Communication resource joint optimization method for wireless charging sensor network

Abstract

The invention provides a communication resource joint optimization method for a wireless charging sensor network. A downlink wireless energy-carrying communication process and an uplink pure communication process in the wireless rechargeable sensor network are comprehensively considered, a sensor energy consumption model of a serial uplink and downlink process is established, and an uplink and downlink communication resource joint optimization model is constructed. The model takes minimization of total energy consumption of a system as an optimization target, ensures uplink and downlink communication quality and downlink energy transmission quality at the same time, and performs joint optimization on uplink and downlink time allocation, transceiving beam forming of a fusion center and sensor power allocation.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of Internet of Things, and in particular to a communication resource joint optimization method for a wireless rechargeable sensor network. BACKGROUND

[0002] As one of the emerging information industries that are currently developing rapidly, the Internet of Things (IoT) integrates the latest technologies of mobile communication and the Internet. In IoT applications, intelligent perception of the real world is of great importance. Wireless Sensor Networks (WSNs) play a core role in intelligent perception due to their low cost, flexible deployment, and strong self-organizing capabilities. However, in some scenarios where power is limited or power replacement is difficult, the limited power reserves of sensors cannot meet the energy consumption requirements for long-term operation. To improve the adaptability of WSNs in such scenarios, researchers have introduced wireless energy transmission technology and proposed the concept of Wireless Rechargeable Sensor Networks (WRSNs). By wirelessly charging sensors, the limitations of local power sources on sensor life are effectively eliminated, thereby enhancing their continuous working capabilities.

[0003] According to the design of the sensor receiver, the working mode of WRSN can be divided into two categories. The first category is wireless charging and wireless communication based on time switching mode, i.e., the Fusion Center (FC) transmits energy to the sensor in the wireless energy transmission phase and transmits information in the wireless communication phase. The second category is wireless power division mode-based wireless power communication, in which the FC transmits energy and information to the sensor simultaneously through wireless power signals. The sensor receiver is equipped with a power divider that divides the received signal into two parts: one part is used for signal decoding, and the other part is supplemented to the local power supply through an energy collector. Compared with the time switching mode, the power division mode can achieve parallel transmission of energy and information, and usually has higher energy efficiency and spectrum utilization. Currently, research on WRSN mainly focuses on the optimization of downlink unidirectional communication resources, usually assuming that the energy consumption of the sensor is a fixed value. However, in actual applications, the working energy consumption of the sensor usually changes dynamically with its uplink communication power. Therefore, joint optimization of uplink and downlink communication resources is a key research direction for improving the performance of WRSN, and is worth further in-depth discussion. SUMMARY

[0004] To solve the limitations and defects of the prior art, the present application provides a communication resource joint optimization method for a wireless rechargeable sensor network, comprising:

[0005] The expression (10) of the optimization model is as follows:

[0006]

[0007] where w k denotes the beamforming vector at the fusion center for downlink, v k denotes the beamforming vector at the fusion center for uplink, p k denotes the power splitting factor of the sensor, P k denotes the uplink power of the sensor, τ1 denotes the time for downlink, τ2 denotes the time for uplink, Pr(·) denotes the probability of an event, denotes the signal-to-interference-plus-noise ratio (SINR) for downlink, denotes the signal-to-interference-plus-noise ratio (SINR) for uplink, denotes the minimum code rate requirement for downlink communication, denotes the minimum code rate requirement for uplink communication, and denotes the maximum outage probability that the system can tolerate, P O,k is the operational energy consumption of the kth sensor, η is the energy harvesting efficiency of the sensor, is the estimated value of the channel between the kth sensor and the fusion center, T is the total time limit;

[0008] The expression (10) is converted into expression (24) using the successive convex approximation method as follows:

[0009]

[0010] where, denotes the optimized value of τ1 for the nth iteration, is the downlink power of the fusion center for the nth iteration;

