QoE (Quality of Experience) driven resource allocation method of RIS (Radio Information System)-assisted three-dimensional offshore communication system

Abstract

The invention discloses a QoE (Quality of Experience)-driven resource allocation method for an RIS (Radio Information System)-assisted three-dimensional offshore communication system, which is specifically implemented according to the following steps: step 1, constructing an RIS-assisted three-dimensional offshore communication system model, and deploying the RIS on the surface of an obstacle; 2, proposing a random channel model based on three-dimensional geometric characteristics for the complexity of an ocean communication environment; step 3, in order to measure the subjective experience of the ship user, constructing an MOS evaluation function as a QoE index; step 4, establishing a multi-user sum MOS maximization problem, and jointly optimizing the transmitting power, the spectrum multiplexing and the RIS reflection coefficient; step 5, decomposing the original optimization problem into three sub-problems by using a block coordinate descent method, and optimizing the transmitting power by using a fractional programming method and optimizing the RIS reflection coefficient by using a semi-definite relaxation method; and solving a convex optimization problem by using CVX to obtain an approximate optimal resource allocation strategy. According to the method, the experience quality of a plurality of ship users can be maximized, and an approximate optimal resource allocation strategy is obtained.

Description

Technical Field

[0001] The present invention belongs to the field of wireless communication technology, and specifically relates to a QoE-driven resource allocation method for a RIS-assisted three-dimensional offshore communication system. Background Art

[0002] With the continuous development of the marine economy, the demand for low-latency and high-reliability marine communication has increased significantly. The growth of marine activities such as marine transportation, marine tourism, and offshore mineral exploration has posed new challenges and requirements for the development of marine communication. Constrained by environmental factors such as signal attenuation, multipath fading, and sparse infrastructure, existing marine communication technologies (such as satellite communication and communication in high-frequency (HF), very-high-frequency (VHF), and ultra-high-frequency (UHF) bands) have significant bottlenecks in terms of data transmission rate, latency, and reliability. To address these challenges, reconfigurable intelligent surface (RIS), as a cutting-edge technology in the sixth-generation (6G) network, shows broad application prospects. By dynamically manipulating electromagnetic waves, RIS can optimize signal transmission, expand the coverage area, and effectively suppress interference in the harsh offshore environment, providing a cost-effective alternative to traditional communication infrastructure.

[0003] In the 6G network, quality of experience (QoE) has become the core indicator for measuring users' satisfaction with communication services. Different from traditional quality of service (QoS) indicators that mainly focus on network performance parameters such as bandwidth, throughput, and latency, QoE emphasizes users' subjective perception of service quality. However, the improvement of data rate does not necessarily lead to better QoE, which indicates that an efficient QoE-based resource allocation strategy is crucial for improving users' satisfaction in the case of limited spectrum resources. Summary of the Invention

[0004] The purpose of the present invention is to provide a QoE-driven resource allocation method for a RIS-assisted three-dimensional offshore communication system, which can maximize the quality of experience of multiple vessel users and obtain an approximately optimal resource allocation strategy.

[0005] The technical solution adopted by the present invention is that the QoE-driven resource allocation method for a RIS-assisted three-dimensional offshore communication system is specifically implemented according to the following steps: Step 1: Construct a RIS-assisted three-dimensional offshore communication system model and deploy the RIS on the surface of obstacles; Step 2: In view of the complexity of the marine communication environment, a random channel model based on three-dimensional geometric characteristics is proposed; Step 3: To measure the subjective experience of vessel users, a MOS evaluation function is constructed as a QoE metric; Step 4: Establish a multi-user sum MOS maximization problem, and jointly optimize the transmit power, spectrum reuse, and RIS reflection coefficient; Step 5: Using the block coordinate descent method, the original optimization problem is decomposed into three sub-problems, and the fractional programming method is used to optimize the transmit power, the successive convex approximation method is used to optimize the spectrum reuse, and the semidefinite relaxation method is used to optimize the RIS reflection coefficient; CVX is used to solve the convex optimization problem to obtain an approximate optimal resource allocation strategy.

[0006] The features of the present invention also lie in: In Step 1, the system model includes a BS, a RIS, and an offshore vessel located on the coast. The direct link between the BS and the vessel may be blocked by building, hill, and tree obstacles. Therefore, a RIS is deployed on the obstacle to enhance the signal transmission between the BS and the vessel. The vessel is equipped with a single antenna, and the BS is equipped with M a uniform linear array of N element antennas, and the RIS is equipped with L an unmanned surface vessel that communicates with the BS through the RIS, and K vessel users adopting the device-to-device communication protocol.

