A large-scale access energy saving optimization method, device and electronic equipment

Abstract

The application discloses a large-scale access energy-saving optimization method and device and electronic equipment. The method comprises the following steps: sending a plurality of response configuration information to a plurality of MTDs, so that each MTD sends uplink control information according to the response configuration information; defining a clustering access definition optimization problem of the uplink control information according to each uplink control information; smoothing the clustering access optimization definition problem by using a spectral radius gradient solution method, so as to strategically group a plurality of MTDs according to a plurality of uplink control information and obtain a strategic grouping scheme; converting the clustering access optimization definition problem into an eigenvalue maximization problem of a correlation matrix, and calculating a feasible solution in an augmented form by introducing an augmented variable, so as to realize a large-scale access energy-saving optimization method with high energy efficiency, low power consumption, small time delay and low operation cost by jointly optimizing discontinuous reception cycles and timing parameters corresponding to each strategic grouping scheme through the eigenvalue maximization problem.

Description

TECHNICAL FIELD

[0001] The present application belongs to the technical field of communication, and particularly relates to a large-scale access energy optimization method and device and electronic equipment. BACKGROUND

[0002] In a large-scale Internet of Things deployment, such as massive Machine-Type Communications (mMTC) in the Fifth Generation (5G) network, a large number of devices need to access the network at the same time. These devices usually have low power consumption requirements and do not need frequent data transmission. Therefore, how to effectively manage the access of these devices while reducing power consumption and shortening the wake-up delay has become an important research topic. Uplink control information is a key component of communication between the network and the terminal, and the scheduling request in it is particularly important for reducing signaling overhead and power consumption. The discontinuous reception mechanism allows the terminal to enter a sleep state during the inactive period to save power, but frequent state switching can also cause a longer wake-up delay. However, existing research only focuses on uplink control information transmission or only focuses on the discontinuous reception mechanism, and lacks a unified framework and joint optimization trade-off method that comprehensively considers the interaction between the two.

[0003] Current large-scale access energy optimization methods are mainly divided into two categories: resource allocation optimization methods based on control signaling mechanisms and multiple access technologies, and discontinuous reception parameter configuration methods based on model prediction. Resource allocation optimization methods based on control signaling mechanisms and multiple access technologies require compression of control information and more accurate synchronization mechanisms, which not only increases the processing complexity of the terminal, but also introduces additional error code risk, affecting the quality of data transmission. Discontinuous reception parameter configuration methods based on model prediction rely on historical observation priors, affecting the effectiveness of parameter configuration and having poor robustness.

[0004] Firstly, the resource allocation optimization method based on control signaling mechanism and multiple access technology mainly reduces the number of bits of control information through compression coding technology, and improves the spectrum efficiency by using more efficient multiple access technology to reduce the number of resource blocks required for control information transmission, thereby reducing the transmission delay and power consumption. B. Li et al. proposed an efficient and low-overhead uplink scheduling algorithm suitable for large-scale Internet of Things applications, described a capacity outer bound under sampling constraints, which only allows a small part of terminals to use control channels for system state reporting at a time, and realizes the capacity outer bound by using the prior information of channel state distribution and instantaneous queue length. The method meets the sampling constraints while reducing the throughput loss. A. Ramadan et al. proposed a hybrid uplink scheduler based on non-orthogonal multiple access and orthogonal multiple access, which can optimize the system under the clustering constraints of heterogeneous network targets and overload scenarios, and allows part of the devices to share the same resources in frequency and time to improve the spectrum and scheduling efficiency of the network. S. Lv et al. studied the energy-saving and secure packet transmission of non-orthogonal multiple access assisted large-scale access network, and developed a decentralized stateless multi-agent reinforcement learning algorithm, which can improve the energy efficiency and reduce the information exchange overhead in combination with non-orthogonal multiple access relay. Y. Meng et al. studied the grant-free non-orthogonal multiple access integrated learning framework with power control scheme in large-scale access network, which aims to minimize the long-term system power consumption under the condition of meeting the packet error rate constraint, and includes two interrelated sub-algorithms to optimize the power level and competitive transmission unit allocation.