[0011] The optimization problem is restructured to obtain expression (26) as follows:

[0012]

[0013] where, denotes all the variables to be optimized in the new optimization problem, Rank(·) denotes the rank of a matrix, and Tr(·) denotes the trace of a matrix;

[0014] The optimization problem is solved using the alternating optimization algorithm to obtain expression (27) as follows:

[0015]

[0016] where, is the set of all the variables to be optimized, and is the auxiliary variable introduced;

[0017] The relaxation variable {α1,k ,α 2,k ,α 3,k , the optimization problem is transformed into expression (28) as follows:

[0018]

[0019] where, is the set of all to-be-optimized variables, and are auxiliary variables introduced.

[0020] Optionally, it further comprises:

[0021] According to the properties of the second conversion, the auxiliary variables are updated as follows:

[0022]

[0023]

[0024] where, is the noise power in the downlink process, is the noise power in the uplink process, is the estimation error of the channel between the ith sensor and the fusion center,

[0025] Optionally, the step of transforming expression (10) into expression (24) using the successive convex approximation method further comprises:

[0026] Taking the logarithm of both sides of expression (24), the constraint expression decoupling the uplink and downlink time length from the remaining variables is as follows:

[0027]

[0028] where, ρ k represents the power splitting factor of the sensor, P k represents the uplink power of the sensor, τ1 represents the time of the downlink, τ2 represents the time of the uplink, P O,k is the operating energy consumption of the kth sensor, η is the energy harvesting efficiency of the sensor, is the estimation value of the channel between the kth sensor and the fusion center,

[0029] Optionally, the step of obtaining expression (10) of the optimization model further comprises:

[0030] Introducing auxiliary variables and The probability constraints are converted into expressions (11-13) as follows:

[0031]

[0032]

[0033]

[0034] where τ1 represents the time of the downlink, τ2 represents the time of the uplink, Pr(·) represents the probability of an event, represents the signal-to-interference-plus-noise ratio of the downlink, represents the signal-to-interference-plus-noise ratio of the uplink, represents the maximum outage probability that the system can tolerate, represents the power of the signal, represents the power of the interference, represents the power of the noise, is the noise power in the downlink process;

[0035] The following variables are defined:

[0036]

[0037] where w k is the downlink beamforming vector of the fusion center, Δh k is the random channel estimation error, ρ k is the sensor power splitting factor,

[0038] Expression (13) is approximated to the following expressions (15-16) according to the variable expression (14):

[0039]

[0040] where is an auxiliary variable introduced, is the expression of the trace of, is the expression of the Frobenius norm of, is the expression of, is the expression of as follows:

[0041]

[0042] Optionally, it also includes:

[0043] The variables and are introduced

[0044]

[0045] where diag(·) denotes the operation of constructing a matrix with the vector elements as the diagonal elements, denotes the Kronecker product, v k is the uplink beamforming vector, P k is the sensor uplink transmit power, is the variance of the channel estimation error;

[0046] The uplink outage probability constraint is converted into the following expressions (19-22) according to expression (18):

[0047]

[0048] where, and are auxiliary variables introduced, τ2 represents the time of the uplink, denotes the minimum code rate requirement of the downlink communication;

[0049] Each term in expressions (19-22) is calculated as follows:

[0050]

[0051] where Pr(·) denotes the probability of an event, is the estimated value of the channel between the kth sensor and the fusion center, denotes the maximum outage probability that the system can tolerate.