[0007] In Step 2, the position vector from the origin to the midpoint of the ULA at the BS is expressed as , and the distance vector from the midpoint of the ULA at the BS to the m th transmitting antenna is expressed as (1) where , represents the antenna element spacing at the BS, and respectively represent the elevation angle and azimuth angle of the ULA. The distance vector from the initial element of the RIS to the n th reflecting element is expressed as (2) where , represents the spacing between the RIS reflecting elements, , the position vector from the origin to the initial element of the RIS is expressed as . Assuming that all the reflecting elements of the RIS array have the same size and are uniformly arranged, the position vector from the origin to the n th reflecting element of the RIS is expressed as (3) For a vessel, the position vectors from the origin to the l th USV, the k th pair of DVU transmitter and receiver are respectively denoted as , and , assuming that all vessels are moving towards the coast at a slow speed. At the beginning of each time slot, the vessels send their position information to the BS, and within the entire time slot, the channel state information of the vessels remains unchanged; To describe the scattering environment of the near - shore communication system, consider the P scattering clusters between the BS and the RIS, the l scattering clusters between the RIS and the U th USV, the k scattering clusters between the R th DVU transmitter and the RIS, and the k scattering clusters between the RIS and the W th DVU receiver. The position vectors of the p th ([[]] ) scatterer in the q th ([[]] ) cluster, the position vectors of the u th ([[]] ) scatterer in the v th ([[]] ) cluster, the position vectors of the r th ([[]] ) scatterer in the s th ([[]] ) cluster, and the position vectors of the w th ([[]] ) scatterer in the t th ([[]] ) cluster are respectively denoted as , , , .

[0008] The channel coefficient matrix between the BS and the RIS in step 2 is denoted as (4) where represents the path loss at the reference distance , is the distance between the RIS and the BS, is the path loss exponent from the RIS to the BS, is the complex matrix related to the small - scale fading, represents the mThe channel coefficient from the n th antenna of the BS to the (5) where represents the Rician factor of the RIS-BS link, represents the LoS component from the m th antenna of the BS to the n th reflecting element of the RIS, and its expression is (6) where is the carrier wavelength, represents the LoS link distance between the midpoint of the ULA at the BS and the n th reflecting element of the RIS, and its expression is (7) represents the unit direction vector from the midpoint of the ULA at the BS to the n th reflecting element of the RIS, and its expression is (8) where and represent the azimuth departure angle and elevation departure angle of the LoS link respectively, represents the NLoS component from the m th antenna of the BS to the n th reflecting element of the RIS, and its expression is (9) where is the phase caused by the scatterers, independent and uniformly distributed within , and represent the distances between the midpoint of the ULA at the BS and the initial element of the RIS and the p th scatterer in the q th cluster respectively, and represent the unit direction vectors from the midpoint of the ULA at the BS and the initial element of the RIS to the p th scatterer in the q th cluster; The channel coefficient from the l th USV to the RIS includes large-scale fading and small-scale fading, and can be expressed as (10) where represents the two-ray channel model related to large-scale fading, and its expression is (11) Among them, d is the distance between the transmitter and the receiver, and respectively represent the antenna heights of the transmitter and the receiver. The channel coefficient from the n th reflecting element of the RIS to the l th USV can be expressed as (12) Among them, is the Rician factor of the USV-RIS link, represents the LoS component from the n th reflecting element of the RIS to the l th USV, and its expression is (13) Among them, and respectively represent the LoS link distance and the unit direction vector from the l th USV to the initial element of the RIS, represents the NLoS component from the n th reflecting element of the RIS to the l th USV, and its expression is (14) Among them, and respectively represent the distances from the initial element of the RIS and the l th USV to the u th scattering point in the v th cluster, represents the unit direction vector from the initial element of the RIS to the u th scattering point in the v th cluster; The channel coefficient from the k th DVU transmitter to the RIS and the channel coefficient from the RIS to the k th DVU receiver are respectively expressed as and , in addition, the channel coefficient from the k th DVU transmitter to the k th DVU receiver and the interference channel coefficient from the l th USV to the k th DVU receiver are respectively expressed as and , which includes the double-ray model related to large-scale fading and the Rayleigh fading model related to small-scale fading.

[0009] In step 2, the phase shift vector of the RIS is expressed as , and the reflection coefficient matrix of the RIS is defined as , where and represent the phase shift and amplitude reflection coefficient of the n -th RIS reflection element respectively. For simplicity of analysis, it is set that to maximize the received power. The signal received by the BS from the l -th USV can be expressed as (15) where and represent the transmission powers of the l -th USV and the k -th DVU transmitter respectively. The transmission information symbols of the l -th USV and the k -th DVU transmitter are represented as and respectively, satisfying . In addition, is additive white Gaussian noise, following a complex Gaussian distribution with a mean of 0 and a variance of , that is . The binary variable represents the spectrum reuse metric. When the k -th DVU reuses the spectrum resources of the l -th USV, , otherwise . The received SINR of the l -th USV is expressed as (16) For the k -th DVU receiver, its received signal can be expressed as (17) where . Therefore, the received SINR of the k -th DVU is expressed as (18) Therefore, the achievable rate of the k -th DVU is expressed as (19).

