A self-sustaining ris-assisted energy-carrying internet of things multi-dimensional resource allocation method

Abstract

The present application belongs to the field of wireless communication network, and relates to a kind of self-sustaining RIS assisted multi-dimensional resource allocation method of energy-carrying internet of things, comprising: constructing energy-carrying internet of things model;Construct information transmission model, energy collection model and power consumption model;According to information transmission model, energy collection model and power consumption model, establish energy efficiency maximization objective function;Two-dimensional resource joint optimization algorithm is used to optimize and solve the objective function, and the multi-dimensional resource allocation strategy is obtained;According to the multi-dimensional resource allocation strategy, the multi-dimensional resources of energy-carrying internet of things are allocated;The present application takes energy efficiency as the index for measuring system performance, and under the constraint conditions of meeting the available power of base station, RIS phase shift, energy causality of RIS, minimum energy requirement of internet of things equipment and information rate requirement, the two-dimensional resources of power domain and space domain are jointly optimized and managed.

Description

Technical Field

[0001] This invention belongs to the field of wireless communication network technology, specifically relating to a self-continuous RIS-assisted method for multi-dimensional resource allocation in the Internet of Things. Background Technology

[0002] With the rapid increase in the number of Internet of Things (IoT) devices, traditional battery-powered devices face severe energy constraints and sustainability challenges. Frequent charging and short battery lifespans impact practical applications and user experience. Meanwhile, Simultaneous Wireless Information and Power Transfer (SWIPT) technology, which transmits energy wirelessly, provides devices with a continuous, external power source, thereby improving device availability and lifespan.

[0003] Reconfigurable Intelligent Surface (RIS) is a wireless communication aid technology with high spectral efficiency, high energy efficiency, and low cost. It has attracted widespread attention in the wireless communication field in recent years and is considered one of the key technologies for next-generation wireless communication. RIS-assisted powered IoT can achieve efficient utilization of multi-dimensional resources such as spectrum, space, and energy, thereby achieving low-cost and high-reliability transmission and extending system uptime. Currently, research on RIS-assisted powered IoT mainly focuses on designing reasonable resource allocation algorithms to improve the information and energy transmission performance between IoT devices and base stations. However, these studies typically do not consider the energy consumption of the RIS itself. Adjusting the RIS state does consume energy, the magnitude of which depends mainly on the number of RIS units and the quantization accuracy. As the number of RIS units increases, its energy consumption becomes significant. Therefore, to achieve sustainable operation of powered IoT, it is necessary to consider the energy supply issues of both the RIS and IoT devices. Therefore, a novel RIS structure is proposed, in which each unit on the RIS is equipped with an energy harvesting module, and each unit has two operating modes: reflection mode and energy harvesting mode. Specifically, by allocating a portion of the units for energy harvesting to support the normal operation of the RIS (Resilient RIS), while the remaining units are used to reflect incident signals, the performance gains brought to IoT devices by the SWIPT (Switch) technology are enhanced. Self-sustaining operation of the RIS is achieved by calling different modes of different units. Currently, self-sustaining RISs mainly follow three operating protocols: Energy Splitting (ES), Time Switching (TS), and Mode Switching (MS). In the ES protocol, each unit on the RIS can simultaneously perform energy harvesting and signal reflection operations, adjusting the allocation ratio of energy harvesting and signal reflection according to different energy splitting ratios. The TS protocol can fully utilize time-domain resources, facilitating the design of passive beamforming for the RIS; however, this protocol has strict requirements for the synchronization capability of unit switching times and requires complex hardware deployment. In the MS protocol, each unit can flexibly adjust its operating mode, performing both energy harvesting and signal reflection, but each unit selects only one mode at a time. Compared to the TS and ES protocols, the MS protocol is easier to implement in practice and has lower signaling overhead. Therefore, the MS protocol is considered the most commonly used protocol in current research on self-sustaining RIS. However, current research on self-sustaining RIS mainly focuses on extending the operating time of the RIS, without comprehensively considering providing power to both IoT devices and the RIS simultaneously, thus failing to truly achieve sustainable operation of the IoT. Summary of the Invention

[0004] To address the problems existing in the prior art, this invention proposes a self-sustaining RIS-assisted multi-dimensional resource allocation method for energy-carrying IoT. This method includes: constructing an energy-carrying IoT model, which consists of a base station, IoT devices, and a self-sustaining RIS; constructing an information transmission model, an energy harvesting model, and a power consumption model based on the energy-carrying IoT model; establishing an energy efficiency maximization objective function based on the information transmission model, energy harvesting model, and power consumption model; optimizing the objective function using a two-dimensional resource joint optimization algorithm that integrates the power domain and spatial domain to obtain the optimal multi-dimensional resource allocation strategy; and allocating the multi-dimensional resources of the energy-carrying IoT according to the optimal multi-dimensional resource allocation strategy.