[0052] Optionally, the step of obtaining expression (10) of the optimization model includes:

[0053] The channel state information estimation error is modeled as a statistical error, and the expression is as follows:

[0054]

[0055] where the channel from the fusion center to the kth sensor in the downlink process is denotes the set of complex numbers, and the channel from the kth sensor to the fusion center in the uplink process is denotes the estimation result of the channel state information, denotes the channel state information estimation error, and CN represents a complex Gaussian distribution;

[0056] The signal component for energy harvesting and the component for signal decoding received by the kth sensor in the downlink process are as follows:

[0057]

[0058] where ρ kis a power splitting factor, is an actual channel, n k is noise of the downlink process, is a symbol sent to the kth sensor, and the power satisfies is is a corresponding beamforming vector, n k is antenna noise of the receiving end, and the distribution obeys

[0059] Optionally, the method further comprises:

[0060] According to expression (2), the energy collected by the sensor in the downlink process is expressed as follows:

[0061]

[0062] wherein, is an estimated value of the channel between the kth sensor and the fusion center, η is the energy collection efficiency of the sensor, w k is a downlink beamforming vector of the fusion center, and τ1 represents the time of the downlink;

[0063] According to expression (2), the component for signal decoding is expanded as:

[0064]

[0065] wherein, Δh k is a random error of the channel estimation, and the power of each signal component is expressed as:

[0066]

[0067] The signal-to-noise ratio of the downlink communication of the sensor is expressed as:

[0068]

[0069] Optionally, the method further comprises:

[0070] The signal of the kth sensor received by the fusion center in the uplink process is:

[0071]

[0072] wherein, is an estimated value of the channel between the kth sensor and the fusion center, is a symbol transmitted by the kth sensor, and the power is is a beamforming vector of the fusion center for receiving the signal of the kth sensor, and the beamforming vector is normalized as follows: ||v k || 2 = 1, denotes the antenna noise of the fusion center.

[0073] Optionally, further comprising:

[0074] The power of each signal component in the uplink process is represented as:

[0075]

[0076] where v k is the uplink beamforming vector, is the estimated value of the channel between the kth sensor and the fusion center, Δh k is the random error of the channel estimation, P k is the sensor uplink transmission power, is the noise power of the uplink process;

[0077] The uplink stage signal-to-noise ratio is represented as:

[0078]

[0079] The present application has the following beneficial effects:

[0080] The present application proposes a communication resource joint optimization method for wireless charging sensor networks. The downlink wireless energy-carrying communication and uplink pure communication process in the wireless rechargeable sensor network are comprehensively considered, a sensor energy consumption model for the series uplink and downlink process is established, and an uplink and downlink communication resource joint optimization model is constructed. The model minimizes the total system energy consumption as the optimization objective, while ensuring the uplink and downlink communication quality and the downlink energy transmission quality, and jointly optimizes the uplink and downlink time allocation, the transmit and receive beamforming of the fusion center, and the sensor power allocation.

[0081] In order to solve the joint optimization problem, the present application first processes the outage probability constraint and other non-convex constraints, decouples part of the optimization variables, and performs convex approximation on the non-convex terms. The present application proposes an alternating optimization algorithm to optimize the fusion center end and the sensor end in the joint model respectively, so as to realize efficient solution. The simulation experimental results provided by the present application have verified the effectiveness of the optimization algorithm, and proved the superior performance of the optimization algorithm in reducing the total system energy consumption. BRIEF DESCRIPTION OF DRAWINGS

[0082] Figure 1 The system model schematic diagram provided by the first embodiment of the present application.

[0083] Figure 2 The optimization algorithm flowchart provided by the first embodiment of the present application.

[0084] Figure 3 The convergence result graph of the algorithm alternating optimization provided by the first embodiment of the present application.

[0085] Figure 4 The simulation result diagram of the algorithm provided by the embodiment one of the present application under different fusion center antenna numbers.

[0086] Figure 5 The simulation result diagram of the algorithm provided by the embodiment one of the present application under different interruption probability tolerances. DETAILED DESCRIPTION

[0087] In order for those skilled in the art to better understand the technical solutions of the present application, the communication resource joint optimization method for wireless charging sensor network provided by the present application is described in detail below in combination with the drawings.