[0010] In step 3, to measure the subjective experience of the DVU, a MOS evaluation function is constructed as a QoE metric to quantify user satisfaction. The MOS value range is , reflecting the QoE level from poor to excellent. When the MOS value reaches 4.5, continuously allocating resources will not improve the user QoE. The k The MOS of the DVU is expressed as (20) where a and b are constants, and their values are determined by the achievable rate and at a given time. and represent the weighted coefficients of the outage probability and the achievable rate respectively, satisfying , is the maximum achievable rate, represents the k th outage probability of the DVU link assisted by RIS, which is used to characterize the transmission reliability and is defined as (21) where represents the k th minimum SINR threshold of the DVU link. The k th outage probability of the DVU link can be approximately expressed as (22) where and represent the and expected values respectively.

[0011] In step 4, by jointly optimizing the transmit power , spectrum reuse and the RIS reflection coefficient , a multi-user sum MOS maximization problem is established, subject to the minimum SINR requirement of the USV and the maximum MOS constraint of the DVU. Therefore, the optimization problem is expressed as (23) where represents the l th minimum SINR threshold of the USV, and represent the l th maximum transmit power of the USV and the k th maximum transmit power of the DVU respectively. The constraint limits that the spectrum of one USV can be shared with at most one pair of DVUs. The constraint limits that at most one pair of DVUs can reuse the spectrum of one USV. The constraint is the unit modulus constraint of the RIS reflection coefficient. The constraint It means that when the MOS value reaches 4.5, no resources will be allocated to the DVU.

[0012] In step 5, the block coordinate descent method is used to decompose the original optimization problem into three sub-problems: transmit power optimization, spectrum reuse optimization, and RIS reflection coefficient optimization. Then, the above three sub-problems are alternately optimized through a QoE-based resource allocation algorithm until the objective function converges. When spectrum reuse and the RIS reflection coefficient are fixed, the optimization sub-problem of the transmit power is expressed as (24) Since the objective function in (24) is a non-concave function, and the constraint conditions and are non-convex constraints, thus, the problem belongs to a non-concave maximization problem. To handle the fractional expression in the objective function , the fractional programming method is adopted. Therefore, by introducing an auxiliary variable , is rewritten as (25) Under the condition that and are fixed, the iterative update expression of the optimal is (26) Therefore, substituting (25) into (24), the objective function becomes a concave function. To handle the non-convexity of the constraint condition , it is converted to (27) In addition, applying the successive convex approximation method, at a specific local point in each iteration, the original function is approximated by a more tractable analytical function. To solve the non-convex constraint in , it is rewritten as the difference between two concave functions. At the specific local points and , a first-order Taylor expansion of is performed to obtain its upper bound, and its expression is (28) where, (29) is a constant term related to the derivative. To handle the non-convex constraint in , at specific local points and at, by using its first-order Taylor expansion, we can obtain the upper bound of, and its expression is (30) where is a constant term related to the derivative. Substituting (28) and (30) into the constraint , thus obtaining (31) This constraint is a convex constraint; Since the objective function is a concave function and all constraints are linear or convex constraints, therefore, this optimization problem is transformed into a convex optimization problem and can be solved by a standard convex optimization solver.

[0013] The beneficial effects of the present invention are: The QoE-driven resource allocation method for the RIS-assisted three-dimensional offshore communication system of the present invention jointly optimizes the transmit power, spectrum reuse, and RIS reflection coefficient to maximize the multi-user sum MOS. To solve this complex non-convex optimization problem, the present invention adopts the fractional programming, successive convex approximation, and semidefinite relaxation methods to alternately optimize the transmit power, spectrum reuse, and RIS reflection coefficient. The simulation results show that compared with the random spectrum reuse, random RIS reflection coefficient, and the no-RIS scheme, the QoE-based resource allocation algorithm proposed by the present invention significantly improves the sum MOS and the achievable sum rate, and effectively reduces the outage probability. In addition, the simulation results verify the convergence of the proposed algorithm and it has a fast convergence speed. Description of the Drawings

[0014] Figure 1 is the flow chart of the present invention; Figure 2 is the scenario model diagram of the RIS-assisted offshore communication system of the present invention; Figure 3 is the three-dimensional channel model diagram of the RIS-assisted offshore communication system of the present invention; Figure 4 is the algorithm flow chart of the present invention. Detailed Embodiments

[0015] The present invention will be described in detail below in conjunction with the drawings and specific embodiments.

