, denotes a single-antenna communication user or a single-antenna potential eavesdropper, denotes a single-antenna communication user, denotes a single-antenna potential eavesdropper; The semantic similarity function is expressed as: (10)； wherein, denotes a semantically similar function , , , and are approximation parameters; No. The semantic transmission rate between a single-antenna communication user and a single-antenna potential eavesdropper is: (11)； wherein, denotes the semantic transmission rate at the nth single-antenna communication user with a single-antenna potential eavesdropper; denotes the semantic transmission rate at the nth single-antenna communication user with a single-antenna potential eavesdropper; No. The semantic security rate for a single-antenna communication user is: (12)； wherein denotes the achievable semantic transmission rate at a single-antenna communication user, denotes the achievable semantic transmission rate at a single-antenna potential eavesdropper, denotes the max operation.

4. The smart reflective surface assisted commensal metacognitive semantic bit secure transmission method of claim 3, wherein, The semantic security rate optimization problem is: (18)； (19)； (20)； (21)； (22)； (23)； wherein, denotes the maximum transmit power, and are the RIS bit transmission and the target-aware quality of service threshold, respectively, denotes the maximum value of is the equivalent channel matrix, is the equivalent channel matrix, is the equivalent reflection coefficient vector; is the RIS phase offset.

5. The smart reflective surface assisted commensal transsensory semantic bit secure transmission method of claim 4, wherein, The specific process of solving the semantic security rate optimization problem is: B1: dividing the semantic security rate optimization problem into four sub-problems, including an optimization problem of a beamforming vector corresponding to communication and sensing, an optimization problem of an RIS reflection coefficient vector, an optimization problem of a receiving beamforming vector, and an optimization problem of an average number of semantic symbol signals after encoding each data segment; B2: solving an optimization problem of a communication and sensing corresponding beamforming vector, an optimization problem of an RIS reflection coefficient vector, an optimization problem of a receiving beamforming vector, and an optimization problem of an average number of semantic symbol signals after encoding each data segment, to obtain an optimal solution .

6. The smart reflective surface assisted commensal transsensory semantic bit secure transmission method of claim 5, wherein, The construction process of the optimization problem of the beamforming vector corresponding to communication and sensing is: Introducing auxiliary variables and satisfies: (25)； (26)； Introducing auxiliary variables and satisfies the following condition: (33)； (34)； (35)； (36)； Finally, the optimization problem of the beamforming vector corresponding to communication and sensing is expressed as: (57)； (19)； (41)； (42)； (43)； (44)； (45)； (48)； (50)； (51)； (53)； (54)； (55)； (56)； wherein , ; denotes taking the real part of a complex number, , and denotes the solution obtained in the th iteration; is an auxiliary variable, , , , and denotes the solution obtained in the th iteration; is the complex baseband equivalent channel from the reflection unit to a potential eavesdropper with a single antenna, , are equivalent channel matrices.

7. The smart reflective surface assisted commensal transsensory semantic bit secure transmission method of claim 6, wherein, The construction process of the optimization problem of the RIS reflection coefficient vector is: Introduce new matrix variables , , , , is the identity matrix of size (M+1) x (M+1), set ; introduce two sets of auxiliary variables and ; Finally, the optimization problem of the RIS reflection coefficient vector is expressed as: (74)； (20)； (63)； (65)； (66)； (67)； (68)； (69)； (70)； (71)； (72)； (73)； wherein denotes the sum of the main diagonal elements of a matrix, , , , , , , , , and is the solution obtained at the th iteration; and is the solution obtained at the th iteration.

8. The smart reflective surface assisted commensal transsensory semantic bit secure transmission method of claim 7, wherein, The construction of the optimization problem of the receiving beamforming vector is: (76)； The optimization problem of the average number of coded semantic symbol signals of each data segment is constructed as: (78)； Among them, the definition = , = .

9. The smart reflective surface assisted commensal transsensory semantic bit secure transmission method of claim 8, wherein, The optimization problem of the beamforming vector corresponding to the communication and perception, the optimization problem of the RIS reflection coefficient vector, the optimization problem of the receiving beamforming vector and the optimization problem of the average number of coded semantic symbol signals of each data segment are solved, specifically: B2.1: In the case of given variables , the beamforming vectors corresponding to communication and sensing are solved iteratively by CVX toolbox; B2.2: The optimization problem of the RIS reflection coefficient vector is solved through the CVX toolbox; Specifically, by employing Gaussian randomization techniques, the optimal solution can be obtained... We obtain a suboptimal solution that maximizes the RIS reflection coefficient vector. Then the suboptimal solution Projecting onto the unit modulus constraint, we obtain... ,in, Representing vectors The (M+1)th component, express The phase angle; A conservative update rule is adopted: (81)； wherein, denotes the RIS reflection coefficient vector at the i-th iteration, denotes the RIS reflection coefficient vector at the i-th iteration, denotes the safety rate when the beamforming vector corresponding to the communication and sensing is taken as the i-th iteration solution, denotes the safety rate when the RIS takes the arbitrary optimized reflection coefficient vector denotes the safety rate when the beamforming vector corresponding to the communication and sensing is taken as the i-th iteration solution, denotes the safety rate when the RIS takes the reflection coefficient vector denotes the safety rate when the beamforming vector corresponding to the communication and sensing is taken as the i-th iteration solution, denotes the safety rate when the RIS takes the reflection coefficient vector denotes the safety rate when the beamforming vector corresponding to the communication and sensing is taken as the i-th iteration solution, denotes the safety rate when the RIS takes the reflection coefficient vector denotes the safety rate when the beamforming vector corresponding to the communication and sensing is taken as the i-th iteration solution, B2.3: The optimization problem of the receiving beamforming vector is solved; Definition The optimal solution to this problem is the principal eigenvector of the matrix and use this principal eigenvector as the optimal solution to the perceptual beamforming vector; B2.4: The optimization problem of the average number of coded semantic symbol signals of each data segment is solved; Regarding the variable is a monotonically decreasing function, whose optimal value is taken at the left end point of the interval, so that is set to its allowed minimum value; B2.5: The optimal solution of each sub-problem is integrated as the final optimal solution.

Images