[0088] Embodiment one

[0089] The embodiment is directed to a wireless rechargeable sensor network (WRSN) in a power division mode, studies the joint optimization problem of uplink and downlink communication resources, and proposes a high energy efficiency and strong robustness joint optimization algorithm for uplink and downlink time allocation, fusion center (FC) beamforming and sensor end power division. While ensuring the uplink and downlink communication quality of the WRSN, the algorithm can effectively reduce the overall communication energy consumption of the system and improve the energy utilization efficiency of the network.

[0090] The embodiment takes the communication process between the FC and the sensor in the WRSN as the research object. Therefore, the embodiment first introduces the overall communication framework and the specific uplink and downlink signal model of the WRSN, and then introduces the uplink and downlink joint optimization model and the corresponding joint optimization algorithm established by the embodiment.

[0091] Figure 1 The system model schematic diagram provided by the embodiment one of the present application. As Figure 1As shown, the embodiment sets the WRSN to be composed of one FC equipped with M antennas and K single-antenna sensors deployed in a scattered manner. The communication between the FC and the sensors in the WRSN is completed based on a Time Division Duplex (TDD) framework, that is, one complete communication cycle is divided into two stages, uplink and downlink, and the time lengths of the two stages are denoted as τ1 and τ2, respectively. In the downlink stage, the FC transmits wireless power-carrying signals to each sensor using wireless power-carrying communication technology. At the sensor end, each sensor is equipped with a power divider, and the signal received by the sensor antenna is first divided into two different signal streams according to a certain power division coefficient by the power divider. The two signal streams will be sent to the energy collector and the information decoder, respectively, and the energy collection and information decoding are performed synchronously. In the uplink stage, the sensor transmits the collected observation values to the FC to complete the information collection work. In the entire workflow, the working energy consumption of the sensor and the communication energy consumption in the uplink process are maintained by the energy collected by the sensor in the downlink process.

[0092] In the downlink stage, the wireless channel from the FC to the kth sensor can be defined as denotes a complex set. According to the reciprocity characteristic of the uplink and downlink communication channels in TDD communication, the channel from the kth sensor to the FC in the uplink process can be directly represented as In an actual system, the channel state information (CSI) obtained through channel estimation and other technologies has an inevitable estimation error from the real channel. The embodiment models the CSI estimation error as a statistical error:

[0093]

[0094] wherein, denotes the estimation result of the CSI, denotes the CSI estimation error, and CN represents a complex Gaussian distribution.

[0095] According to the above channel model, in the downlink process, the signal component received by the kth sensor for energy collection and the component for signal decoding are represented as:

[0096]

[0097] wherein, ρ k is a power division factor, is a symbol sent to the kth sensor, and the power thereof satisfies is a corresponding beamforming vector, n k is the antenna noise of the receiving end, and the distribution thereof obeys

[0098] Based on equation (2), the energy collected by the sensor during the downlink procedure can be expressed as:

[0099]

[0100] where η is the efficiency of the energy harvester. Similarly, based on equation (2), the component used for signal decoding can be expanded as:

[0101]

[0102] The power of each signal component in the equation can be expressed as:

[0103]

[0104] The signal-to-noise ratio of the downlink communication at the sensor end can be expressed as:

[0105]

[0106] Similarly, during the uplink procedure, the signal received by the FC from the kth sensor is:

[0107]

[0108] is the symbol transmitted by the kth sensor, and its power is: is the beamforming vector used by the FC to receive the signal from the sensor, which is normalized as: k || 2 = 1, denotes the noise at the FC's antenna. Based on this, the power of each signal component during the uplink communication procedure can be expressed as:

[0109]

[0110] The signal-to-noise ratio during the uplink phase can be expressed as:

[0111]

[0112] Based on the above analysis, the optimization problem of the present embodiment can be modeled as follows:

[0113]

[0114] where and denote the minimum code rate requirements for the downlink and uplink communication, and represent the maximum outage probability that the system can tolerate, P O,k is the operating energy consumption of the kth sensor.