[0005] The beneficial effects of this invention are:

[0006] This invention allows units on the self-continuous RIS (Resonance System) to select different operating modes: some units are used for energy harvesting to support the normal operation of the RIS, while the remaining units are used to reflect incident signals, thereby improving the performance gain of SWIPT technology for IoT devices. Both the RIS and the IoT device terminal in this invention employ nonlinear energy harvesting models. This invention uses energy efficiency as a performance indicator, and under constraints such as base station available power, RIS phase shift, RIS energy causality, minimum energy requirements of IoT devices, and information rate requirements, it jointly optimizes and manages two-dimensional resources in the power domain (PS ratio) and spatial domain (active / passive beamforming, RIS unit operating mode). This invention proposes an iterative optimization algorithm combining penalty functions, fractional quadratic transformation methods, successive convex approximation (SCA), and semidefinite relaxation (SDR). The multi-dimensional resource management strategy proposed in this invention can effectively improve the energy efficiency of the system and has strong application value and practical significance. Attached Figure Description

[0007] Figure 1 This is a schematic diagram of the system model structure of the present invention;

[0008] Figure 2 This is a schematic diagram illustrating the working principle of the self-persistent RIS introduced in this invention.

[0009] Figure 3 A graph showing the relationship between system energy efficiency and RIS quantization accuracy;

[0010] Figure 4 A comparison chart of energy efficiency between continuous phase shift and 3-bit discrete phase shift;

[0011] Figure 5 The graph shows the relationship between the energy efficiency and the number of iterations for different schemes with different numbers of RIS reflector units.

[0012] Figure 6A graph showing the relationship between system energy efficiency and the number of RIS reflector units;

[0013] Figure 7 A graph showing the relationship between system energy efficiency and the number of IoTD users;

[0014] Figure 8 A graph showing the relationship between the energy efficiency and the number of iterations of the proposed solutions for different numbers of IoTD users;

[0015] Figure 9 A graph showing the relationship between the energy efficiency of Scheme 1 and the number of iterations for different numbers of IoTD users;

[0016] Figure 10 A graph showing the relationship between the energy efficiency and the number of iterations of Scheme 2 for different numbers of IoTD users;

[0017] Figure 11 The relationship between the maximum transmit power of a base station and its energy efficiency;

[0018] Figure 12 A schematic diagram showing the signal energy collected at IoTD users under different energy harvesting models;

[0019] Figure 13 A schematic diagram showing the signal energy collected by RIS under different energy harvesting models;

[0020] Figure 14 The diagram illustrates the impact of different energy harvesting models on system energy efficiency under varying minimum energy harvesting requirements. Detailed Implementation

[0021] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0022] A self-sustaining RIS-assisted multi-dimensional resource allocation method for energy-carrying IoT includes: constructing an energy-carrying IoT model, which consists of base stations, IoT devices, and a self-sustaining RIS; constructing an information transmission model, an energy harvesting model, and a power consumption model based on the energy-carrying IoT model; establishing an energy efficiency maximization objective function based on the information transmission model, energy harvesting model, and power consumption model; optimizing the objective function using a two-dimensional resource joint optimization algorithm that integrates the power domain and spatial domain to obtain the optimal multi-dimensional resource allocation strategy; and allocating the multi-dimensional resources of the energy-carrying IoT according to the optimal multi-dimensional resource allocation strategy.

[0023] In this embodiment, the system model structure is composed as follows: Figure 1 As shown, the system model consists of a base station equipped with M antennas, K single-antenna IoTDs, and a RIS with N units. In this network scenario, each IoTD belongs to a set. Each unit in RIS belongs to a set Between the base station and the IoTD, the RIS acts as an auxiliary device, providing SWIPT services to all IoTDs to ensure power supply and reliable communication. Each IoTD is equipped with a power-splitting receiver architecture, which can divide the received signal into two parts proportionally, one part being ρ. k Used for information transmission, another part 1-ρ k Used for energy harvesting. It is worth noting that when the number of cells on the RIS increases, the power consumption of the RIS also increases accordingly; therefore, the impact of RIS power consumption on the system cannot be ignored. To address this issue, this invention employs a novel RIS structure, namely a self-sustaining RIS. For example... Figure 2 As shown, each unit on this RIS is equipped with two circuits, one for signal reflection and the other for energy harvesting. When a unit operates in reflection mode, it reflects all incident signals back to the corresponding IoTD; when a unit operates in energy harvesting mode, it harvests the energy carried by the incident signals to maintain the RIS's operation. It is important to note that each unit can only operate in one mode at a time. For ease of representation, the variable α is used. n To define the mode selection of cells in RIS:

[0024]

[0025] exist Figure 1 In the system model shown, it is assumed that the reflection matrix of RIS is defined as Θ = AΦ, where It is a diagonal matrix. For the mode selection matrix of RIS, θ n ∈[0,2π). In this invention, the reflection coefficient of all elements is assumed to be 1. Furthermore, it is assumed that all channels experience frequency-flat quasi-static fading and have perfect CSI. The equivalent channels from BS to the k-th IoTD, from BS to RIS, and from RIS to the k-th IoTD are respectively represented as... as well as The signal emitted from BS can be represented as: in and These represent the active beamforming vector from BS to the k-th IoTD and the transmitted data symbols, respectively. The maximum transmit power at BS is P. max ,and

[0026] In this embodiment, the information transmission model of the system is constructed, and the signal received at the k-th IoTD can be represented as:

[0027]

[0028] in, This represents the Gaussian white noise at the k-th IoTD. The k-th IoTD uses a PS receiver architecture to convert the signal power ρ k The ∈(0,1) part is used for information decoding, and the remaining 1-ρ k Used for energy harvesting; For the equivalent channel from BS to the k-th IoTD, Let Θ be the reflection matrix of the RIS, G be the equivalent channel from the BS to the RIS, and x be the signal emitted from the BS. Therefore, the information decoding signal used for the k-th IoTD can be expressed as:

[0029]

[0030] in, Let represent the noise generated by the k-th IoTD device performing information decoding. Therefore, the signal-to-interference-plus-noise ratio (SIR) and corresponding data rate of the k-th IoTD are:

[0031]

[0032] in, The equivalent channel from BS to the k-th IoTD is added to the equivalent channel from BS to RIS and then from RIS to the k-th IoTD, w k For the active beamforming vector from BS to the k-th IoTD, w i For the active beamforming vector from BS to the i-th IoTD, Let be the Gaussian white noise power at the k-th IoTD. The noise power ρ generated by the information decoding operation performed by the k-th IoTD device k The signal power division factor. For all IoTD devices; SINR k Let be the signal-to-interference-plus-noise ratio (SIR) of the k-th IoT device.

[0033] Establish a system energy harvesting model. For the k-th IoTD, the signal used for energy harvesting is represented as:

[0034]

[0035] The power received by the k-th IoTD for energy harvesting can be calculated as follows:

[0036]

[0037] The total power of the signal received on the self-persistent RIS is calculated and represented as follows:

[0038]

[0039] Among them, A EH =I N -A represents the binary matrix used for energy harvesting. This represents the thermal noise generated at the RIS. Based on the above formula, the power of the RF signal received at the RIS can be calculated.

[0040]

[0041] in, This represents the operation of calculating the expectation, s k For data symbols transmitted from BS to the k-th IoTD.

[0042] This invention employs a nonlinear saturated energy harvesting model based on the Logistic function. Therefore, the actual energy received at the k-th IoTD and RIS can be calculated as follows:

[0043]

[0044] Where the subscripts k and r represent the k-th IoTD and RIS respectively, M {k,r} This represents the maximum power that the energy receiver can receive when the EH circuit is saturated, a {k,r} and b {k,r} These are two constants, the magnitude of which depends on the physical characteristics of the radio frequency circuit; For nonlinear saturated energy harvesting power, X {k,r} and Y {k,r} For parameters related to the model, This refers to the power received by the k-th IoT device for energy harvesting or the power of the radio frequency signal received at the RIS.

[0045] A system power consumption model is established. The total power consumption of the system consists of the static power consumption of IoTDs, RIS, and BS, the dynamic power consumption of all units on the RIS, and the power consumption consumed by the base station in transmitting signals. It can be expressed as:

[0046]

[0047] Where, p c,k P c,BS P c,RIS P represents the static power consumption of the k-th IoTD, BS, and RIS, respectively.D,r This represents the dynamic power consumption of each RIS unit during operation, where 0≤η≤1 represents the power amplification factor of the BS.

[0048] This embodiment proposes a self-sustaining, reconfigurable smart surface-assisted multi-dimensional resource management strategy for energy-carrying IoT. Under the constraints of total available power at the base station, phase shift and energy causality at the RIS (Radio Frequency Identifier), minimum energy requirements at the IoTD (Internet Device), and information rate requirements, it maximizes the energy efficiency of the IoT by jointly optimizing active / passive beamforming, the PS (Power Surge) ratio, and the operating mode of the cells at the RIS. The specific optimization problem can be expressed as:

[0049]

[0050] Where A represents the RIS mode selection strategy matrix, w k θ represents the active beamforming vector from the BS to the k-th IoTD. n ρ represents the RIS reflection angle. k P represents the PS scaling factor. max SINR represents the system's maximum transmitter power. k,req P represents the minimum required signal-to-interference-plus-noise ratio for the k-th IoTD. k,req And represents the minimum energy harvesting threshold value for the k-th IoTD. Constraint C1 limits the maximum transmit power of the base station. C2 is used to guarantee the quality of service for each IoTD. C3 and C4 are used to ensure that the energy received by each IoTD and RIS is greater than the minimum energy required for their operation. C5 is the power split ratio constraint. C6 is the phase shift constraint for the RIS. C7 constrains the operating mode of each unit on the RIS.

[0051] Because the objective function and constraints in the problem are non-convex, it is difficult to solve directly. Therefore, a two-dimensional resource joint optimization algorithm integrating the power domain (PS scaling factor) and the spatial domain (active / passive beamforming, RIS cell operating modes) is proposed. To solve the resulting non-convex optimization problem, an iterative optimization algorithm integrating penalty functions, fractional quadratic transformation methods, successive convex approximation, and semidefinite relaxation is designed. The specific operation is as follows:

[0052] First, let's transform the above problem. To solve the coupling problem between the mode selection matrix A and the passive beamforming Φ in the reflection matrix Θ, we define a reflection vector. and set The signal power received at the k-th IoTD is:

[0053]

[0054] in, and Rank(W k ) = 1, Then the signal-to-interference-plus-noise ratio of equation (4) is transformed into equation (14).

[0055]

[0056] Therefore, the data transmission rate of the k-th IoTD can be converted as follows:

[0057]

[0058] Similarly, the power received at the i-th IoTD and RIS is processed. Equation (7) can be transformed into the following:

[0059]

[0060] Equation (9) can be equivalently transformed into:

[0061]

[0062] Wherein, binary variable m n =1 indicates that the nth RIS unit is selected for energy harvesting, otherwise m n =0. T n =diag(t) n ), A vector that is 1 only at the nth element and 0 at the rest, i.e.:

[0063]

[0064] In the objective function, the denominator P s It can be converted as follows:

[0065]

[0066] Based on the above transformation, relax Rank(W) k After ) = 1, Problems can be transformed into problems

[0067]

[0068] Due to variable m n The problem is solved by using discrete 0-1 variables. It remains difficult to solve. To address this problem, we can work around the variable m. n Make the following changes, at this time It is equivalent to equation (21).