[0016] Embodiment 1 The QoE-driven resource allocation method for the RIS-assisted three-dimensional offshore communication system of the present invention, as Figure 1 shown, is specifically implemented according to the following steps: Step 1: Construct an RIS-assisted three-dimensional offshore communication system model, and deploy the RIS on the surface of obstacles to enhance the signal transmission between the BS and the vessels. Step 2: In view of the complexity of the marine communication environment, propose a stochastic channel model based on three-dimensional geometric characteristics, as Figure 3 shown, to characterize the wireless propagation characteristics of the offshore communication system. Step 3: To measure the subjective experience of vessel users, construct an MOS evaluation function as a QoE metric to quantify user satisfaction. Step 4: Establish a multi-user sum MOS maximization problem, and jointly optimize the transmit power, spectrum reuse, and RIS reflection coefficient. Step 5: Use the block coordinate descent method to decompose the original optimization problem into three sub-problems, and respectively use the fractional programming method to optimize the transmit power, the successive convex approximation method to optimize the spectrum reuse, and the semidefinite relaxation method to optimize the RIS reflection coefficient; use CVX to solve the convex optimization problem to obtain an approximate optimal resource allocation strategy.

[0017] Example 2 The QoE-driven resource allocation method for the RIS-assisted three-dimensional offshore communication system of the present invention, as Figure 2 shown, in Step 1, the system model includes a BS, an RIS, and offshore vessels located on the coast. The direct link between the BS and the vessels may be blocked by building, hill, and tree obstacles. Therefore, deploy the RIS on the obstacles to enhance the signal transmission between the BS and the vessels. The vessels are equipped with single antennas, and the BS is equipped with M a uniform linear array of N antennas, and the RIS is equipped with a uniform rectangular array composed of L passive reflection elements. The offshore communication system includes K unmanned surface vessels that communicate with the BS through the RIS, and

[0018] Example 3 The QoE-driven resource allocation method for the RIS-assisted three-dimensional offshore communication system of the present invention, in which in Step 2, the position vector from the origin to the midpoint of the ULA at the BS is expressed as , and the distance vector from the midpoint of the ULA at the BS to the m th transmitting antenna is expressed as (1) where , denotes the antenna element spacing at the BS, and respectively denote the elevation angle and azimuth angle of the ULA, and the distance vector from the initial element of the RIS to the n th reflection element is expressed as (2) Among them, , represents the spacing between RIS reflection elements, , and the position vector from the origin to the initial element of the RIS is expressed as . Assuming that all the reflection elements of the RIS array have the same size and are evenly arranged, the position vector from the origin to the n th reflection element of the RIS is expressed as (3) For a ship, the position vectors from the origin to the l st USV, the k th pair of DVU transmitter and receiver are respectively expressed as , and . Assuming that all ships are moving towards the coast at a slow speed, at the beginning of each time slot, the ship will send its position information to the BS, and within the entire time slot, the channel state information of the ship remains unchanged; To describe the scattering environment of the offshore communication system, consider the P scattering clusters between the BS and the RIS, the l scattering clusters between the RIS and the U th USV, the k scattering clusters between the R th DVU transmitter and the RIS, and the k scattering clusters between the RIS and the W th DVU receiver. The position vectors from the origin to the p th ( ) scattering point in the q th ( ) cluster, the position vector from the origin to the u th ( ) scattering point in the v th ( ) cluster, the position vector from the origin to the r th ( ) scattering point in the s th ( ) cluster, and the position vector from the origin to the w th ( ) scattering point in the t th ( ) cluster are respectively expressed as , , , .

[0019] The channel coefficient matrix between the BS and the RIS is expressed as (4) Among them, represents the path loss at the reference distance , is the distance between the RIS and the BS, is the path loss exponent from the RIS to the BS, is a complex matrix related to small-scale fading, represents the channel coefficient from the m th antenna of the BS to the n th reflecting element of the RIS, and its expression is (5) Among them, represents the Rician factor of the RIS-BS link, represents the LoS component from the m th antenna of the BS to the n th reflecting element of the RIS, and its expression is (6) Among them, is the carrier wavelength, represents the LoS link distance between the midpoint of the ULA at the BS and the n th reflecting element of the RIS, and its expression is (7) represents the unit direction vector from the midpoint of the ULA at the BS to the n th reflecting element of the RIS, and its expression is (8) Among them, and respectively represent the azimuth departure angle and elevation departure angle of the LoS link, and their expressions are respectively (9) (10) represents the NLoS component from the m th antenna of the BS to the n th reflecting element of the RIS, and its expression is (11) Among them, is the phase caused by the scatterers, independent and uniformly distributed within , and respectively represent the midpoint of the ULA at the BS and the initial element of the RIS and the p th within theq the distance between scatterers and respectively represent the unit direction vectors from the midpoint of the ULA at the BS and the initial element of the RIS to the p th q scatterer within the th l The channel coefficient from the (12) where represents the two-ray channel model related to large-scale fading, and its expression is (13) where d is the distance between the transmitter and the receiver, and respectively represent the antenna heights of the transmitter and the receiver. The channel coefficient from the n th l reflecting element of the RIS to the (14) where is the Rician factor of the USV-RIS link, represents the LoS component from the n th l reflecting element of the RIS to the (15) where and respectively represent the LoS link distance and the unit direction vector from the l th USV to the initial element of the RIS, n represents the NLoS component from the l th (16) where and respectively represent the distances from the initial element of the RIS and the l th u USV to the v th scatterer within the u th v cluster, from the kThe channel coefficients from the k th DVU transmitter to the RIS and from the RIS to the th DVU receiver are denoted as and k respectively. In addition, the channel coefficient from the k th DVU transmitter to the l th DVU receiver and the interference channel coefficient from the k th USV to the th DVU receiver are denoted as and