[0115] Since the outage probability constraint in expression (10) is difficult to solve directly, this embodiment will be approximately transformed to obtain a form of deterministic constraints that can be processed.

[0116] Regarding the processing of the downlink outage probability constraint, first introduce auxiliary variables and Transform the probability constraint into the following form:

[0117]

[0118]

[0119]

[0120] Among them, expression (11) is a convex constraint that can be solved directly, expression (12) realizes the decoupling of variables, and the subsequent non-convex term approximation processing will be performed in the next section, and expression (13) still exists in the form of unsolvable probability. In order to promote the transformation of expression (13), the following variables are defined:

[0121]

[0122] Then expression (13) can be approximated as:

[0123]

[0124] Among them, the trace and Frobenius norm of, and and The expressions of and are as follows:

[0125]

[0126] Through the above expressions, it can be verified that expression (15) and expression (16) are convex constraints that can be processed. In summary, the downlink outage probability constraint can be transformed into expression (11), expression (12), expression (15) and expression (16).

[0127] Similarly, in order to process the uplink outage probability constraint, first introduce and and define the following variables:

[0128]

[0129] Among them, diag(·) represents the operation of constructing a matrix with the elements of the vector as the main diagonal elements, denotes the Kronecker product. Based on the above definitions, the uplink outage probability constraint can be transformed into:

[0130]

[0131] where each term is calculated as follows:

[0132]

[0133] In summary, the uplink outage probability constraint can be transformed into expressions (19)-(22).

[0134] The embodiment eliminates the difficult-to-solve probabilistic form constraint, but the original optimization problem still has the following problems: the optimization objective is still non-convex, and the uplink and downlink time lengths in the sensor energy constraint are still coupled with the remaining variables. First, to solve the problem of non-convex optimization objective, the embodiment uses the Successive Convex Approximation (SCA) method to approximate it. The SCA method is a non-convex optimization method that transforms a non-convex optimization item into its convex approximation item through iterative approximation. Using the SCA method, the optimization objective in expression (10) can be transformed into the following convex approximation objective:

[0135]

[0136] where, represents the τ1 optimization value of the n-th iteration, and other similar superscripts in the following text have similar meanings and will not be repeated. For the sensor energy constraint, the embodiment takes the logarithm of both sides of the inequality, and can obtain the constraint that decouples the uplink and downlink time lengths from the remaining variables:

[0137]

[0138] In summary, the original optimization problem can be restructured in the following form:

[0139]

[0140] where, represents all the variables to be optimized in the new optimization problem. Considering that there are still coupling problems of {W k ,V k}, {ρ k ,P k} and three pairs of variables in expression (26), the embodiment designs an alternating optimization algorithm to solve the optimization problem.

[0141] First, fix {ρ k ,P k} and optimize {W k ,V k} and the remaining auxiliary variables. In the relaxation of {W k ,Vk} using SCA approximation, the final sub-optimization problem can be expressed as:

[0142]

[0143] where, is the set of all optimization variables in this problem. This problem is a fully convex optimization problem and can be solved directly using existing convex optimization algorithms, which will not be discussed in this embodiment. It should be noted that after the convergence of the alternating optimization, the rank-one decomposition of {W k ,V k} is required to obtain the real optimization variables {w k ,v k}. Rank-one decomposition is a mature matrix decomposition method, which will not be discussed here.

[0144] Subsequently, this embodiment fixes {W k ,V k} and optimizes {p k ,P k}. Since {p k ,P k} is not directly related to the original optimization objective, this optimization is a feasible point detection problem. In order to accelerate the convergence speed of the alternating optimization, this embodiment introduces the relaxation variables {a 1,k ,a 2,k ,a 3,k}, and converts the optimization problem into the following form:

[0145]

[0146] where, is the set of all optimization variables in this problem. This problem is a fully convex optimization problem and can be solved directly.

