Intelligent reflecting surface assisted cooperative energy transfer and spectrum sharing method

Abstract

The application relates to a smart reflector assisted cooperative energy transmission and spectrum sharing method, which runs in a cognitive radio network. First, an AP obtains complete channel state information according to pilot signals transmitted by the AP, ST, SR and PU; then, the AP calculates a target rate and a target time allocation coefficient, constructs an optimization problem, and maximizes the achievable rate of a secondary system under the premise that the achievable rate of a primary system meets the target rate; the AP obtains an optimal phase angle in an energy collection stage, calculates optimal phase angles of the primary system and the secondary system in an information transmission stage, and calculates an optimal time allocation coefficient and; whether the achievable rate of the primary system meets the target rate is judged; if the target rate requirement is met, transmission is carried out according to the steps of the first case; otherwise, transmission is carried out according to the steps of the second case; through the fusion design of cooperative energy collection and dynamic spectrum sharing, efficient resource allocation is realized.

Description

Technical Field

[0001] This invention belongs to the field of wireless communication technology, specifically relating to the collaborative optimization technology of energy harvesting and spectrum resource utilization, and particularly to a cooperative wireless energy transmission and spectrum sharing system and optimization method based on intelligent reflective surface (IRS) assistance. Background Technology

[0002] The large-scale deployment of 5G and future 6G networks is driving wireless networks towards a deep integration of the physical and digital worlds. Disruptive applications such as Integrated Sensing and Communication (ISAC), Holographic Multiple-Input Multiple-Output (HMIMO), and Tactile Internet (TI) require both a continuous and stable power supply and more spectrum resources to meet the ever-increasing demands for wireless transmission. If the massive number of IoT devices in future ultra-dense networks rely on traditional battery power, it will result in unbearable maintenance costs and environmental burdens, limiting the scale of network deployment and long-term reliability. Wireless energy harvesting (EH) technology can provide near-permanent power to devices by capturing environmental radio frequency signals, but its efficiency is limited by long-distance path loss and low radio frequency power density. In terms of spectrum resources, the mid-to-low frequency bands below 6 GHz offer advantages such as wide coverage, strong penetration, and relatively low path loss. However, this frequency band has been largely allocated through licensing, and a significant amount of the allocated spectrum remains idle in both time and space. The transmission modes are also relatively fixed, resulting in low spectrum utilization efficiency and difficulty in meeting the access needs of a massive number of devices. To address this challenge, cognitive radio technology has emerged. In cognitive radio networks, primary users who have obtained spectrum licenses have priority access rights, while secondary users can access the spectrum in the following ways: first, through spectrum sensing, utilizing idle spectrum not occupied by primary users; second, sharing spectrum with primary users under strict control of transmit power and access parameters; and third, obtaining spectrum usage opportunities by assisting primary users in transmitting data. These mechanisms can maximize the throughput of secondary systems while ensuring the communication quality of the primary system, thus achieving a win-win situation for both the primary and secondary systems.

[0003] In recent years, Intelligent Reflectors (IRS) have provided a novel approach to solving this problem. An IRS consists of numerous low-cost, passive programmable reflective units. Its core principle lies in the independent, real-time adjustment of the phase shift of each reflective unit by a central controller. This allows for precise intelligent control of the propagation direction and phase of incident electromagnetic waves, thereby actively reconstructing the wireless channel with near-zero power consumption. In terms of energy harvesting, IRS can focus and reflect radio frequency signals transmitted by the base station or ambient radio frequency energy to energy harvesting equipment through coherent superposition, significantly improving the radio frequency power density at the receiver, overcoming long-distance path loss, and achieving efficient, directional wireless power transmission. Regarding spectrum sharing, IRS can enhance the primary user signal strength through beamforming, improve the perceived reliability of secondary user signals in spectrum-sharing scenarios, and create high-gain, low-interference directional links for secondary user access by forming directional beams. This enhances the security and speed of dynamic spectrum access under strict interference constraints. Addressing the dual challenges of scarce spectrum resources and limited device energy in wireless communication networks, existing technologies have not adequately studied the synergistic improvement of spectrum efficiency and energy efficiency, lacking a cross-time, primary-secondary system collaboration framework. They have failed to fully consider cooperative transmission schemes where idle secondary users assist the primary system in improving energy harvesting and information transmission efficiency. Although IRS technology has been introduced into communication systems to enhance channel quality and suppress co-channel interference, its application paradigm is typically limited to optimizing the performance of the energy harvesting and information transmission phases in isolation. It has failed to systematically construct an IRS-assisted primary-secondary system collaborative energy harvesting and information transmission system, creating system-level redundant resources such as spectrum and time, improving overall system efficiency and performance, and achieving dynamic reuse of licensed spectrum and refined allocation of time resources. Summary of the Invention

[0004] To address the shortcomings of existing technologies, this invention proposes a method for cooperative energy transfer and spectrum sharing assisted by an intelligent reflective surface.

[0005] This invention is applicable to communication scenarios with high demands for energy efficiency and spectrum utilization. Through the integrated design of collaborative energy harvesting and dynamic spectrum sharing, it achieves efficient resource allocation.

[0006] The technical solution of this invention is as follows: A method for cooperative power transfer and spectrum sharing assisted by a smart reflector is proposed, operating within a cognitive radio network. The cognitive radio network comprises a primary system and a secondary system. The primary system includes an access point (AP) and a primary user (PU), while the secondary system includes a secondary transmitter (ST) and a secondary receiver (SR). Data transmission time is divided into equal-length time blocks, each lasting... Seconds; in a time block The transmission process includes: First, the AP obtains complete channel state information based on the pilot signals transmitted by the AP, ST, SR, and PU, namely the channel vectors between AP-PU, ST-PU, ST-SR, AP-IRS, PU-IRS, ST-IRS, and SR-IRS; then, the AP calculates the target rate. and target time allocation coefficient The optimization problem is to maximize the achievable rate of the secondary system while ensuring that the achievable rate of the primary system meets the target rate; the AP obtains the optimal phase angle during the energy harvesting phase. And calculate the optimal phase angle for the information transmission stages of the main system and the secondary system. , and optimal time allocation coefficient and Determine whether the reachable rate of the main system meets the target rate: if it meets the target rate requirement, then proceed with the transmission according to the steps of the first case; otherwise, proceed with the transmission according to the steps of the second case. The first scenario includes the following: Energy harvesting phase; AP and ST send energy signals to PU, IRS configures phase angle It also reflects energy signals; the PU collects energy from the direct channels of the AP and ST and the IRS reflection channel for a duration of [duration missing]. ; During the main system information transmission phase, the PU uses the collected energy to transmit data to the AP via the uplink channel, while the IRS configures the phase angle. In addition to auxiliary reflected data signals, the AP receives and decodes PU data from the direct channel and the IRS reflected channel for a duration of [duration missing]. ; In the secondary system information transmission phase, ST transmits data to SR, while IRS configures the phase angle. In addition to assisting in the reflection of data signals, the SR receives and decodes the ST data from the direct channel and the IRS reflection channel, with a duration of [duration missing]. ; The second scenario includes the following: During the energy harvesting phase, the AP sends an energy signal to the PU via the downlink channel, and the PU directly harvests the energy for a duration of [duration missing]. ; During the main system information transmission phase, the PU uses the collected energy to transmit data to the AP via the uplink channel. The AP receives and decodes the data, and the duration is... .

[0007] According to a preferred embodiment of the present invention, a nonlinear energy harvesting model is employed: ; in, Represents an exponential function. This indicates the RF power received by the primary user PU. This indicates the maximum power of the main user's PU energy harvesting circuit when it is close to saturation. To set specific relevant constants for a specific circuit.