[0020] respectively, which include the two-ray model related to large-scale fading and the Rayleigh fading model related to small-scale fading. The phase shift vector of the RIS is denoted as , and the reflection coefficient matrix of the RIS is defined as where and n represent the phase shift and amplitude reflection coefficient of the th RIS reflection element respectively. For simplicity of analysis, l is set to maximize the received power. The signal received by the BS from the (17) where and represent the transmission powers of the l th USV and the k th DVU transmitter respectively. The transmission information symbols of the l th USV and the k th DVU transmitter are denoted as and respectively, satisfying . In addition, is additive white Gaussian noise, following a complex Gaussian distribution with a mean of 0 and a variance of , that is . The binary variable represents the spectrum reuse metric. When the k th pair of DVUs reuse the spectrum resources of the l th USV, , otherwise . The received SINR of the l th USV is denoted as (18) For the k th DVU receiver, its received signal can be expressed as (19) where , therefore, the received SINR of the DVU is expressed as k (20) k Therefore, the achievable rate of the DVU is expressed as (21) Example 4 For the QoE-driven resource allocation method of the RIS-assisted three-dimensional offshore communication system of the present invention, in step 3, in order to measure the subjective experience of the DVU, a MOS evaluation function is constructed as a QoE metric to quantify user satisfaction. The MOS value range is , reflecting the QoE level from poor to excellent. When the MOS value reaches 4.5, continuous resource allocation will not improve the user QoE. The k MOS of the DVU is expressed as (22) where a and b are constants, and their values are determined at the given achievable rate and and and k k k (23) where k (24) where is a step function. The smooth approximation function of the step function is where the non-negative smoothing parameter can be used to control the approximation error. Therefore, the k outage probability of the DVU link can be further rewritten as (25) Therefore, (23) is rewritten as (26) Due to the concavity of the step function and by using Jensen's inequality, it can be obtained that (27) where and . For , assume that , and are mutually independent. Since , and , then there is , where represents the Rayleigh fading component related to small-scale fading. Therefore, is re-expressed as (28) Similarly, is re-expressed as (29) Substituting (28) and (29) into (27), the approximate expression of the outage probability for the k th DVU link can be obtained as (30) where and respectively represent and the expected values of

[0021] Example 5 For the QoE-driven resource allocation method of the RIS-assisted three-dimensional offshore communication system of the present invention, in step 4, by jointly optimizing the transmit power , spectrum reuse and the RIS reflection coefficient , a multi-user sum MOS maximization problem is established and satisfies the minimum SINR requirement of the USV and the maximum MOS constraint of the DVU. Therefore, the optimization problem is expressed as

[0022] where represents the minimum SINR threshold of the l th USV, and respectively represent the l th USV and the kFor the maximum transmit power of the DVU, Constraint (31d) restricts that the spectrum of one USV can be shared with at most one pair of DVUs, Constraint (31e) restricts that one pair of DVUs can multiplex the spectrum of at most one USV, Constraint (31f) is the unit modulus constraint for the RIS reflection coefficient, and Constraint (31g) means that when the MOS value reaches 4.5, no resources are allocated to the DVU anymore.

[0023] Embodiment 6 The QoE-driven resource allocation method for the RIS-assisted three-dimensional offshore communication system of the present invention is as Figure 4 shown. In step 5, due to the high coupling between the optimization variables, as well as the binary variables in Constraints (31d) and (31e), the unit modulus constraint in Constraint (31f), and the non-convex constraints in Constraints (31a) and (31g), the problem belongs to a mixed-integer non-convex optimization problem. To efficiently solve this complex optimization problem, the block coordinate descent method is used to decompose the original optimization problem into three sub-problems: transmit power optimization, spectrum multiplexing optimization, and RIS reflection coefficient optimization. Then, a QoE-based resource allocation algorithm is proposed to alternately optimize the above three sub-problems until the objective function converges; When the spectrum multiplexing and the RIS reflection coefficient are fixed, the optimization sub-problem of the transmit power is expressed as