[0147] At the end of each round of alternating optimization, the update of the auxiliary variable needs to be completed. Based on the properties of the quadratic transformation, the update rule can be expressed as:

[0148]

[0149] In summary, the uplink and downlink communication resource joint optimization algorithm for WRSN proposed in this embodiment can be summarized as Figure 2 . As shown in the figure, the algorithm proposed in this embodiment first uses the alternating optimization method to optimize {W k ,V k ,p k ,P kThe optimization is performed on {τ1,τ2}. After the alternating optimization converges, the preliminary optimization result {W} is obtained. k V k ,ρ k ,P k {W} in ,τ1,τ2} k V k} will be further obtained through rank-one decomposition {w k ,v k}. {w k ,v k} and {ρ k ,P k The final optimization result is composed of τ1, τ2}.

[0150] This embodiment will demonstrate the simulation results of the optimization algorithm and other comparative algorithms to verify the effectiveness and advantages of the proposed algorithm.

[0151] First, we introduce the channel model used in this simulation. The channel modeling between the FC and the sensor includes both large-scale fading and small-scale fading. Large-scale fading follows a logarithmic distance path loss model: A(d / d0). -α Where α = 2.2 is the distance loss exponent, and the reference distance d0 and the corresponding reference fading A are set to 1m and -30dB, respectively. Small-scale fading is modeled as a Rayleigh fading model, i.e., following a standard complex Gaussian distribution CN(0,1). Sensors are uniformly and randomly distributed within a circular region with a radius of 1m centered at (20,0)m. Other simulation parameters are shown in Table 1.

[0152] Table 1 Simulation Parameters

[0153]

[0154] In addition to simulating the algorithm proposed in this embodiment, this embodiment also simulates a comparative algorithm to objectively evaluate the performance of the proposed algorithm and verify its effectiveness: This comparative algorithm does not optimize the uplink and downlink times, and the duration is fixed at τ1 = τ2 = 10ms. Since the duration is fixed, the optimization objective is modified to minimize the downlink transmit power, and the other optimization constraints are the same as those of the proposed algorithm.

[0155] First, this embodiment verifies the iterative convergence characteristics of the two algorithms through simulation experiments. For example... Figure 3 As shown, both algorithms converge within 10 iterations. The comparative algorithm converges slightly faster than the proposed algorithm because it does not consider the duration of uplink and downlink communication during the alternating optimization process, thus reducing the number of optimization variables and leading to faster convergence. However, precisely because of this, the comparative algorithm's optimization performance in terms of total system energy consumption is inferior to the proposed algorithm, indicating that the proposed algorithm has superior performance in energy consumption optimization.

[0156] Subsequently, this embodiment compares the performance of the two algorithms under different numbers of FC antennas. For example... Figure 4 As shown, the total system power consumption obtained by both optimization algorithms gradually decreases with the increase of the number of antennas. This phenomenon indicates that increasing the number of antennas at the FC end can effectively improve signal transmission efficiency, allowing the optimization algorithm to further reduce the total system power consumption while satisfying various constraints. Furthermore, the decreasing trend of total power consumption gradually slows down with the increase of the number of antennas, indicating that there is an upper limit to the improvement in signal gain from multiple antennas. At the same time, increasing the number of antennas also increases the computational complexity of the optimization algorithm. Therefore, in practical system design, a balance should be struck between antenna configuration and optimization performance, and the number of antennas should be rationally selected to achieve a balance between power consumption optimization and computational complexity.

[0157] at last, Figure 5 Simulation results of the two algorithms under different outage probability tolerances are presented. Outage probability tolerance reflects the stringency of communication quality requirements; a higher tolerance means a greater acceptable outage probability, thus allowing for lower transmit power. As shown in the figure, the total system energy consumption optimized by both algorithms decreases with increasing outage probability tolerance, which is consistent with theoretical analysis. Furthermore, under all set outage probability tolerances, the optimized energy consumption of the proposed algorithm is consistently lower than that of the comparative algorithm, further validating its effectiveness and superiority in energy consumption optimization.