[0069]

[0070] Fractional Programming and Optimization Variable Decoupling: Next, we will focus on dealing with non-convex fractional objective functions. A commonly used method for handling fractional objective functions is the Tinkelbach method. This method introduces an auxiliary parameter to equivalently transform the fractional objective function into a subtractive form. The transformed subtractive optimization problem is solved in the inner loop, while the auxiliary parameter is iteratively updated in the outer loop. If the inner optimization problem can be solved globally, the algorithm will converge to the global optimum of the original problem. However, if the optimization problem in the subtractive form is still non-convex, a suboptimal solution to the inner optimization problem can usually only be obtained with acceptable computational complexity. In this case, the convergence of this method cannot be guaranteed. Furthermore, the Tinkelbach method is not suitable for solving the sum of the logarithms of fractional functions. To overcome these problems, this invention proposes an iterative algorithm based on quadratic transformation to handle the planned fractional problem. First, the problem is transformed using a quadratic transformation... This transforms into an equivalent optimization problem.

[0071] Due to interference between users, the numerator in the objective function is not a concave function. To address this challenge, an auxiliary variable γ is introduced. k Using variable γ k To replace SINR in the objective function numerator k SINR k With γ k The relationship can be represented as follows:

[0072]

[0073] To handle fractional problems in C8, an auxiliary variable β will be introduced. k Then, C8 is decomposed into C8a and C8b as follows.

[0074]

[0075] At this point, γ in equation (23) k ,β k There is variable coupling between them, and C8a can be equivalently transformed into equation (25).

[0076]

[0077] It is not difficult to see from C8a that variable W k , There is still a coupling relationship between them. To solve this problem, Lemma 1 will be used to apply the relationship between C8a and... Perform the transformation.

[0078] Lemma 1: For any two Hermitian matrices of the same size... and available:

[0079]

[0080] The right side of equation (26) can be calculated as follows:

[0081]

[0082] (a) indicates that the equation holds when both X and Y are Hermitian matrices.

[0083] According to Lemma 1, the following equation can be obtained:

[0084]

[0085] Furthermore, due to the variable ρ k W k ,m, The coupling relationship between them means that the denominator in the objective function is not a convex function. To solve this problem, a new auxiliary variable l will be introduced. k ,l r And respectively using variable l k and l r To replace the denominator of the objective function and and With l k and l r The relationship is as follows:

[0086]

[0087] P l NL ≥l r (30)

[0088] Substituting the mathematical expression of the nonlinear energy harvesting model into equation (29), we can obtain

[0089]

[0090] By introducing intermediate variables Equation (31) can be transformed as follows:

[0091]

[0092]

[0093] Observing equation (32), it can be seen that C9a is difficult to process directly. Therefore, an intermediate variable is introduced. Decompose C9a into equations (34) and (35).

[0094]

[0095] Similarly, substituting the mathematical expression of the nonlinear energy harvesting model into equation (30), we can obtain

[0096]

[0097] Similarly, introducing intermediate variables Equation (36) can be transformed as follows:

[0098]

[0099] In constraint C10a, the variable W can be found. k and m n There is still a coupling relationship between them. To solve this problem, an auxiliary variable will be introduced. The Big-M method is then used to convert C10a into a set of equivalent constraints.

[0100]

[0101] Among them, M big >>1 is a sufficiently large constant.

[0102] After the above transformation, the problem It can be equivalently transformed into a problem

[0103]

[0104] in, l={l k ,l r}, and

[0105] For the fractional problem of the objective function, the objective function can be transformed into the following form using the quadratic transformation method:

[0106]

[0107] Therefore, the problem It can be further converted as follows:

[0108]

[0109] Next, we will introduce a multidimensional resource management algorithm based on SCA and SDR. This will address the constraints... C8a,C8b, The non-convexity of C9b, C10b, C12, and C14 can be transformed into convex form, allowing us to obtain a suboptimal solution using the SCA method. First, we transform the rank-1 constraint of C14 using Lemma 2.

[0110] Lemma 2: The rank-1 constraint of C14 can be equivalently transformed into

[0111]

[0112] in, and They represent The kernel function and spectral function.

[0113] because If it is a Hermitian matrix, then in For matrix The i-th singular value of the matrix. Therefore, the matrix has a singular value if and only if ... When the rank is 1, equation (44) holds.

[0114] Solve using the penalty function method The problem involves equality constraints. Compare equation (44) with... Substitution In the objective function, the optimization problem Can be converted

[0115]

[0116] When the penalty terms λ→∞ and χ→∞, the problem The solution satisfies The equality constraints of equation (44). However, if the initial values ​​of λ and χ are too large, The objective function may be dominated by the penalty term, which could significantly affect the objective function. To avoid this effect, this invention first chooses relatively small initial values ​​λ and χ, and gradually increases λ and χ to a sufficiently large value during the iteration process until a satisfactory result is obtained. A feasible solution to equation (44).