[0024] Since the objective function in (32) is a non-concave function and Constraints (32a) and (32d) are non-convex constraints, the problem belongs to a non-concave maximization problem. To handle the fractional expression in the objective function , the fractional programming method is adopted. Therefore, by introducing an auxiliary variable , is rewritten as (33) Under the condition that and are fixed, the iterative update expression of the optimal is (34) Therefore, substituting (33) into (32), the objective function becomes a concave function. To handle the non-convexity of Constraint (32a), it is converted into (35) In addition, by applying the sequential convex approximation method, at a specific local point in each iteration, the original function is approximated by a more tractable analytical function. To address the in the non-convex constraint (32d), it is rewritten as the difference between two concave functions. At the specific local points and , a first-order Taylor expansion of is performed to obtain its upper bound, and its expression is (36) where, (37) are constants related to the derivative. To handle the in the non-convex constraint (32d), at the specific local points and , its upper bound can be obtained by using its first-order Taylor expansion, and its expression is The upper bound of (38) where, are constants related to the derivative. Substituting (36) and (38) into the constraint condition (32d), we obtain (39) This constraint condition is a convex constraint; therefore, the problem can be rewritten as

[0025] Since the objective function is a concave function and all constraint conditions are linear or convex constraints, the problem is a convex optimization problem and can be solved by a standard convex optimization solver (such as CVX).

[0026] When the transmit power and the RIS reflection coefficient are fixed, the optimization sub-problem of spectrum reuse is expressed as

[0027] To simplify the complex binary variables in (41b) and (41c) in the problem , by relaxing the binary variables to continuous variables, an upper bound for the objective value of the problem is provided. Therefore, the problem can be rewritten as

[0028] However, due to the non-concavity of the objective function and the non-convexity of the constraint conditions (41a) and (41d), the problem remains a non-convex optimization problem. To handle the non-concave terms of the objective function in (42) , a lower bound is obtained at a specific local point through the first-order Taylor expansion of , and its expression is (43) where (44) are constant terms related to the derivative. Substituting (43) into (42), the objective function is thus reconstructed into a concave function form. To transform the non-convex constraint (41d) into a convex constraint, the first-order Taylor expansion is applied at a specific local point to obtain the upper bounds of and , and their upper bounds are respectively expressed as (45) (46) where (47) and are constant terms related to the derivative. Substituting (45) and (46) into the constraint condition (41d), a convex constraint is obtained, and its expression is (48) Therefore, the problem can be further rewritten as

[0029] In the problem , the objective function is concave, and all constraint conditions are convex. Therefore, the problem is a convex optimization problem and can be solved by a standard convex optimization solver (such as CVX).

[0030] When the transmit power and the spectrum reuse are fixed, the optimization sub-problem of the RIS reflection coefficient is expressed as

[0031] In solving the problem When, the main challenges include the non-concavity of the objective function, the non-convexity of the constraint conditions (50a) and (50c), and the unit modulus constraint in the constraint condition (50b). To effectively solve the non-concavity of the objective function, the semidefinite relaxation technique is adopted. Let , where . To facilitate the processing of in the objective function and , define and . Therefore, first rewrite as , where . Introduce two auxiliary variables, defined respectively as (51) (52) Further rewrite as (53) where, is a positive semidefinite matrix and satisfies . Similarly, we can get . Therefore, in (50) can be rewritten as (54) To handle the fraction in the objective function (50), the fractional programming method is adopted to rewrite as (55) Under the condition that is fixed, the iterative update expression of the optimal is (56) To solve the non-concave term in the objective function, define , , , and . Therefore, rewrite as (57) where, (58) . Substitute (55) and (57) into (50), and the objective function is transformed into a concave function. Define and To transform the non - convex constraint (50a) into a linear constraint, it is reformulated as (59) where (60) Then, by applying the first - order Taylor expansion at a specific local point the upper bounds of and are obtained, and their upper bounds are respectively expressed as (61) (62) where and are constants related to derivatives. Therefore, substituting (61) and (62) into (50c), the constraint (50c) can be transformed into (63) Therefore, the problem can be transformed into

[0032] The problem is a standard convex semidefinite programming problem and can be solved by a standard convex optimization solver (such as CVX). Usually, the optimal solution of the relaxed problem may not be a rank - 1 solution. If , then the optimal solution can be directly obtained by performing eigenvalue decomposition on . Otherwise, if , then a rank - 1 solution can be constructed by the Gaussian randomization method.

Claims

1. A QoE driven resource allocation method for a RIS-assisted three-dimensional offshore communication system, characterized in that: Follow the steps below to implement it: Step 1: construct a RIS-assisted three-dimensional offshore communication system model and deploy the RIS on the obstacle surface; Step 2: In view of the complexity of the marine communication environment, a random channel model based on three-dimensional geometric characteristics is proposed; Step 3: To measure the subjective experience of ship users, a MOS evaluation function is constructed as a QoE indicator; Step 4: Establish a multi-user total MOS maximization problem and jointly optimize the transmit power, spectrum reuse and RIS reflection coefficient; Step 5. Use the block coordinate descent method to decompose the original optimization problem into three sub-problems, and use the fractional programming method to optimize the transmission power, the continuous convex approximation method to optimize the spectrum reuse, and the semidefinite relaxation method to optimize the RIS reflection coefficient. Use CVX to solve the convex optimization problem and obtain the approximate optimal resource allocation strategy.