[0008] According to a preferred embodiment of the present invention, in a cognitive radio network, in a single time block Internally, the energy harvesting and information transmission process is divided into two stages: the first stage is the energy harvesting stage, and the second stage is the information transmission stage. The optimized optimal achievable rate is defined as the target rate, and its mathematical expression is: ; in, To understand the signal-to-noise ratio of radio networks, This represents the channel amplitude between the AP and PU, and its statistical characteristics follow a Nakagami-m distribution. This refers to the transmit power of the PU during the information transmission phase. This represents the power received by the PU during the energy harvesting phase. Regarding the time allocation coefficient The optimal value is obtained by traversal, and the specific process is as follows: Will The range of values Discretize by fixed step size, set traversal step size Initialize target rate and optimal time allocation coefficient ; right Execute sequentially: and Substituting into the target rate formula (2), we get Compare the current rate and ,like Then update , traversal iteration Obtain the final target rate .

[0009] According to a preferred embodiment of the present invention, the AP acquires complete channel state information, and the specific communication process is divided into three stages, the transmission process of each stage being as follows: Collaborative energy harvesting phase: The time allocation coefficient for this phase is... The energy collected by the PU comes not only from the AP, but also from the direct link of the ST, and from the reflected link between the AP and ST via the IRS. The received power of the PU is expressed as: ; in, , , , and These represent the channel vectors between AP-PU, ST-PU, AP-IRS, ST-IRS, and PU-IRS, respectively. Main system information transmission phase: The time allocation coefficient for this phase is... The PU utilizes all the energy acquired during the cooperative energy harvesting phase to transmit data to the AP via the uplink channel; the reachable information rate of the main system is: ; in, The optimal signal-to-noise ratio for AP to receive PU signals. Indicates the IRS number Optimal phase configuration of each component during the main system information transmission phase Indicates equipment and The phase angle of the channel between them. , indicating the first intelligent reflective surface Units, The power output of the PU during the main system's information transmission phase, utilizing the energy collected. Subsystem information transmission phase: ST transmits data to SR with the assistance of IRS. The achievable rate of the subsystem is: ; in, This indicates the optimal signal-to-noise ratio for SR reception. For the IRS Optimal phase configuration of each component during the subsystem information transmission phase; To compare the effectiveness of the optimization algorithm proposed in this invention, this invention proposes a random phase scheme. The achievable speed of the main system is... ,in The optimal signal-to-noise ratio for the AP to receive the PU. Let be the transmit power of PU, where:

[0010] ; in, For the energy harvesting phase in the first scenario, the IRS employs a randomized phase configuration. and Let the time allocation coefficients for a random allocation scheme satisfy the following conditions: .

[0011] According to a preferred embodiment of the present invention, when the information transmission rate of the main system reaches the target rate, the saved time resources will be allocated to the secondary system for information transmission. The optimization problem is expressed as: ; in, This represents the optimal signal-to-noise ratio for SR reception in the subsystem. The achievable speed of the main system For the target rate, Indicates the duration of the energy harvesting phase. Indicates the time of information transmission in the main system. Indicates the IRS phase angle during the energy harvesting phase. This represents the achievable rate threshold coefficient of the main system.

[0012] According to a preferred embodiment of the present invention, the optimal phase angle is obtained and configured by calculating the input channel state based on the TD3 algorithm; including: The interaction between the agent and its environment is described as a Markov decision process, defined as a quintuple. ,in and They represent time. A finite set of states and actions; Indicates the state Take action below Then transferred to The transition probability is expressed as... ; Indicates time Instant reward, A discount factor is used to determine the importance of future rewards; the agent's behavior is determined by the policy. Decision, it means in the state Select action The probability or deterministic mapping is divided into stochastic policies and deterministic policies; in time... The cumulative rewards obtained are The goal of an intelligent agent is to learn the optimal policy. Maximize the expected cumulative discount return; in the state Follow the strategy The expected cumulative reward function is defined as the state value function, i.e.: ; In state Follow the strategy Take action The expected cumulative reward is defined as the action value function, i.e.: ; The TD3 algorithm includes an online Actor network. Learning Deterministic Strategies Two independent Critic networks A target Actor network and two target Critic networks ; in time The environment provides the state. The intelligent agent takes action Earn rewards by interacting with the environment and the next state ; Transform tuples Stored in the playback buffer In the middle, it is used to replay sampling and update network parameters; The Actor network consists of an online Actor network and a target Actor network; the TD3 algorithm uses an approximation function from the online Actor network. This represents the strategy to be learned; it is used to determine the strategy based on the current state. Directly output a defined, continuous action. Interacting with the environment; during the exploration phase, additional noise is added to generate exploratory behavior, i.e. , The noise is Ornstein-Uhlenbeck; the update objective of the Actor network is to adjust the policy parameters. Maximize the Critic network The assessed expected cumulative reward: ; exist At time t, the update of the Actor network follows the deterministic policy gradient theorem as follows: ; in, Indicates the input status of the online Actor network. The output action after , This represents the action of the online Critic network in response to the output of the online Actor network. and state Output value estimate This indicates information about the parameters of the online Actor network. Gradient operator for partial derivatives Represents the Critic network Regarding the action Gradient operator for partial derivatives It is an experience replay buffer. Indicates the experience replay buffer Sampling One transformed tuple sample; Critic network is a value function An estimator for accurately assessing the state. Next action The expected long-term return that can be obtained; the Critic network consists of two online Critic networks and two objective Critic networks, with the minimum Q-value of the two objective Critic networks as the objective value. At time t, the target Q-value is calculated using the target Critic network, with the following formula: ; in, and These represent the experience replay buffer, respectively. Small batches of samples The corresponding immediate reward and next state, This represents the cumulative discount factor. This indicates taking the minimum of the two target Q values. Indicates the first The parameters of a Critic network, The target action obtained after policy smoothing regularization is shown below: ; in, It is OU truncation noise added to the target action. Indicates the noise variance. This represents a truncation function that limits the noise range to... between, Represents the parameters of the target Actor network; The loss function of the Critic network is: ; in, This represents the Q-value of the online Critic network. Indicates the first Q-estimates for each Critic network; The TD3 algorithm employs a delayed update strategy, where the Critic network updates its network every time it needs to perform a delayed update. Next, the Actor network updates its parameters once via gradient ascent: ; in, The learning rate of the Actor network is used to update the Critic network parameters by minimizing the loss function between the current Q-estimate and the target Q-value, i.e., by updating them through gradient descent. , is represented as: ; in, The learning rate is used for the Actor network; the target network, including the target Actor network and the target Critic network, updates its parameters using a soft update method, as shown below: ; ; in, This is a soft update coefficient; Based on the TD3 algorithm architecture, the optimization problem is given. The state, action, and reward functions are designed as follows: State space: in time The state contains the channel state information at that moment. and the previous step The channel state information is ,in, , , , and These represent the channel vectors between AP-PU, ST-PU, AP-IRS, ST-IRS, and PU-IRS, respectively; the channel state information input to the Actor network is represented as... , and These represent taking the real and imaginary parts respectively; in At any given time, the state is represented as The channel amplitude follows a Nakagami-m distribution and is generated in each communication scenario. Action space: in time The action taken is represented as , ,in Indicates the IRS number Optimized phase angle values ​​for each component For continuous actions; Reward function: in time The TD3 algorithm obtains rewards through interaction; the reward function is designed as follows: ; in, This is the scaling factor. For the main user at every time The received power obtained by using the phase angle randomly generated by the IRS is, i.e., Equation (6). TD3 algorithm in time The learned optimal phase is By adjusting the phase angle of each reflective element in the IRS, the received power of the PU can be maximized.

[0013] According to a preferred embodiment of the present invention, the input channel state is calculated based on the SDR algorithm to obtain the optimal phase angle and then configured; specifically as follows: phase angle optimization problem This can be summarized as a complex quadratic optimization problem with a unit modulus constraint; to maximize the received power, the received power... That is, expression (3) expands into a complex quadratic form: ; in, For the variable to be optimized, The constraint becomes ,and , , , The objective function is rewritten as: ; After expanding and combining like terms, we obtain the objective function: ; in , , Introducing augmented vectors: ; The objective function is then expressed as a quadratic form of the augmented vector: ; in Defined as: ; Therefore, the optimization problem Simplified to: ; First of all, Solution Perform eigenvalue decomposition. ,in It is the identity matrix. It is a real diagonal matrix; The suboptimal solution to the problem is ,in , It exhibits a circularly symmetric complex Gaussian distribution; The objective solution to the problem is generated independently through a Gaussian randomization process. Select the maximum value, that is: ; Therefore, the optimal phase angle of the IRS is expressed as: .