[0117] Given arbitrary penalty factors λ and χ, due to the penalty term in the objective function and constraints... C12 is still non-convex, causing problems. It is also not convex. To solve this problem, the SCA method will be used. Process it.

[0118] In the t-th iteration of SCA, for a given Using a first-order Taylor expansion, a convex upper bound for the penalty term can be obtained:

[0119]

[0120] in, yes The eigenvector corresponding to the largest eigenvalue. Similarly, by considering the eigenvector with respect to m... n The non-convex part of the penalty term is expanded using a first-order Taylor series.

[0121] For any feasible point The following equation holds true.

[0122]

[0123] Therefore, regarding m n The penalty item is transformed into the following form:

[0124]

[0125] Furthermore, in C9b and C10b, and For variables and For convexity, their lower bounds can be obtained by performing a first-order Taylor expansion at any feasible point. Given feasible points in the t-th iteration... and The corresponding first-order Taylor expansion is as follows:

[0126]

[0127] Based on this, C9b and C10b can be converted into the following form:

[0128]

[0129] Similarly, given initial values and The following formula can be obtained:

[0130]

[0131] Similarly, the convex upper bounds of equations (27) and (28) can be obtained using the SCA method.

[0132]

[0133] Therefore, C8a and C8b can be transformed into the following form:

[0134]

[0135] Using the same method and The conversion is performed as follows:

[0136]

[0137] Finally, for a given initial value question It can be converted to:

[0138]

[0139] Currently, the problem The only non-convex factor is the rank-1 constraint of C12. This section uses the SDR method to relax the rank-1 constraint. At this point, the problem... This can be solved using the CVX solver. Assume that the channel vectors of all users are statistically independent, and P... max >0, then the optimization problem The rank-1 constraint in the equation is always satisfied.

[0140] Next, this invention designs an iterative algorithm based on double-loop penalty to solve the problem. In the outer iteration, there are two penalty terms λ and χ, which gradually increase according to the rules λ = ω1λ and χ = ω2χ, where ω1 and ω2 > 1. The algorithm terminates when the penalty terms satisfy the condition shown in equation (60).

[0141]

[0142] Where ε > 0 is a predefined precision that satisfies the iteration termination condition. In the inner loop, The algorithm is optimized through iterative iterations under a certain penalty factor, and the objective function is non-decreasing. When λ and χ approach infinity and the objective function is non-decreasing, the conditions of equations (46) and (47) can be satisfied. The steps of this iterative algorithm are summarized in detail in Algorithm 1.

[0143] Table 1 Iterative Algorithm Based on Penalty Function

[0144]

[0145] The main complexity of Algorithm 1 stems from solving the problem. The internal loop solves the problem. Specifically, the complexity of each iteration of this algorithm can be calculated as... in, and ε represents the number of inequalities and variables in the proposed algorithm, respectively, and ε is the convergence accuracy of the algorithm.

[0146] The following describes the acquisition of discrete phase shifts. In practical applications, due to hardware limitations, employing continuous phase shifts on RIS systems is very challenging. To address this issue, this invention uses a phase shift quantization method to obtain discrete phase shifts. Assuming b represents the quantization precision of the cell phase shift in the RIS, the discrete phase shift set can be represented as... As is well known, exhaustive search can precisely obtain the optimal discrete phase shift for every unit in a RIS, but its computational complexity is high, specifically... Furthermore, the computational complexity of this method increases exponentially with the number of RIS units and the quantization precision b of the RIS units. To obtain discrete phase shifts with lower complexity, this invention employs a low-complexity discrete phase shift quantization method, the specific steps of which are as follows: By calculating the following formula... Optimal continuous phase shift Projection to set The closest discrete phase shift point in the middle distance. The computational complexity of this discrete phase shift quantization method is O(n). Much smaller than exhaustive search It is more suitable for practical applications.

[0147] Simulation results were used to evaluate the performance of a self-sustaining RIS-assisted multidimensional resource allocation scheme in a powered Internet of Things (IoT). To verify the effectiveness of the proposed scheme, the following four benchmark schemes were used:

[0148] 1) Comparison Scheme 1 (Fix PS ratio): A resource allocation scheme for powered IoT devices with a self-sustaining RIS-assisted system using a fixed power split ratio. In this scheme, the power split ratio at all IoTDs is 0.5, i.e., ρ = 0.5. Furthermore, the active beamforming vector, phase shift at the RIS, and mode selection are all obtained using the method proposed in this scheme.

[0149] 2) Comparison Scheme 2 (All elements at the RIS working in reflection mode): A resource allocation scheme for powered IoT equipped with a total reflection RIS. This scheme does not consider the self-sustainability of the RIS. Furthermore, the active beamforming vector, phase shift at the RIS, and IoTD power split ratio optimization are obtained using the method proposed in this scheme.

[0150] 3) Comparison Scheme 3 (Half elements at the RIS working in reflection mode): A resource allocation scheme for a self-sustaining RIS-assisted powered IoT with a fixed RIS operating mode. In this scheme, half of the units on the RIS are used for energy harvesting operations, and the other half are used for signal reflection operations. Furthermore, the active beamforming vector, phase shift at the RIS, and IoTD power split ratio optimization are obtained using the method proposed in this scheme.

[0151] 4) Comparison Scheme 4 (MRT-based active beamforming): A resource allocation scheme based on Maximum Ratio Transmission (MRT). In this scheme, the active beamforming vector at the base station is obtained using the MRT method. Other variables, such as phase shift at RIS, RIS operating mode, and power division ratio at IoTD, are calculated using the method proposed in this scheme.