2. The QoE driven resource allocation method of the RIS-assisted three-dimensional offshore communication system according to claim 1, characterized in that: The system model in step 1 includes a BS, a RIS and an offshore ship located on the coast. The direct link between the BS and the ship may be blocked by obstacles such as buildings, hills and trees. Therefore, the RIS is deployed on the obstacles to enhance the signal transmission between the BS and the ship. The ship is equipped with a single antenna and the BS is equipped with M Uniform linear array of element antennas, RIS equipped N A uniform rectangular array of passive reflective elements. The offshore communication system includes L An unmanned surface vessel communicating with the BS via the RIS, and K For ship users adopting device-to-device communication protocol.

3. The QoE driven resource allocation method of the RIS-assisted three-dimensional offshore communication system according to claim 1, characterized in that: The position vector from the origin to the midpoint of the ULA at the BS in step 2 is expressed as , from the midpoint of ULA at BS to the m The distance vector of the transmitting antennas is expressed as (1) in, , represents the antenna unit spacing at the BS, and denote the elevation and azimuth of the ULA, respectively, from the initial RIS element to the n The distance vector of each reflective element is expressed as (2) in, , represents the spacing between RIS reflective elements, , the position vector from the origin to the initial element of the RIS is expressed as , assuming that all reflective elements of the RIS array have the same size and are evenly arranged, then the distance from the origin to the RIS n The position vector of the reflective element is expressed as (3) For a ship, from the origin to the l USV, k The position vectors of the DVU transmitter and receiver are expressed as , and ,Assuming that all ships move towards the coast at a slow speed, at the beginning of each time slot, the ship will send its position information to the BS, and the channel state information of the ship remains unchanged throughout the time slot; In order to describe the scattering environment of offshore communication systems, consider the P Scattering clusters, RIS and l Between USVs U Scattering clusters, k The connection between the DVU transmitter and the RIS R The scattering clusters and the RIS and k Between DVU receivers W scattering clusters, from the origin to the p indivual( ) within the cluster q indivual( ) The position vector of the scattering point, u indivual( ) within the cluster v indivual( ) The position vector of the scattering point, r indivual( ) within the cluster s indivual( ) The position vector of the scattering point, w indivual( ) within the cluster t indivual( ) The position vectors of the scattering points are expressed as , , , .

4. The QoE driven resource allocation method of the RIS-assisted three-dimensional offshore communication system according to claim 3, characterized in that: The channel coefficient matrix between BS and RIS in step 2 is expressed as: (4) in, Indicates the reference distance The path loss at is the distance between RIS and BS, is the path loss exponent from RIS to BS, is the complex matrix associated with small-scale fading, Indicates that from BS m Antenna to RIS n The channel coefficient of the reflective element is expressed as (5) in, represents the Rician factor of the RIS-BS link, Indicates that from BS m Antenna to RIS n The LoS component of the reflective element is expressed as (6) in, is the carrier wavelength, Indicates the ULA midpoint at BS and the RIS n The LoS link distance between the reflective elements is expressed as (7) Indicates the distance from the ULA midpoint at BS to the RIS n The unit direction vector of the reflective element is expressed as (8) in, and They represent the azimuth departure angle and elevation departure angle of the LoS link respectively, Indicates that from BS m Antenna to RIS n The NLoS component of a reflecting element is expressed as (9) in, is the phase caused by the scatterer, independent and in Evenly distributed inside, and Respectively represent the ULA midpoint at BS and the RIS initial element and the p In-cluster q The distance between the scattering points, and Respectively represent the distance from the ULA midpoint at BS and the RIS initial element to the p In-cluster q The unit direction vector of each scattering point; From l The channel coefficient from USV to RIS includes large-scale fading and small-scale fading, which can be expressed as (10) in, represents the two-ray channel model associated with large-scale fading, which is expressed as (11) in, d is the distance between the transmitter and the receiver, and represent the antenna heights of the transmitter and receiver respectively, from RIS n The reflective element to the l The channel coefficient of a USV can be expressed as (12) in, is the Rician factor of the USV-RIS link, Indicates that from RIS n The reflective element to the l The LoS component of a USV is expressed as (13) in, and Respectively represent from l The LoS link distance and unit direction vector from the USV to the RIS initial element, Indicates that from RIS n The reflective element to the l The NLoS component of a USV is expressed as (14) in, and Respectively represent the initial element and the first l USV to u In-cluster v The distance between the scattering points, Indicates the number of elements from the RIS initial element to the u In-cluster v The unit direction vector of each scattering point; From k The channel coefficients from the DVU transmitter to the RIS and from the RIS to the k The channel coefficients of the DVU receivers are expressed as and In addition, from k DVU transmitter to k The channel coefficients of the DVU receiver and the l USV to k The interference channel coefficients of the DVU receivers are expressed as and , which includes the two-ray model related to large-scale fading and the Rayleigh fading model related to small-scale fading.