[0158] This embodiment proposes a joint optimization method for communication resources in wireless charging sensor networks. Considering both downlink wireless power-carrying communication and uplink pure communication processes in a wireless rechargeable sensor network, a sensor energy consumption model is established that serializes the uplink and downlink processes, and a joint optimization model for uplink and downlink communication resources is constructed. This model aims to minimize the total system energy consumption while ensuring both uplink and downlink communication quality and downlink energy transmission quality. It jointly optimizes uplink and downlink time allocation, transmit / receive beamforming at the fusion center, and sensor power allocation.

[0159] To solve this joint optimization problem, this embodiment first processes the interruption probability constraint and other non-convex constraints to decouple some optimization variables and performs convex approximation on the non-convex terms. This embodiment proposes an alternating optimization algorithm to optimize the fusion center and sensor ends in the joint model separately, achieving efficient solution. Simulation results provided in this embodiment have verified the effectiveness of the optimization algorithm, demonstrating its superior performance in reducing the total system energy consumption.

[0160] It is understood that the above embodiments are merely for illustrating the principles of the present invention.

[0161] The exemplary embodiments used are not limited thereto. For those in the art...

[0162] to one of ordinary skill in the art, without departing from the spirit and essential characteristics of the present application.

[0163] various modifications and improvements can be made without departing from the spirit and scope of the application.

Claims

1. A method for joint optimization of communication resources for a wireless charging sensor network, the method comprising: Comprising: The expression (10) of the optimization model is obtained as follows: where w k denotes the beamforming vector for downlink at the fusion center, v k denotes the beamforming vector for uplink at the fusion center, p k denotes the power splitting factor of the sensor, P k denotes the uplink power of the sensor, τ1 denotes the time for downlink, τ2 denotes the time for uplink, Pr(·) denotes the probability of an event, denotes the signal-to-interference-and-noise ratio (SINR) for downlink, denotes the SINR for uplink, denotes the minimum code rate requirement for downlink communication, denotes the minimum code rate requirement for uplink communication, and denotes the maximum outage probability that the system can tolerate, P O,k is the operational energy consumption of the kth sensor, and η is the energy harvesting efficiency of the sensor, is the estimated value of the channel between the kth sensor and the fusion center, T is the total time limit; The expression (10) is transformed into expression (24) using a continuous convex approximation method as follows: wherein, τ1opt, n represents the optimized value of τ1 for the n-th iteration, Pc,n is the fusion center downlink power for the n-th iteration. The optimization problem is reconstructed to obtain expression (26) as follows: wherein Rank( ) represents the rank of a matrix, and Tr( ) represents the trace of a matrix. The optimization problem is solved using an alternating optimization algorithm to obtain expression (27) as follows: wherein, is the set of all variables to be optimized, and are introduced auxiliary variables; Introducing the slack variables {α 1,k ,α 2,k ,α 3,k}, the optimization problem is transformed into the expression (28) as follows: wherein, is the set of all optimization variables, and are introduced auxiliary variables.