[0014] According to a preferred embodiment of the present invention, the main system uses the IRS phase angle to achieve optimized received power and erosion time allocation coefficients. and To obtain the optimal time allocation coefficients and determine whether the achievable rate of the main system meets the target rate requirement, the process includes: after obtaining the optimal phase angle through the SDR or TD3 algorithm, fixing the optimal phase angle configuration, and optimizing it through a traversal method. and Specifically, traversal , For all feasible combinations, calculate the information transmission rate of the main system under each combination, as follows: Will The range of values Discretize by fixed step size, set traversal step size , Initialize the optimal reachable rate and optimal time allocation coefficient , ; For each traversal And execute in sequence: and Substituting into the target rate formula (4), we get Compare the current rate With target rate ,like Then, the achievable rate of the subsystem is calculated according to equation (5). ,like ,but , , ;when and After the traversal was completed, there was no and The combination enables the main system to achieve a certain speed. Once the target rate is reached, the system will transmit at that rate.

[0015] A computer device includes a memory and a processor, the memory storing a computer program, and the processor executing the computer program to implement the steps of the above-described intelligent reflector-assisted cooperative energy transfer and spectrum sharing method.

[0016] A computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps of the above-described intelligent reflector-assisted cooperative energy transfer and spectrum sharing method.

[0017] The beneficial effects of this invention are as follows: 1. This invention introduces a mechanism for IRS-assisted energy harvesting, information transmission, and secondary system collaborative access to spectrum sharing in the primary system, effectively overcoming the limitations of energy, time, and spectrum resource isolation between primary and secondary systems in traditional cognitive radio networks. Specifically, the primary system employs an energy harvesting and information transmission protocol, leveraging the passive beamforming gain of the IRS and the assistance of the ST (Speed ​​Streaming Station) to integrate AP-PU and ST-PU direct links with AP-IRS-PU and ST-IRS-PU reflection links to enhance the energy harvesting process. Simultaneously, information transmission is optimized through PU-AP and PU-IRS-AP composite links. This invention provides a specific communication transmission scheme: Provided the primary system meets the target rate, the secondary system is allowed to access the primary system's licensed spectrum, utilizing remaining time slots to complete its own data transmission, achieving resource synergy and performance improvement between the primary and secondary systems. Conversely, when the primary system fails to meet the target rate requirement, the secondary system is prohibited from accessing the primary system's spectrum, and the system transmits at the original target rate. This achieves flexible allocation of energy, time, and spectrum resources between the primary and secondary systems while ensuring the communication quality of the primary system.

[0018] 2. This invention constructs an optimization problem for cooperative energy transfer and spectrum sharing assisted by an intelligent reflector. The objective is to maximize the achievable rate of the subsystem, with the constraint that the achievable rate of the main system meets the target rate requirement. An optimization problem and solution algorithm based on this system are designed. Addressing the non-convex optimization challenges involved in the system, this invention proposes an independent optimization algorithm framework. For the non-convex optimization problem of the IRS phase angle, this framework integrates a semi-definite relaxation algorithm and a dual-delay deep deterministic strategy gradient algorithm to achieve the globally optimal configuration of the IRS phase during the energy harvesting phase. For the time allocation coefficient variable, an ergodic method is used to solve for the optimal time slot allocation scheme. This algorithm framework not only solves the NP-hard problem of non-convex issues but also ensures the global optimality of resource allocation, thus providing an efficient and feasible technical path for optimizing key parameters of the IRS-assisted system.

[0019] 3. This invention fully reveals the impact of network parameters on the system performance. By constructing a system achievable rate model and conducting in-depth simulation analysis, it systematically reveals the coupling correlation between key variables such as access point and secondary transmitter transmit power, IRS spatial deployment location, primary system rate threshold coefficient, number of IRS reflective units, and Nakagami-m fading parameters and the primary and secondary systems in the system. It clarifies the characteristics of the constraint relationship between the above parameters and the primary system achievable rate and target rate, as well as their gain contribution to the secondary system achievable rate, providing theoretical support and performance reference for the practical engineering deployment and parameter configuration of IRS-assisted cognitive radio networks. Attached Figure Description

[0020] Figure 1 This is a schematic diagram of the architecture of a cognitive radio network. Figure 2 Flowchart of the collaborative energy transfer process assisted by intelligent reflective surfaces; Figure 3 A schematic diagram showing the locations of the primary and secondary transmitters, receivers, and IRS; Figure 4 PU received power A schematic diagram illustrating the convergence changes over time steps; Figure 5 For the achievable rate of the subsystem With AP transmission power A schematic diagram illustrating the changes; Figure 6 For the achievable rate of the subsystem With ST's transmission power A schematic diagram illustrating the changes; Figure 7 For the achievable rate of the subsystem With IRS horizontal position A schematic diagram illustrating the changes; Figure 8 For the achievable rate of the subsystem With the vertical position of the IRS A schematic diagram illustrating the changes; Figure 9 For the achievable rate of the subsystem With the distance between AP and PU A schematic diagram illustrating the changes; Figure 10 For the achievable rate of the subsystem With IRS channel parameters A schematic diagram illustrating the changes; Figure 11 For the achievable rate of the subsystem With the number of IRS A schematic diagram illustrating the changes; Figure 12For the achievable rate of the subsystem With threshold coefficient A diagram illustrating the changes. Detailed Implementation

[0021] The present invention will be further defined below with reference to the accompanying drawings and embodiments, but is not limited thereto.

[0022] Example 1 A method for cooperative power transfer and spectrum sharing assisted by a smart reflector is proposed, operating within a cognitive radio network. The cognitive radio network comprises a primary system and a secondary system. The primary system includes an access point (AP) and a primary user (PU), while the secondary system includes a secondary transmitter (ST) and a secondary receiver (SR). Data transmission time is divided into equal-length time blocks, each lasting... Seconds; First, consider a model with only the main system. The main user (PU) is powered by energy harvesting and uses time-division multiplexing to transmit information to the access point (AP). The durations of the energy harvesting phase and the information transmission phase are respectively... and By traversing time The optimal achievable rate of the system Defined as the target rate, the optimal time allocation coefficient is: In this invention's system, the primary user (PU) acquires energy through two direct links, AP-PU and ST-PU, and simultaneously utilizes the passive beamforming gain of the IRS to compensate for channel attenuation through two reflection links, AP-IRS-PU and ST-IRS-PU, thereby enhancing the energy harvesting effect. The duration of the energy harvesting phase is... Subsequently, the PU utilizes the collected energy to transmit the main system signal to the access point (AP) via the PU-AP direct link and the PU-IRS-AP reflective link. The duration of this information transmission phase is... By optimizing the phase angle of the IRS, the energy harvested by the PU during the energy harvesting phase is significantly improved, while the signal-to-noise ratio (SNR) during the information transmission phase is greatly improved. This allows the main system to appropriately reduce the time allocation coefficient while meeting the target rate requirements. and When the achievable rate of the primary system meets the target rate requirement, the secondary system is permitted to access the licensed spectrum of the primary system and utilize the remaining time with the assistance of the IRS. The primary system completes its own information transmission; if the achievable rate of the primary system does not reach the target rate threshold, the secondary system is prohibited from accessing the primary system's spectrum, and the primary system transmits at the target rate. A schematic diagram of the transmission scheme is shown below. Figure 2 As shown. In a time block The transmission process includes:

[0023] First, the AP obtains complete channel state information based on the pilot signals transmitted by the AP, ST, SR, and PU, namely the channel vectors between AP-PU, ST-PU, ST-SR, AP-IRS, PU-IRS, ST-IRS, and SR-IRS; then, the AP calculates the target rate. and target time allocation coefficient The optimization problem is to maximize the achievable rate of the subsystem while ensuring that the achievable rate of the main system meets the target rate. The AP obtains the optimal phase angle during the energy harvesting phase using the optimization algorithm proposed in this invention. And calculate the optimal phase angle for the information transmission stages of the main system and the secondary system. , and optimal time allocation coefficient and Determine whether the reachable rate of the main system meets the target rate: if it meets the target rate requirement, then proceed with the transmission according to the steps of the first case; otherwise, proceed with the transmission according to the steps of the second case. The first scenario includes the following: Energy harvesting phase; AP and ST send energy signals to PU, IRS configures phase angle It also reflects energy signals; the PU collects energy from the direct channels of the AP and ST and the IRS reflection channel for a duration of [duration missing]. ; During the main system information transmission phase, the PU uses the collected energy to transmit data to the AP via the uplink channel, while the IRS configures the phase angle. In addition to auxiliary reflected data signals, the AP receives and decodes PU data from the direct channel and the IRS reflected channel for a duration of [duration missing]. ; In the secondary system information transmission phase, ST transmits data to SR, while IRS configures the phase angle. In addition to assisting in the reflection of data signals, the SR receives and decodes the ST data from the direct channel and the IRS reflection channel, with a duration of [duration missing]. ; The second scenario includes the following: During the energy harvesting phase, the AP sends an energy signal to the PU via the downlink channel, and the PU directly harvests the energy for a duration of [duration missing]. ; During the main system information transmission phase, the PU uses the collected energy to transmit data to the AP via the uplink channel. The AP receives and decodes the data, and the duration is... .

[0024] The transmission powers of the access point and the secondary transmitter are respectively , The Nakagami-m parameter is The path loss index is The signal-to-noise ratio and noise power at the receiving end are Single time block It lasts for 1 second.

[0025] Consider as Figure 1 The cognitive radio network shown coexists with a primary system (AP-PU) and a secondary system (ST-SR) in a two-dimensional plane. The primary system holds licensed spectrum usage rights, while the secondary system establishes spectrum sharing by assisting the primary system in transmitting energy signals: the primary system allocates licensed spectrum resources and saved transmission time to the secondary system while ensuring its own transmission quality requirements. Smart reflectors are deployed throughout the transmission phase of both the primary and secondary systems to optimize energy harvesting and data transmission performance. This invention considers both large-scale path loss and small-scale power fading characteristics during signal transmission in wireless channels. Figure 3 A schematic diagram showing the locations of primary and secondary transmitters, receivers, and IRS; large-scale path loss modeling is as follows: ,in It is the path loss index. Indicates the transmitting end and receiving end The transmission distance between them. (Transmitter) and receiving end The magnitude of small-scale power fading between them is denoted as Its statistical properties follow a Nakagami-m distribution. The phase adjustment of the IRS is achieved through the phase matrix. Achieve, where the phase angle , These correspond to the adjustment phases of the main system energy harvesting phase, the main system information transmission phase, and the secondary system information transmission phase, respectively. This indicates that it is a complex number. The number of IRS reflector units. Each user node in the network is affected by additive white Gaussian noise (AWGN), which follows a zero-mean complex Gaussian distribution with a power spectral density of... To address the nonlinear characteristics of energy harvesting circuits in practical engineering scenarios, this invention employs a nonlinear energy harvesting model to process the energy signals collected by the subsystem, thereby improving the model's practical adaptability.

[0026] Example 2 The difference between the method for cooperative energy transfer and spectrum sharing assisted by a smart reflective surface as described in Example 1 and the method described in Example 1 is as follows: This invention establishes a model of a cooperative energy transfer and spectrum sharing system assisted by an intelligent reflector. Based on this model, an optimization problem is constructed, and a solution algorithm based on this system is designed. The objective of the optimization problem is to maximize the reachable rate of the subsystem, with the constraint that the reachable rate of the main system meets the target rate requirement. Based on the independence between the optimization variables, this invention designs two independent optimization algorithms: First, for the non-convex optimization problem of the IRS phase angle during the energy harvesting phase of the main system, the semidefinite relaxation (SDR) algorithm and the twin-delayed deep-deterministic policy gradient (TD3) algorithm in deep reinforcement learning (DRL) are used to obtain the optimal IRS phase configuration. Second, for the time allocation coefficient problem between the main and subsystems, an ergodic method is used to determine the optimal transmission time allocation coefficient. This invention proposes an algorithm combining SDR and the ergodic method, as well as an algorithm combining TD3 and the ergodic method, effectively achieving optimized solutions for system performance. By developing a cognitive radio network simulation program, the impact of key factors such as access point transmit power, secondary transmitter transmit power, horizontal and vertical location of the IRS, primary system threshold coefficient, number of smart reflectors, and Nakagami-m fading parameters on network performance was revealed. To demonstrate the advantages of this invention in network performance, three benchmark schemes were proposed. Numerical results show that, compared with the benchmark schemes, the proposed scheme and algorithm can significantly improve the achievable rate of the secondary system.

[0027] In traditional wireless energy harvesting system research, linear energy harvesting models are widely used to simplify analysis. This model assumes that the radio frequency to DC energy conversion efficiency is a constant, meaning that the harvested power and the received radio frequency power are always linearly related. However, this idealized model cannot accurately reflect the nonlinear characteristics present in actual circuits, such as the saturation effect and sensitivity threshold of the rectifier circuit. Therefore, to consider the characteristics of actual energy harvesting circuits, a nonlinear energy harvesting model is adopted:

[0028] ; in, Represents an exponential function. This indicates the RF power received by the primary user PU. This indicates the maximum power of the main user's PU energy harvesting circuit when it is close to saturation. Specific constants are set for each specific circuit. This nonlinear energy harvesting model is used for energy conversion at the PU end, converting the power received during the energy harvesting stage into a power value that better reflects the characteristics of the actual nonlinear circuit, and then using the harvested energy to transmit data to the AP.

[0029] In cognitive radio networks, in a single time block Internally, the energy harvesting and information transmission process is divided into two stages: the first stage is the energy harvesting stage, where the primary user (PU) receives radio frequency signals and completes energy harvesting through the downlink channel of the access point (AP), lasting for a duration of [duration missing]. The second stage is the information transmission stage. The PU uses the energy collected in the previous stage to transmit information to the AP via the uplink channel, lasting for [duration missing]. The optimized optimal achievable rate is defined as the target rate, and its mathematical expression is:

[0030] ; in, To understand the signal-to-noise ratio of radio networks, This represents the channel amplitude between the AP and PU, and its statistical characteristics follow a Nakagami-m distribution. This refers to the transmit power of the PU during the information transmission phase. This represents the power received by the PU during the energy harvesting phase. Regarding the time allocation coefficient The optimal value is obtained by traversal, and the specific process is as follows: Will The range of values Discretize by fixed step size, set traversal step size Initialize target rate and optimal time allocation coefficient ; right Execute sequentially: and Substituting into the target rate formula (2), we get Compare the current rate and ,like Then update , traversal iteration Obtain the final target rate .

[0031] Will This serves as a performance benchmark for other subsequent transmission schemes.

[0032] The AP obtains complete channel state information. The specific communication process is divided into three stages, and the transmission process of each stage is as follows: Collaborative energy harvesting phase: The time allocation coefficient for this phase is... The energy collected by the PU comes not only from the AP, but also from the direct link of the ST, and from the reflected link between the AP and ST via the IRS. The received power of the PU is expressed as: ; in, , , , and These represent the channel vectors between AP-PU, ST-PU, AP-IRS, ST-IRS, and PU-IRS, respectively. Main system information transmission phase: The time allocation coefficient for this phase is... The PU utilizes all the energy acquired during the cooperative energy harvesting phase to transmit data to the AP via the uplink channel; the reachable information rate of the main system is: ; in, The optimal signal-to-noise ratio for AP to receive PU signals. Indicates the IRS number Optimal phase configuration of each component during the main system information transmission phase Indicates equipment and The phase angle of the channel between them. , indicating the first intelligent reflective surface Units, The power output of the PU during the main system's information transmission phase, utilizing the energy collected. In the secondary system information transmission phase, the collaboration between ST and IRS effectively improves the energy harvesting and information transmission efficiency of the main system, enabling the main system to meet the performance requirements of the target system in a shorter time. Therefore, under the constraint of meeting the main system's communication quality requirements (i.e., achieving a rate no lower than the target rate), the saved time period... During this period, the primary system licenses spectrum to the secondary system. During this phase, the ST transmits data to the SR via the IRS, and the achievable rate of the secondary system is:

[0033] ; in, This indicates the optimal signal-to-noise ratio for SR reception. For the IRS Optimal phase configuration of each component during the subsystem information transmission phase; To evaluate the auxiliary effect of the intelligent reflector on the main system, two benchmark schemes were designed for comparative analysis in this experiment. Simultaneously, to compare the effectiveness of the optimization algorithm of this invention, a random-time, random-phase scheme was used to compare the performance advantages of the algorithm.