5. The QoE driven resource allocation method of the RIS-assisted three-dimensional offshore communication system according to claim 2, characterized in that: In step 2, the phase shift vector of RIS is expressed as , the reflection coefficient matrix of RIS is defined as ,in, and Respectively represent n The phase shift and amplitude reflection coefficient of each RIS reflection element are set to simplify the analysis. To maximize the received power, the BS receives the l The signal of a USV can be expressed as (15) in, and Respectively represent l USV and k The transmission power of the DVU transmitter, l USV and k The transmission information symbols of the DVU transmitters are represented as and ,satisfy ,also, is additive white Gaussian noise with mean 0 and variance The complex Gaussian distribution of , binary variable Indicates the spectrum reuse index. k Multiplexing of DVU l When the spectrum resources of a USV are ,otherwise , No. l The receiving SINR of a USV is expressed as (16) For k The received signal of a DVU receiver can be expressed as (17) in, , therefore, k The received SINR for DVU is expressed as (18) Therefore, the k The achievable rate for DVU is expressed as (19)。 6. The QoE driven resource allocation method of the RIS-assisted three-dimensional offshore communication system according to claim 1, characterized in that: In step 3, in order to measure the subjective experience of DVU, a MOS evaluation function is constructed as a QoE indicator to quantify user satisfaction. The MOS value range is , reflecting the QoE level from poor to excellent. When the MOS value reaches 4.5, continuous resource allocation will not improve user QoE. k The MOS of DVU is expressed as (20) in, a and b is a constant whose value is determined by the given achievable rate and When determined, and Represent the weighted coefficients of the interruption probability and the achievable rate, respectively, satisfying , is the maximum achievable rate, Indicates k The interruption probability of the DVU link under the assistance of RIS is used to characterize the transmission reliability and is defined as (21) in, Indicates k The minimum SINR threshold for the DVU link, k The interruption probability of the DVU link can be approximately expressed as (22) in, and Respectively and expected value.

7. The QoE driven resource allocation method of the RIS-assisted three-dimensional offshore communication system according to claim 1, characterized in that: In step 4, the transmission power is jointly optimized , spectrum reuse and RIS reflection coefficient , establish the multi-user total MOS maximization problem, and meet the minimum SINR requirement of USV and the maximum MOS constraint of DVU, so the optimization problem is expressed as （23） in, Indicates l The minimum SINR threshold of a USV, and Respectively represent l USV and k Maximum transmit power for DVU, constraints The spectrum of a USV can be shared with at most one pair of DVUs. Limit a pair of DVUs to reuse the spectrum of one USV at most. is the unit modulus constraint of the RIS reflection coefficient, and the constraint condition This means that when the MOS value reaches 4.5, no more resources will be allocated to the DVU.

8. The QoE driven resource allocation method of the RIS-assisted three-dimensional offshore communication system according to claim 1, characterized in that: In step 5, the block coordinate descent method is used to decompose the original optimization problem into three sub-problems: transmit power optimization, spectrum reuse optimization, and RIS reflection coefficient optimization. Then, the three sub-problems are alternately optimized by the QoE-based resource allocation algorithm until the objective function converges. When spectrum reuse and RIS reflection coefficient When fixed, the transmission power The optimization subproblem is expressed as （24） Since the objective function in (24) is a non-concave function and the constraint and is a non-convex constraint, so the problem It is a non-concave maximization problem. In order to deal with the objective function The fractional expression of ,right Rewrite it as (25) exist and Under fixed conditions, the optimal The iterative update expression is (26) Therefore, substituting (25) into (24), the objective function becomes a concave function. In order to deal with the constraint The non-convexity of (27) In addition, a continuous convex approximation method is applied to approximate the original function with a more tractable analytical function at a specific local point in each iteration in order to solve the non-convex constraints. In , rewritten as the difference of two concave functions, at a specific local point and Place, right Perform a first-order Taylor expansion to obtain its upper bound, which is expressed as (28) in, (29) is a constant term related to the derivative, in order to deal with non-convex constraints In , at a specific local point and Using its first-order Taylor expansion, we can get The upper bound of is expressed as (30) in, is the constant term related to the derivative, substituting (28) and (30) into the constraints , thus obtaining (31) This constraint is a convex constraint; Since the objective function is a concave function and all constraints are linear or convex constraints, the optimization problem is transformed into a convex optimization problem and can be solved by a standard convex optimization solver.

Images