[0152] In terms of system deployment, a two-dimensional coordinate system is adopted. Specifically, the BS and RIS are located at coordinates (0,0) meters and (5,5) meters, respectively. K IoTDs are randomly distributed within an area with a radius of 5 meters centered at coordinate (5,0). Furthermore, all channels experience both large-scale and small-scale fading. Large-scale fading uses a distance-dependent path loss model, while small-scale fading follows a Rayleigh distribution. The fading coefficients for the base station to RIS link, base station to IoTD link, and RIS to IoTD link are set to 2.2, 3.6, and 2.2, respectively. Unless otherwise specified, the default settings for this invention are: 20 RIS units, 4 IoTDs in the network, 6 BS antennas, a minimum signal-to-interference-plus-noise ratio (SNR) of 10 dB at the IoTDs, a minimum energy harvesting threshold of -20 dBm at the IoTDs, a discrete phase-shift quantization accuracy of 3-bit, and a maximum transmit power of 38 dBm at the BS. Specific simulation parameter settings are given in Table 2.

[0153] Table 2 Simulation parameter settings for self-sustaining RIS-assisted powered Internet of Things

[0154]

[0155]

[0156] Next, we will analyze the simulation results, focusing first on the relationship between different quantization accuracies of the RIS unit and the system energy efficiency.

[0157] Figure 3The energy efficiency comparison results of the proposed solution under different RIS quantization accuracies are presented. As can be observed from the figure, when the number of RIS cells is 10 and 20, the system energy efficiency using 3-bit quantization accuracy is higher than that using 2-bit and 4-bit. This indicates that as the RIS cell quantization accuracy increases, the system energy efficiency first increases and then decreases. This is because as the quantization accuracy increases, the system's sum rate first increases and then tends to stabilize, but the RIS power consumption continuously increases. When the quantization accuracy is within a small range, as the quantization accuracy increases, the system's sum rate gain exceeds the introduced power consumption, thereby improving the system's energy efficiency. However, when the quantization accuracy is high, although the achievable system sum rate increases, the increase in power consumption exceeds the rate gain, thus reducing the system's energy efficiency.

[0158] Figure 4 The impact of different schemes on energy efficiency when using 3-bit quantization precision and continuous phase shift is presented. In this simulation, the dynamic power consumption of the RIS cell is set to 1.5mW. As can be seen from the figure, the energy efficiency with 3-bit quantization precision is slightly lower than with continuous phase shift. This is because discrete phase shift introduces a certain precision loss, resulting in a slight reduction in the total sum rate compared to an ideal continuous phase shift. However, the discrete loss values ​​for the five different schemes are 1.34%, 1.84%, 0.33%, 4.4%, and 0.63%, respectively, all within acceptable ranges. Furthermore, continuous phase shift has high power consumption and complexity, making it difficult to implement in practical applications. In summary, the simulation results show that in the system considered in this invention, using 3-bit quantization precision for the RIS achieves the optimal trade-off between sum rate and power consumption and is easy to implement. Therefore, subsequent simulations in this invention all use 3-bit quantization precision.

[0159] The relationship between RIS scale and system energy efficiency. Figure 5 This demonstrates the impact of various comparative schemes on system energy efficiency under different RIS cell sizes N. From Figure 5 As can be seen from the graph, for RIS with 5, 10, 15, 20, 25, and 30 cells, the energy efficiency of all compared schemes initially increases and then stabilizes with increasing iterations, indicating that the algorithm has convergence. The convergence of different schemes shows that the scheme proposed in this invention converges relatively slowly but has the best performance. This also indicates that the more resources used in joint optimization, the better the energy efficiency, but more computation time is required to achieve optimal convergence. Furthermore, the graph shows that the energy efficiency of the scheme proposed in this invention is superior to other schemes at different RIS scales. This is because the scheme proposed in this invention can simultaneously perform joint optimization of multiple variables, dynamically adjusting the value of each variable according to requirements, thereby achieving optimal system energy efficiency.

[0160] Figure 6 The results show how system energy efficiency changes with the number of RIS units when using different comparative schemes. The results indicate that, within a certain range, the system energy efficiency initially increases with the increase in the number of RIS units. This is because the system speed gained by the RIS units to some extent exceeds the energy consumption introduced by the RIS units, thus improving the system's energy efficiency. Furthermore, it can be seen that for any number of units N, the scheme proposed in this invention is superior to other comparative schemes, further demonstrating the advantages of the proposed scheme.

[0161] Figure 7 The relationship between system energy efficiency and the number of users is presented when different schemes are adopted. From Figure 7 As can be seen, in each scheme, the system's energy efficiency increases with the number of users. Compared with the other four schemes, the scheme proposed in this invention exhibits better energy efficiency performance under different user numbers. This is because this scheme performs joint optimization of multi-dimensional resources, which can dynamically adjust the allocation of multi-dimensional resources according to demand. In particular, compared with the resource allocation scheme using a fixed power division ratio at the user, the scheme proposed in this invention can reasonably allocate the power division ratio at the user according to different channel states, that is, allocate different information decoding and energy harvesting ratios to different users. The optimization of the power allocation ratio mainly affects the trade-off between information rate and energy harvesting at the IoTD, thereby effectively improving the system's energy efficiency performance. The scheme proposed in this invention can make full use of multi-dimensional resources and effectively improve system performance. Furthermore, its performance advantage will become more obvious as the number of users increases.