[0034] First benchmark scheme: The IRS is deployed during the energy harvesting phase, while it is not used during the information transmission phase of the main system. The achievable rate of this baseline model is expressed as... ,in This refers to the signal-to-noise ratio of the AP receiving the PU signal in this scheme. This refers to the transmit power of the PU during the main system information transmission phase. This refers to the received power of the PU during the energy harvesting stage in the present invention.

[0035] Second benchmark scheme: The IRS is deployed during the information transmission phase of the main system, while it is not used during the energy harvesting phase. The achievable rate of this baseline model is expressed as... ,in The optimal signal-to-noise ratio for AP to receive PU data. This refers to the transmit power of the PU during the main system information transmission phase. The power received by the PU without IRS assistance during the energy harvesting phase.

[0036] Random Time Random Phase (RTRP) scheme: To verify the algorithm's effectiveness, the IRS phase angle during the energy harvesting phase is randomly configured; considering the optimization of the time coefficient, a random allocation scheme is adopted for the time coefficient. The achievable rate of the main system is... ,in This refers to the emission power of the PU. The optimal signal-to-noise ratio of the AP receiving the PU in the scheme of this invention is:

[0037] ; This refers to the power collected by the PU during the energy harvesting phase using the IRS random phase scheme in the present invention. For the energy harvesting phase in the first scenario, the IRS employs a randomized phase configuration. and Let the time allocation coefficients for a random allocation scheme satisfy the following conditions: .

[0038] This invention, based on the target system, introduces ST (Spirit Stereo) to actively assist in sending energy signals to PU (Power Unit). Simultaneously, intelligent reflective surfaces are deployed to assist in the energy harvesting phase, information transmission phase, and information transmission phase of the main system. When the information transmission rate of the main system reaches the target rate, the saved time resources are allocated to the secondary system for information transmission. The optimization problem is expressed as:

[0039] ; in, This represents the optimal signal-to-noise ratio for SR reception in the subsystem. The achievable speed of the main system For the target rate, Indicates the duration of the energy harvesting phase. Indicates the time of information transmission in the main system. Indicates the IRS phase angle during the energy harvesting phase. This represents the reachable rate threshold coefficient of the main system; similarly, for the first baseline scheme, the IRS is deployed in the energy harvesting phase of the main system, while the information transmission phase of the main system does not use IRS assistance. The constraint (a) of problem P1 is expressed as follows: For the second baseline scheme, the IRS is deployed in the information transmission phase of the main system, while the energy harvesting phase of the main system does not use IRS assistance. The constraint (a) of problem P1 is expressed as follows: For a random-time, random-phase scheme, the main system randomly generates time allocation coefficients and IRS phase angles during each optimization. If the random parameter combination enables the main system to achieve the target transmission rate, i.e. When the primary system completes communication, it allows secondary systems to access its spectrum and use the remaining time slots for information transmission; otherwise, secondary systems will be prohibited from accessing the spectrum, and the primary system will still independently complete information transmission at the original target rate.

[0040] The optimal phase angle is obtained and configured based on the input channel state using the TD3 algorithm; this includes: Based on the independence between optimization variables, this invention performs separate optimization of the IRS phase angle and time allocation coefficients: For the non-convex optimization problem of the IRS phase angle in the energy harvesting stage, the non-convex problem is solved by two methods: the semidefinite relaxation (SDR) method and the dual-delay deep deterministic policy gradient (TD3) algorithm in deep reinforcement learning to obtain the optimal phase configuration; based on the optimized received power, the global optimal solution of the time allocation coefficients is determined by an ergonomic method, and then it is verified whether the proposed system meets the preset target rate constraint, and the performance gain of the subsystem is quantitatively analyzed.

[0041] The optimization objective is to maximize the achievable rate of the subsystem, with the constraint that the achievable rate of the primary system meets the target rate requirement. The target rate is the optimal achievable rate of a traditional energy harvesting and information transmission scheme that only includes the AP-PU link, i.e., the primary user uses a time-division multiplexing approach, harvesting energy only through the AP's downlink and transmitting data through the AP's uplink. The optimization variables include two core parameters: one is the phase angle coefficient of the intelligent reflector unit of the IRS during the energy harvesting phase. ,in The first is the number of IRS units; the second is the time allocation coefficient for the main system transmission phase. and .

[0042] The interaction between the agent and the environment is described as a Markov Decision Process (MDP), defined as a quintuple. ,in and They represent time. A finite set of states and actions; Indicates the state Take action below Then transferred to The transition probability is expressed as... ; Indicates time Instant reward, A discount factor is used to determine the importance of future rewards; the agent's behavior is determined by the policy. Decision, it means in the state Select action The probability or deterministic mapping is divided into stochastic policies and deterministic policies; in time... The cumulative rewards obtained are The goal of an intelligent agent is to learn the optimal policy. Maximize the expected cumulative discount return; in the state Follow the strategy The expected cumulative reward function is defined as the state value function, i.e.: ; In state Follow the strategy Take action The expected cumulative reward is defined as the action value function, i.e.: ; The TD3 algorithm effectively alleviates problems such as Q-value overestimation and large training fluctuations by introducing techniques such as a dual Q-network, objective policy smoothing regularization, and delayed policy updates. The TD3 algorithm includes an online Actor network. Learning Deterministic Strategies Two independent Critic networks Provides value estimation for a target Actor network. and two target Critic networks ; in time The environment provides the state. The intelligent agent takes action Earn rewards by interacting with the environment and the next state ; Transform tuples Stored in the playback buffer In the middle, it is used to replay sampling and update network parameters;

[0043] The Actor network consists of an online Actor network and a target Actor network. The online Actor network interacts with the environment in real time, selecting actions based on the current policy. The target Actor network provides a stable action estimate for the next state, avoiding the training oscillations and instability issues of the online Actor network. The TD3 algorithm uses an approximation function from the online Actor network. This represents the strategy to be learned; it is used to determine the strategy based on the current state. Directly output a defined, continuous action. Interacting with the environment; during the exploration phase, additional noise is added to generate exploratory behavior, i.e. , For Ornstein-Uhlenbeck (OU) noise; the update objective of the Actor network is to adjust the policy parameters. Maximize the Critic network The assessed expected cumulative reward:

[0044] ; exist At time t, the update of the Actor network follows the deterministic policy gradient theorem as follows: ; in, Indicates the input status of the online Actor network. The output action after , This represents the action of the online Critic network in response to the output of the online Actor network. and state Output value estimate This indicates information about the parameters of the online Actor network. Gradient operator for partial derivatives Represents the Critic network Regarding the action Gradient operator for partial derivatives It is an experience replay buffer. Indicates the experience replay buffer Sampling One transformed tuple sample; Critic network is a value function An estimator for accurately assessing the state. Next action The expected long-term return that can be obtained; the TD3 algorithm adopts a dual-Q network architecture. The system also employs a target network mechanism, comprising two online Critic networks and two target Critic networks. The minimum Q-value of the two target Critic networks is used as the target value to mitigate the problem of overestimation of value. At time t, the target Q-value is calculated using the target Critic network, with the following formula:

[0045] ; in, and These represent the experience replay buffer, respectively. Small batches of samples The corresponding immediate reward and next state, This represents the cumulative discount factor. This indicates taking the minimum of the two target Q values. Indicates the first The parameters of a Critic network, The target action obtained after policy smoothing regularization is shown below: ; in, It is OU truncation noise added to the target action. Indicates the noise variance. This represents a truncation function that limits the noise range to... between, Represents the parameters of the target Actor network; The loss function of the Critic network is: ; in, This represents the Q-value of the online Critic network. Indicates the first Q-estimates for each Critic network; The TD3 algorithm employs a delayed update strategy, where the Critic network updates its network every time it needs to perform a delayed update. Next, the Actor network updates its parameters once via gradient ascent: ; in, The learning rate of the Actor network is used to update the Critic network parameters by minimizing the loss function between the current Q-estimate and the target Q-value, i.e., by updating them through gradient descent. , is represented as: ; in, The learning rate is used for the Actor network; the target network, including the target Actor network and the target Critic network, updates its parameters using a soft update method, as shown below: ; ; in, This is a soft update coefficient; Based on the TD3 algorithm architecture, the optimization problem is given. The state, action, and reward functions are designed as follows: State space: in time The state contains the channel state information at that moment. and the previous step The channel state information is ,in, , , , and These represent the channel vectors between AP-PU, ST-PU, AP-IRS, ST-IRS, and PU-IRS, respectively; the channel state information input to the Actor network is represented as... , and These represent taking the real and imaginary parts respectively; in At any given time, the state is represented as The channel amplitude follows a Nakagami-m distribution and is generated in each communication scenario. Action space: in time The action taken is represented as , ,in Indicates the IRS number Optimized phase angle values ​​for each component For continuous actions; Reward function: in time The TD3 algorithm obtains rewards through interaction. Since nonlinear energy harvesting is a monotonically increasing function, to correctly guide the agent's training process and alleviate the problem of unclear training objectives caused by the small improvement effect of the reward function, the reward function is designed as follows: ; in, This is the scaling factor. For the main user at every time The received power obtained by using the phase angle randomly generated by the IRS is, i.e., Equation (6). TD3 algorithm in time The learned optimal phase is By adjusting the phase angle of each reflective element in the IRS, the received power of the PU can be maximized.

[0046] The optimal phase angle is calculated and configured based on the input channel state using the SDR (semi-definite relaxation method); the details are as follows: phase angle optimization problem This can be summarized as a complex quadratic optimization problem with a unit modulus constraint; to maximize the received power, the received power... That is, expression (3) expands into a complex quadratic form: ; in, For the variable to be optimized, The constraint becomes ,and , , , The objective function is rewritten as: ; After expanding and combining like terms, we obtain the objective function: ; in , , Therefore, the problem is represented as a Quadratic Constrained Quadratic Programming (QCQP) problem. Since the unit modulus constraint is non-convex, it is an NP-hard problem. SDR transforms it into a convex optimization problem by using slack variables. Augmented vectors are introduced:

[0047] ; The objective function is then expressed as a quadratic form of the augmented vector: ; in Defined as: ; However, optimization problem It is an NP-hard problem, making ,satisfy and Due to the nonconvexity of the rank-1 constraint, a positive semidefinite relaxation is used to relax the rank-1 constraint. Therefore, the optimization problem... Simplified to: ; Due to optimization issues This is a standard convex semi-positive definite programming problem, and the optimal value can be obtained using existing convex optimization tools. However, the optimization problem... The obtained optimal solution may not satisfy the rank-1 constraint. First, for Solution Perform eigenvalue decomposition. ,in It is the identity matrix. It is a real diagonal matrix; The suboptimal solution to the problem is ,in , It is a circularly symmetric complex Gaussian (CSCG) distribution. The objective solution to the problem is generated independently through a Gaussian randomization process. Select the maximum value, that is:

[0048] ; Therefore, the optimal phase angle of the IRS is expressed as: .

[0049] The main system uses the IRS phase angle to achieve optimized received power and erosion time allocation coefficients. and To obtain the optimal time allocation coefficients and determine whether the achievable rate of the main system meets the target rate requirement, the process includes: after obtaining the optimal phase angle through the SDR or TD3 algorithm, fixing the optimal phase angle configuration, and optimizing it through a traversal method. and Specifically, traversal , For all feasible combinations, calculate the information transmission rate of the main system under each combination, as follows: Will The range of values Discretize by fixed step size, set traversal step size , Initialize the optimal reachable rate and optimal time allocation coefficient , ; For each traversal And execute in sequence: and Substituting into the target rate formula (4), we get Compare the current rate With target rate ,like Then, the achievable rate of the subsystem is calculated according to equation (5). ,like ,but , , ;when and After the traversal was completed, there was no and The combination enables the main system to achieve a certain speed. Once the target rate is reached, the system will transmit at that rate.

[0050] The TD3-based IRS phase optimization algorithm learns the optimal IRS phase strategy for different channel scenarios through interactive training and parameter updates of the policy network, dual-review network, and target network. In its specific implementation, the policy network is first initialized. and two comment networks , And construct corresponding target networks for these networks. , and During initialization, the target network parameters are directly aligned with the original network parameters, and an experience replay pool is created simultaneously. Used to store training samples; then it enters a multi-scenario loop, where for each different channel scenario, real-time channel state information is first obtained through channel estimation techniques. The reinforcement learning environment is randomly reset and the current channel state information is integrated to obtain the initial state. Then it enters the iterative training phase; in each iteration, the policy network adjusts its training based on the current state. Output superimposed Gaussian noise action The reward fed back by the observation system after the phase action is performed. and the channel state at the next moment Transition tuple at this moment Store in the experience replay pool Then, random sampling is performed from the pool. A set of transformation tuple samples is used for network updates; first, the target policy network outputs the action for the next state. Add truncated noise to stabilize training, combined with a discount factor. Calculate the target value Then, the parameters of the two comment networks are updated by minimizing the mean squared error loss function; at each interval In the next iteration, the policy gradient is calculated based on the value output of the comment network. The policy network parameters are updated through gradient ascent, while the coefficients are updated softly. The target network parameters are updated slowly to ensure a stable training process; after all scenarios and iterations are completed, the phase output by the policy network during the energy harvesting phase is the optimal phase of the IRS. .

[0051] The SDR-based IRS phase optimization algorithm, based on semidefinite relaxation, is a model-driven algorithm that solves for the optimal IRS phase using convex optimization methods. Its core is to transform the non-convex phase constraint problem into a convex optimization problem. In its implementation, it first constructs the core optimization terms, including those derived from the channel combination vector. , Constructed quadratic form matrix From the system parameter vector , Constructed first-order term vector and the constant term consisting of vector norms. To transform the optimization objective into a tractable quadratic form, the dimension is... The original phase vector Expanded to And construct the augmented matrix At this point, the original optimization objective can be expressed as ; Due to the original problem The phase constraint is non-convex, and a matrix is ​​introduced through a semidefinite relaxation technique. The optimization objective is transformed into the form of matrix trace. At the same time, relax the constraints to and This transforms the problem into a solvable convex optimization problem; subsequently, convex optimization tools such as CVX are used to solve the relaxed problem, yielding the optimal semidefinite matrix. Eigenvalue decomposition The eigenvalue matrix is ​​obtained. Sample circularly symmetric complex Gaussian vectors and construct vectors Multiple independent generation processes are performed using Gaussian randomization. Select the maximum value to obtain the target solution Finally, the optimal phase vector of the IRS that satisfies the non-convex constraint is obtained. .

[0052] By relaxing the rank-1 constraint and using Gaussian randomization, the original nonconvex problem can be approximated with lower complexity. The global optimal solution.

[0053] This embodiment uses Python for simulation analysis. The number of iteration steps for each scenario in the TD3 algorithm is... Second, channel independent implementation Next, learning rate , buffer The size is Batch sampling size Discount coefficient Strategy delay update steps Target network soft update coefficient Unless otherwise specified, all system parameters are set to... , , , , W, , , , , , , , , .