2. The method of claim 1, wherein, Also comprising: According to the nature of the quadratic transformation, the auxiliary variable is updated The expression is as follows: wherein is the noise power in the downlink process, is the noise power in the uplink process, is the estimation error of the channel between the ith sensor and the fusion center, 3. The method of claim 2, wherein, The step of transforming the expression (10) into expression (24) using a continuous convex approximation method further comprises: Taking the logarithm of both sides of the inequality of expression (24), the constraint expression of the uplink and downlink time length decoupling from the remaining variables is obtained as follows: wherein, p k denotes the power splitting factor of the sensor, P k denotes the uplink power of the sensor, τ1 denotes the time of the downlink, τ2 denotes the time of the uplink, P O,k is the working energy consumption of the kth sensor, η is the energy harvesting efficiency of the sensor, is the estimated value of the channel between the kth sensor and the fusion center, 4. The method of claim 3, wherein, The step of obtaining the expression (10) of the optimization model further comprises: Introducing auxiliary variables and The probability constraints are transformed into expressions (11-13) as follows: wherein τ1 represents the time of the downlink, τ2 represents the time of the uplink, Pr(·) represents the probability of an event, denotes the signal-to-interference-plus-noise ratio of the downlink, denotes the signal-to-interference-plus-noise ratio of the uplink, denotes the maximum outage probability tolerable by the system, denotes the power of the signal, denotes the power of the interference, denotes the power of the noise, is the noise power in the downlink, The following variables are defined: where w k is the downlink beamforming vector of the fusion center, Ah k is the random channel estimation error, p k is the sensor power splitting factor, According to the variable expression (14), the expression (13) is approximated to the following expressions (15-16): wherein is an introduced auxiliary variable, an expression for the trace of an expression for the Frobenius norm of an expression for an expression for is as follows:

5. The method of claim 4, wherein, Also comprising: Introducing variables and Also define the following variables: where diag(·) denotes the operation of constructing a matrix with the vector elements as the main diagonal elements, denotes the Kronecker product, v k is an uplink beamforming vector, P k is a sensor uplink transmit power, is the variance of the channel estimation error; According to expression (18), the uplink interruption probability constraint is transformed into the following expressions (19-22): wherein, and are introduced auxiliary variables, τ2 represents the time of uplink, represents the lowest code rate requirement of downlink communication; The calculation of each term in expressions (19-22) is as follows: where Pr(·) denotes the probability of an event, is an estimate of the channel between the kth sensor and the fusion center, denotes the maximum outage probability that the system can tolerate.

6. The method of claim 5, wherein, The step of obtaining the expression (10) of the optimization model further comprises: The channel state information estimation error is modeled as a statistical error, and the expression is as follows: wherein the channel from the fusion center to the kth sensor in the downlink is denotes the set of complex numbers, and the channel from the kth sensor to the fusion center in the uplink is denotes the estimate of the channel state information, denotes the estimation error of the channel state information, and CN represents a complex Gaussian distribution; The signal component received by the kth sensor for energy collection and the component for signal decoding in the downlink process are respectively as follows: wherein ρ k is a power division factor, is an actual channel, n k is a noise of a downlink process, is a symbol sent to the kth sensor, and the power satisfies is is a corresponding beamforming vector, n k is an antenna noise of a receiving end, and the distribution obeys 7. The method of claim 6, wherein, Also comprising: According to expression (2), the energy collected by the sensor in the downlink process is expressed as follows: wherein, is an estimated value of the channel between the kth sensor and the fusion center, η is the energy harvesting efficiency of the sensor, w k is a downlink beamforming vector of the fusion center, τ1 represents the time of the downlink; According to expression (2), the component for signal decoding is expanded as follows: where Δh k is the random error of the channel estimate, where the power of each signal component is given by: The signal-to-noise ratio of the sensor downlink communication is expressed as:

8. The method of claim 7, wherein, Also comprising: The signal of the kth sensor received by the fusion center in the uplink process is: wherein is an estimate of the channel between the kth sensor and the fusion center, is the symbol transmitted by the kth sensor with power is a beamforming vector used by the fusion center to receive the kth sensor signal, said beamforming vector being normalized as follows: ||v k || 2 = 1, denotes the antenna noise at the fusion center.

9. The method of claim 8, wherein, Also comprising: The power of each signal component in the uplink process is expressed as: wherein v k is an uplink beamforming vector, is an estimate of the channel between the kth sensor and the fusion center, Ah k is a random error in the channel estimate, P k is the sensor uplink transmit power, is the noise power of the uplink process; The signal-to-noise ratio in the uplink stage is expressed as:

Images