[0162] Figures 8-10 The figures show the convergence performance of the proposed solution, comparative solution 1, and comparative solution 2 under different IoTD user numbers. As can be seen from these three figures, the convergence performance of all three solutions gradually decreases with the increase in the number of users. This indicates that with an increasing number of users, all three solutions require more iterations to achieve the optimal resource allocation result. Furthermore, comparing the convergence of the three solutions shows that the proposed solution has a relatively slow convergence speed. This indicates that more resources jointly optimized can improve energy efficiency, but it also requires sacrificing more computation time to reach the optimal convergence state.

[0163] from Figure 11The relationship between maximum transmit power and system energy efficiency can be observed. In each scheme, the system energy efficiency decreases as the maximum transmit power at the base station increases. This is because, according to equation (11), the system power consumption is closely related to the power consumption consumed by beam transmission, and this power consumption accounts for a large proportion in the energy efficiency formula. Therefore, as the transmit power increases, the power consumption in the denominator of the energy efficiency formula also increases. Although increasing the transmit power can improve the system's "sum rate," the increase in rate is relatively small compared to the increase in power consumption. Therefore, as the maximum transmit power increases, the system energy efficiency decreases. However, from Figure 12 Looking at the slope of the energy efficiency decrease, the rate of decrease in system energy efficiency gradually decreases as the maximum transmit power increases. Furthermore, compared to the other four schemes, the scheme proposed in this invention exhibits better energy efficiency performance at different maximum transmit powers. This is because the scheme of this invention employs joint optimization of multiple variables, which helps to improve the system's energy efficiency performance.

[0164] Compared to other solutions, the proposed solution exhibits superior energy efficiency under varying minimum energy harvesting requirements. This further validates that the present invention achieves optimal energy efficiency by jointly optimizing multiple variables and dynamically adjusting resource allocation based on demand. Regardless of changes in minimum energy harvesting requirements, the proposed solution maintains significant advantages. Figure 13 A comparison between the nonlinear and linear energy harvesting models at the RIS under the given parameters is presented. The figure shows that, under the set parameters, when the received signal power is less than 0.1W, the nonlinear energy harvesting model nonlinearly enhances the received energy.

[0165] To comprehensively analyze the impact of the energy harvesting model on the system's energy efficiency, this invention evaluated the energy efficiency performance of the proposed scheme under four different conditions through simulation experiments. These four conditions are:

[0166] Case 1: Both IoTD and RIS use linear energy harvesting models.

[0167] Case 2: Both IoTD and RIS employ nonlinear saturated energy harvesting models.

[0168] Case 3: IoTD uses a linear energy harvesting model, while RIS uses a nonlinear saturated energy harvesting model.

[0169] Case 4: RIS uses a linear energy harvesting model, while IoTD uses a nonlinear saturated energy harvesting model.

[0170] Figure 14The graph shows the energy efficiency comparison results for the four scenarios above under different user minimum energy harvesting requirements. It can be observed from the graph that, under different user minimum energy harvesting requirements, Case 3 has the highest system energy efficiency, while Case 4 has the lowest. This is because, according to... Figure 13 and Figure 14 As a result, for IoTD users, the energy harvesting power under the nonlinear energy harvesting model is less than that under the linear energy harvesting model. However, for the RIS (Radio Recycling System), when the received signal power is less than 0.08W, the power harvested by the RIS under the nonlinear energy harvesting model is greater than that under the linear energy harvesting model. Therefore, Case 3 has a higher energy conversion efficiency than Case 4. In summary, Case 3 exhibits the highest system energy efficiency, while Case 4 exhibits the lowest.

[0171] The above-described embodiments further illustrate the purpose, technical solution, and advantages of the present invention. It should be understood that the above-described embodiments are merely preferred embodiments of the present invention and are not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made to the present invention within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A self-sustaining RIS-assisted multi-dimensional resource allocation method for the energy-carrying Internet of Things, characterized in that, include: A power-carrying IoT model is constructed, consisting of base stations, IoT devices, and a self-sustaining RIS (Resource Identifier). Based on the power-carrying IoT model, an information transmission model, an energy harvesting model, and a power consumption model are constructed. An objective function for maximizing energy efficiency is established based on the information transmission model, energy harvesting model, and power consumption model. A two-dimensional resource joint optimization algorithm that integrates the power domain and spatial domain is used to optimize and solve the objective function, thereby obtaining the optimal multi-dimensional resource allocation strategy. The multidimensional resources of the energy-carrying Internet of Things are allocated according to the optimal multidimensional resource allocation strategy. Energy harvesting models include: obtaining the first The signal used for energy harvesting in the first IoT device; calculate the first The power of the IoT device used for energy harvesting; calculate the total power of the self-continuous RIS received signal, and calculate the RF signal power based on the total power of the self-continuous RIS received signal; according to the first The power and radio frequency signal power of the IoT device used for energy harvesting are calculated using a nonlinear saturated energy harvesting model based on the Logistic function. The energy actually received by each IoT device and the self-contained RIS; The power consumption model is as follows: ； in, , , They represent the first The static power consumption of each IoTD, BS, and RIS This indicates the dynamic power consumption of each RIS unit during operation. This represents the power amplification factor of BS. The total power consumption of the system is This represents the total static power consumption of the system. Each unit on the RIS forms a set. Select variables for the mode of each unit on the RIS. From the base station to the Active beamforming vector for an IoT device For nonlinear energy harvesting power, This refers to Internet of Things (IoT) devices or RIS (Reference Equipment). The objective function for maximizing energy efficiency is established as follows: ； in, The matrix representing the pattern selection strategy of RIS Indicates BS to the number Active beamforming vector of an IoTD Indicates the RIS reflection angle. Indicates the PS scaling factor. This indicates the system's maximum transmitter power. Representing the The minimum signal-to-interference-plus-noise ratio required for an IoTD Representing the Minimum energy harvesting threshold for each IoTD; constraints The maximum transmission power of the base station was limited; Used to ensure the quality of communication services for each IoTD; and These are used to ensure that the energy received by each IoTD and RIS is greater than the minimum energy required for them to operate; Power segmentation ratio constraint; It is a phase shift constraint on RIS; This constrains the operating mode of each unit on the RIS.