[0054] Figure 4 To demonstrate the convergence characteristics of the subsystem's achievable rate with increasing time steps under different AP transmit powers, the TD3 algorithm gradually converges with increasing time steps. Comparing different optimization schemes for smart reflectors, the SDR and TD3 schemes significantly outperform other benchmark schemes.

[0055] Figure 5 The achievable rate of the subsystem varies with the transmit power of the access point. The overall performance of the system of this invention is significantly better than other benchmark systems. The achievable rates of the subsystems for all schemes are... All follow The increase is showing a decreasing trend. The reason is... Increasing the boost level increases the PU energy harvesting capacity and raises the target rate threshold, thereby increasing the main system time slot ratio and compressing the secondary system transmission time, leading to... decline.

[0056] Figure 6 The system performance of this invention is optimal as the achievable rate of the subsystem varies with the transmit power of the subtransmitter. (Subsystem achievable rate) Follow It increases as it grows. The reason is... This improvement simultaneously enhances the energy harvesting efficiency of the main system and the signal-to-noise ratio of the secondary system, saving main system transmission time and increasing secondary system transmission time while maintaining the target rate. Meanwhile, [the following text appears to be incomplete and requires further context: "by..."] Figure 7 It is evident that increasing the power and quantity of assistive devices will improve their assistive effect.

[0057] Figure 7 and Figure 8 The system's achievable rate under different AP and ST transmit powers varies with the horizontal and vertical positions of the smart reflector. The fading is greatest when the smart reflector is in the middle horizontal position, and less when it is close to the AP or PU. Furthermore, as the transmit power of the access point increases, leading to a higher target rate threshold, [further details are needed]. Improvements can compensate for performance losses caused by degradation.

[0058] Figure 9 The achievable speed of the subsystem varies with distance The changes indicate that the intelligent reflector link is significantly affected by dual large-scale fading, suffering severe losses at long distances, but has advantages in small-scale networks where the distance to the communication equipment is relatively short.

[0059] Figure 10 The achievable rate of the subsystem varies with channel parameters. The subsystem performance is... The channel performance is better than The channel data shows that the cooperation between ST and IRS significantly increases the gain of the differential channel.

[0060] Figure 11 The achievable rate of the subsystem varies with the number of IRS reflective units. The change in the number of intelligent reflective surfaces. The addition of beamforming enhances the beamforming capability of the IRS, improves the transmission efficiency of the main system, and frees up more resources for the secondary system.

[0061] Figure 12 The achievable rate of the subsystem varies with the threshold coefficient. The changes. Increasing the number of IRS units makes the main system rate constraints more stringent and requires more time resources, thereby compressing the transmission time slots of the secondary system. However, more IRS units can alleviate the rate attenuation.

[0062] Example 3 A computer device includes a memory and a processor. The memory stores a computer program, and the processor executes the computer program to implement the steps of the intelligent reflective surface-assisted cooperative energy transfer and spectrum sharing method described in Embodiment 1 or 2.

[0063] Example 4 A computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps of the intelligent reflective surface-assisted cooperative energy transfer and spectrum sharing method described in Embodiment 1 or 2.

Claims

1. A method for cooperative energy transfer and spectrum sharing assisted by a smart reflective surface, characterized in that, Operating within a cognitive radio network, which comprises a primary system and a secondary system; the primary system includes access points (APs) and primary users (PUs), while the secondary system includes secondary transmitters (STs) and secondary receivers (SRs). Data transmission time is divided into equal-length time blocks, each lasting... Seconds; in a time block The transmission process includes: First, the AP obtains complete channel state information based on the pilot signals transmitted by the AP, ST, SR, and PU, namely the channel vectors between AP-PU, ST-PU, ST-SR, AP-IRS, PU-IRS, ST-IRS, and SR-IRS; then, the AP calculates the target rate. and target time allocation coefficient The optimization problem is to maximize the achievable rate of the secondary system while ensuring that the achievable rate of the primary system meets the target rate; the AP obtains the optimal phase angle during the energy harvesting phase. And calculate the optimal phase angle for the information transmission stages of the main system and the secondary system. , and optimal time allocation coefficient and Determine whether the reachable rate of the main system meets the target rate: if it meets the target rate requirement, then proceed with the transmission according to the steps of the first case; otherwise, proceed with the transmission according to the steps of the second case. The first scenario includes the following: Energy harvesting phase; AP and ST send energy signals to PU, IRS configures phase angle It also reflects energy signals; the PU collects energy from the direct channels of the AP and ST and the IRS reflection channel for a duration of [duration missing]. ; During the main system information transmission phase, the PU uses the collected energy to transmit data to the AP via the uplink channel, while the IRS configures the phase angle. In addition to auxiliary reflected data signals, the AP receives and decodes PU data from the direct channel and the IRS reflected channel for a duration of [duration missing]. ; In the secondary system information transmission phase, ST transmits data to SR, while IRS configures the phase angle. In addition to assisting in the reflection of data signals, the SR receives and decodes the ST data from the direct channel and the IRS reflection channel, with a duration of [duration missing]. ; The second scenario includes the following: During the energy harvesting phase, the AP sends an energy signal to the PU via the downlink channel, and the PU directly harvests energy for a duration of [duration missing]. ; During the main system information transmission phase, the PU uses the collected energy to transmit data to the AP via the uplink channel. The AP receives and decodes the data, and the duration is... .

2. The method for cooperative energy transfer and spectrum sharing assisted by a smart reflective surface according to claim 1, characterized in that, A nonlinear energy harvesting model is adopted: ； in, Represents an exponential function. This indicates the RF power received by the primary user PU. This indicates the maximum power of the main user's PU energy harvesting circuit when it is close to saturation. To set specific relevant constants for a specific circuit.

3. The method for cooperative energy transfer and spectrum sharing assisted by a smart reflective surface according to claim 1, characterized in that, In cognitive radio networks, in a single time block Internally, the energy harvesting and information transmission process is divided into two stages: the first stage is the energy harvesting stage, and the second stage is the information transmission stage. The optimized optimal achievable rate is defined as the target rate, and its mathematical expression is: ； in, To understand the signal-to-noise ratio of radio networks, This represents the channel amplitude between the AP and PU, and its statistical characteristics follow a Nakagami-m distribution. This refers to the transmit power of the PU during the information transmission phase. This represents the power received by the PU during the energy harvesting phase. Regarding the time allocation coefficient The optimal value is obtained by traversal, and the specific process is as follows: Will The range of values Discretize by fixed step size, set traversal step size Initialize target rate and optimal time allocation coefficient ; right Execute sequentially: and Substituting into the target rate formula (2), we get Compare the current rate and ,like Then update , traversal iteration Obtain the final target rate .

4. The method for cooperative energy transfer and spectrum sharing assisted by a smart reflective surface according to claim 1, characterized in that, The AP obtains complete channel state information. The specific communication process is divided into three stages, and the transmission process of each stage is as follows: Collaborative energy harvesting phase: The time allocation coefficient for this phase is... The energy collected by the PU comes not only from the AP, but also from the direct link of the ST, and from the reflected link between the AP and ST via the IRS. The received power of the PU is expressed as: ； in, , , , and These represent the channel vectors between AP-PU, ST-PU, AP-IRS, ST-IRS, and PU-IRS, respectively. Main system information transmission phase: The time allocation coefficient for this phase is... The PU utilizes all the energy acquired during the cooperative energy harvesting phase to transmit data to the AP via the uplink channel; the reachable information rate of the main system is: ； in, The optimal signal-to-noise ratio for AP to receive PU signals. Indicates the IRS number Optimal phase configuration of each component during the main system information transmission phase Indicates device and The phase angle of the channel between them. , indicating the first intelligent reflective surface Units, The power output of the PU during the main system's information transmission phase, utilizing the energy collected. Subsystem information transmission phase: ST transmits data to SR with the assistance of IRS. The achievable rate of the subsystem is: ； in, This indicates the optimal signal-to-noise ratio for SR reception. For the IRS Optimal phase configuration of each component during the subsystem information transmission phase; The achievable speed of the main system is ,in The optimal signal-to-noise ratio for the AP to receive the PU. Let be the transmit power of PU, where: ； in, For the energy harvesting phase in the first scenario, the IRS employs a randomized phase configuration. and Let the time allocation coefficients for a random allocation scheme satisfy the following conditions: .

5. The method for cooperative energy transfer and spectrum sharing assisted by a smart reflective surface according to claim 1, characterized in that, When the information transmission rate of the main system reaches the target rate, the saved time resources will be allocated to the secondary system for information transmission. The optimization problem is expressed as: ； in, This represents the optimal signal-to-noise ratio for SR reception in the subsystem. The achievable speed of the main system For the target rate, Indicates the duration of the energy harvesting phase. Indicates the time of information transmission in the main system. Indicates the IRS phase angle during the energy harvesting phase. This represents the achievable rate threshold coefficient of the main system.

6. The method for cooperative energy transfer and spectrum sharing assisted by a smart reflective surface according to claim 1, characterized in that, The optimal phase angle is obtained and configured based on the input channel state using the TD3 algorithm; this includes: The interaction between the agent and its environment is described as a Markov decision process, defined as a quintuple. ,in and They represent time. A finite set of states and actions; Indicates the state Take action below Then transferred to The transition probability is expressed as... ; Indicates time Instant reward, A discount factor is used to determine the importance of future rewards; the agent's behavior is determined by the policy. Decision, it means in the state Select action The probability or deterministic mapping is divided into stochastic policies and deterministic policies; in time... The cumulative rewards obtained are The goal of an intelligent agent is to learn the optimal policy. Maximize the expected cumulative discount return; in the state Follow the strategy The expected cumulative reward function is defined as the state value function, i.e.: ； In state Follow the strategy Take action The expected cumulative reward is defined as the action value function, i.e.: ； The TD3 algorithm includes an online Actor network. Learning Deterministic Strategies Two independent Critic networks A target Actor network and two target Critic networks ; in time The environment provides the state. The intelligent agent takes action Earn rewards by interacting with the environment and the next state ; Transform tuples Stored in the playback buffer In the middle, it is used to replay sampling and update network parameters; The Actor network consists of an online Actor network and a target Actor network; the TD3 algorithm uses an approximation function from the online Actor network. This represents the strategy to be learned; it is used to determine the strategy based on the current state. Directly output a defined, continuous action. Interacting with the environment; during the exploration phase, additional noise is added to generate exploratory behavior, i.e. , The noise is Ornstein-Uhlenbeck; the update objective of the Actor network is to adjust the policy parameters. Maximize the Critic network The assessed expected cumulative reward: ； exist At time t, the update of the Actor network follows the deterministic policy gradient theorem as follows: ； in, Indicates the input status of the online Actor network. The output action after , This represents the action of the online Critic network in response to the output of the online Actor network. and state Output value estimate This indicates information about the parameters of the online Actor network. Gradient operator for partial derivatives Represents the Critic network Regarding the action Gradient operator for partial derivatives It is an experience replay buffer. Indicates the experience replay buffer Sampling One transformed tuple sample; Critic network is a value function An estimator for accurately assessing the state. Next action The expected long-term return that can be obtained; the Critic network consists of two online Critic networks and two objective Critic networks, with the minimum Q-value of the two objective Critic networks as the objective value. At time t, the target Q-value is calculated using the target Critic network, with the following formula: ； in, and These represent the experience replay buffer, respectively. Small batches of samples The corresponding immediate reward and next state, This represents the cumulative discount factor. This indicates taking the minimum of the two target Q values. Indicates the first Parameters of a Critic network The target action obtained after policy smoothing regularization is shown below: ； in, It is OU truncation noise added to the target action. Indicates the noise variance. This represents a truncation function that limits the noise range to... between, Represents the parameters of the target Actor network; The loss function of the Critic network is: ； in, This represents the Q-value of the online Critic network. Indicates the first Q-estimates for each Critic network; The TD3 algorithm employs a delayed update strategy, where the Critic network updates its network every time it needs to perform a delayed update. Next, the Actor network updates its parameters once via gradient ascent: ； in, The learning rate of the Actor network is used to update the Critic network parameters by minimizing the loss function between the current Q-estimate and the target Q-value, i.e., by updating them through gradient descent. , is represented as: ； in, The learning rate is used for the Actor network; the target network, including the target Actor network and the target Critic network, updates its parameters using a soft update method, as shown below: ； ； in, This is a soft update coefficient; Based on the TD3 algorithm architecture, the optimization problem is given. The state, action, and reward functions are designed as follows: State space: in time The state contains the channel state information at that moment. and the previous step The channel state information is ,in, , , , and These represent the channel vectors between AP-PU, ST-PU, AP-IRS, ST-IRS, and PU-IRS, respectively; the channel state information input to the Actor network is represented as... , and These represent taking the real and imaginary parts respectively; in At any given time, the state is represented as The channel amplitude follows a Nakagami-m distribution and is generated in each communication scenario. Action space: in time The action taken is represented as , ,in Indicates the IRS number Optimized phase angle values ​​for each component For continuous actions; Reward function: in time The TD3 algorithm obtains rewards through interaction; the reward function is designed as follows: ； in, This is the scaling factor. For the main user at every time The received power obtained by using the phase angle randomly generated by the IRS is, i.e., Equation (6). TD3 algorithm in time The learned optimal phase is By adjusting the phase angle of each reflective element in the IRS, the received power of the PU can be maximized.

7. The method for cooperative energy transfer and spectrum sharing assisted by a smart reflective surface according to claim 1, characterized in that, The optimal phase angle is obtained and configured based on the input channel state using the SDR algorithm; the details are as follows: phase angle optimization problem This can be summarized as a complex quadratic optimization problem with a unit modulus constraint; to maximize the received power, the received power... That is, expression (3) expands into a complex quadratic form: ； in, For the variable to be optimized, The constraint becomes ,and , , , The objective function is rewritten as: ； After expanding and combining like terms, we obtain the objective function: ； in , , Introducing augmented vectors: ； The objective function is then expressed as a quadratic form of the augmented vector: ； in Defined as: ； Therefore, the optimization problem Simplified to: ； First of all, Solution Perform eigenvalue decomposition. ,in It is the identity matrix. It is a real diagonal matrix; The suboptimal solution to the problem is ,in , It has a circularly symmetric complex Gaussian distribution; The objective solution to the problem is generated independently through a Gaussian randomization process. Select the maximum value, that is: ； Therefore, the optimal phase angle of the IRS is expressed as: .

8. A method for cooperative energy transfer and spectrum sharing assisted by a smart reflective surface according to any one of claims 1-7, characterized in that, The main system uses the IRS phase angle to achieve optimized received power and traversal time allocation coefficients. and To obtain the optimal time allocation coefficients and determine whether the achievable rate of the main system meets the target rate requirement, the process includes: after obtaining the optimal phase angle through the SDR or TD3 algorithm, fixing the optimal phase angle configuration, and optimizing it through a traversal method. and Specifically, traversal , For all feasible combinations, calculate the information transmission rate of the main system under each combination, as follows: Will The range of values Discretize by fixed step size, set traversal step size , Initialize the optimal reachable rate and optimal time allocation coefficient , ; For each traversal And execute in sequence: and Substituting into the target rate formula (4), we get Compare the current rate With target rate ,like Then, the achievable rate of the subsystem is calculated according to equation (5). ,like ,but , , ;when and After the traversal was completed, there was no and The combination enables the main system to achieve a certain speed. Once the target rate is reached, the system will transmit at that rate.

9. A computer device comprising a memory and a processor, wherein the memory stores a computer program, characterized in that, When the processor executes the computer program, it implements the steps of the intelligent reflective surface-assisted cooperative energy transfer and spectrum sharing method according to any one of claims 1-8.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by the processor, it implements the steps of the intelligent reflective surface-assisted cooperative energy transfer and spectrum sharing method according to any one of claims 1-8.

Images