2. The self-sustaining RIS-assisted multi-dimensional resource allocation method for energy-carrying IoT according to claim 1, characterized in that, Constructing the information transmission model includes: obtaining the signal-to-interference-plus-noise ratio (SIR) and corresponding data rate for each IoT device, and constructing the information transmission model based on the SIR and data rate; where the first... The signal-to-interference-plus-noise ratio of each IoT device is: ； The data rate is: ； in, For the first The sum of the channel gain vectors of each IoT device, From the base station to the Active beamforming vector for an IoT device From the base station to the Active beamforming vector for an IoT device For the first Gaussian white noise at the location of an IoT device The noise generated by the information decoding operation performed by the first IoT device. The signal power split ratio, It is the collection of all Internet of Things (IoT) devices; For the first The signal-to-interference-plus-noise ratio of an IoT device.

3. The self-sustaining RIS-assisted multi-dimensional resource allocation method for the energy-carrying Internet of Things according to claim 1, characterized in that, Calculate the first The actual energy received by each IoT device and the self-sustaining RIS is: ； Among them, subscript and Representing the first One IoTD and RIS, This indicates the maximum power that the energy receiver can receive when the EH circuit is saturated. and These are two constants; For nonlinear saturated energy harvesting power, , These are all parameters related to the model. For the first An IoT device receives power for energy harvesting or radio frequency signal power received at the RIS.

4. The self-sustaining RIS-assisted multi-dimensional resource allocation method for the energy-carrying Internet of Things according to claim 1, characterized in that, The optimization of the objective function using a two-dimensional resource joint optimization algorithm includes: transforming the maximization objective function P1 into a fractional objective function problem P2; using an iterative algorithm based on quadratic transformation to process the planned fractional problem, i.e., transforming problem P2 into an equivalent optimization problem P3; transforming the objective function using the quadratic transformation method; transforming problem P3 into problem P4 based on the transformed objective function; constructing a penalty function and using the penalty function to transform problem P4 into problem P5; processing problem P5 using the SCA method to obtain problem P6; and solving problem P6 using an iterative algorithm based on double-loop penalty to obtain the optimal solution, where the optimal solution is the optimal multidimensional resource allocation strategy.

5. A self-sustaining RIS-assisted multi-dimensional resource allocation method for energy-carrying Internet of Things according to claim 4, characterized in that, The formula for question P6 is: ； in, For the reflection vector covariance, From the base station to the Covariance of the active beamforming vector of an IoT device This is the power allocation ratio factor. , , All of these are introduced auxiliary variables. To introduce new auxiliary variables to replace those in the denominator of the objective function and , To represent the energy harvesting of RIS cells, a vector composed of binary variables is selected. For the introduction of intermediate variables, To transform the objective function into its current form using the quadratic transformation method, and All are penalty items. For kernel function, Let n be the spectral function and n be the RIS unit. The set of the total number of RIS units. For the first A binary variable indicating whether a RIS unit is selected for energy harvesting. For base station transmit power limitations, The quality of communication services for each IoTD The energy received by the IoTD is greater than the minimum energy required for operation. The RIS receives more energy than the minimum energy required for it to operate. , , , , , , , , These are all constraints generated when problem P6 was transformed after introducing auxiliary variables.

6. The self-sustaining RIS-assisted multi-dimensional resource allocation method for energy-carrying Internet of Things according to claim 5, characterized in that, The iterative algorithm based on double loop penalty is used to solve problem P6, which includes an outer loop iteration and an inner loop. In the outer iteration, two penalty terms are set. and and in accordance with and The patterns gradually increased, among which This is the factor that increases the penalty term; the iteration ends when the penalty term satisfies the condition; the condition is: ； in, It is a predefined precision that satisfies the iteration termination condition. and Representing the covariance of the reflection vector respectively kernel function and spectral function, Indicates the first A binary variable is selected for energy harvesting for each RIS unit, and vice versa. ; Represents a set of RIS units; In the inner loop, The optimization is performed iteratively under a certain penalty factor, and the objective function exhibits non-decreasing behavior; whereby... = indicates from BS to the... The active beamforming covariance matrix of an IoTD; The auxiliary variable introduced; S when and When the objective function approaches infinity and is non-decreasing, the optimal solution to the objective problem P6 is obtained by updating the values ​​of feasible variables and the penalty factor.

Images