[0005] Secondly, the model prediction-based discontinuous reception parameter configuration method dynamically adjusts the discontinuous reception parameters based on the device's historical behavior and prediction model, allowing the device to reduce sleep time during high data transmission demand and increase sleep time during low data transmission demand to adapt to the power consumption requirements in different application scenarios. N. Mysore Balasubramanya et al. proposed an improved discontinuous reception parameter configuration method, which introduced a fast sleep mode and a wake-up signal, allowing the device to quickly enter sleep state when there is no data transmission task, and quickly wake up when needed, effectively improving the energy efficiency of Internet of Things devices. H. W. Ferng et al. proposed a discontinuous reception configuration scheme with adaptive parameters, allowing device terminals to impose constraints on delay or power consumption, and the base station can estimate the packet arrival rate in an iterative manner and find the feasible parameters of the device terminal to meet the constraints with acceptable convergence speed. J. Song et al. proposed a multi-agent deep reinforcement learning method with integrated self-attention mechanism to reduce device terminal power consumption, allowing the model to optimize discontinuous reception parameters independently of the base station's scheduling algorithm, enhancing its adaptability to dynamically changing network conditions. J. Zhou et al. proposed an online learning-based discontinuous reception parameter configuration mechanism to improve the service energy efficiency of Internet of Things devices by adapting to different traffic patterns. In the proposed mechanism, time is divided into multiple time intervals, and the discontinuous reception cycle is adjusted by learning the traffic statistics at the beginning of each time interval. To speed up the learning process, the symmetric sampling method is used, which significantly improves the delay and energy efficiency. J. Wu et al. proposed a machine learning algorithm based on an expert library to predict discontinuous reception parameter configuration, which uses only the observed packet arrival time to predict the next packet arrival time without making any assumptions about the underlying traffic process, and dynamically adjusts the sleep interval using the predicted next packet arrival time. Since different Internet of Things services have different requirements for energy saving and data transmission delay, a loss function is designed by combining these two performance indicators, allowing the discontinuous reception parameter configuration to support a wide range of application services.

[0006] First, in large-scale access scenarios, the implementation complexity and cost of resource allocation optimization methods based on control signaling mechanisms and multiple access technologies are very high. The compression and optimization of control information will dramatically increase the processing complexity of terminals, leading to an increase in the cost of terminal devices and potentially affecting the reliability and stability of the devices. At the same time, more efficient multiple access technologies require new hardware or software support, which will also increase the implementation cost and difficulty. On the other hand, the compression encoding of control information may introduce additional error risk. If the compression encoding is not proper, it will increase the error rate of control information, and non-orthogonal multiple access technology will also increase the interference between devices. In the case of dense deployment, more interference may occur between devices, which will affect the reliability and quality of data transmission.

[0007] Secondly, for large-scale access energy saving optimization, the model prediction based discontinuous reception parameter configuration method needs a large amount of historical data for training and prediction, and it usually needs a large amount of data and computing resources to train an accurate prediction model, and it is very complex to collect and process these data in a large-scale network environment, and the quality and integrity of the data directly affect the accuracy of the prediction model. In addition, the model prediction based method needs a certain time to collect data, train the model and update the discontinuous reception parameters, and different devices and networks may have different behavior patterns, and as the network environment and device behavior change, the prediction model needs to be updated regularly, which greatly affects the energy efficiency and delay of implementation.

[0008] Therefore, if a large-scale access energy saving optimization method with high energy efficiency, low power consumption, small time delay and low operation cost is provided, it becomes an important problem. SUMMARY

[0009] In order to solve the above problems existing in the prior art, the present application provides a large-scale access energy saving optimization method, device and electronic equipment.

[0010] The technical problem to be solved by the present application is solved by the following technical scheme:

[0011] In a first aspect, the present application provides a large-scale access energy saving optimization method, which comprises:

[0012] sending a plurality of response configuration information to a plurality of MTDs, so that each MTD sends uplink control information according to the response configuration information; the response configuration information is determined according to the task request sent by each MTD;

[0013] defining a clustering access definition optimization problem according to each uplink control information, and smoothing the clustering access optimization definition problem by using a spectral radius gradient solution method, so as to perform strategic grouping on the plurality of MTDs according to the plurality of uplink control information, and obtain a strategic grouping scheme; the clustering access definition optimization problem takes minimizing each element in a similarity metric matrix in the strategic grouping scheme as an optimization objective; each element in the similarity metric matrix represents the difference between two observation variables in the uplink control information;

[0014] transforming the clustering access optimization definition problem into an eigenvalue maximization problem of an association matrix, and substituting a feasible solution in an augmented form into the eigenvalue maximization problem to realize large-scale access energy saving optimization by jointly optimizing discontinuous reception cycles and timing parameters corresponding to each strategic grouping scheme; the feasible solution in the augmented form is calculated by introducing an augmented variable; each element in the association matrix is determined according to the number of groups in the strategic grouping.

[0015] Optionally, the smoothing of the cluster access optimization definition problem by using the spectral radius gradient solving method comprises:

[0016] The smoothing of the cluster access optimization definition problem is achieved by minimizing the spectral radius gradient of the association matrix in the cluster access optimization definition problem and setting a trace penalty term in the cluster access optimization definition problem; the trace penalty term is set according to the cumulative observation number of the uplink control information.

[0017] Optionally, the cluster access optimization definition problem is:

[0018]

[0019] wherein S represents the similarity measure matrix; ρ represents the trace penalty term; I represents the unit matrix; A represents the association matrix; 1 represents the all-one column vector, and A1 = 1 represents that the sum of the elements of each row of A is equal to 1; <·, ·> represents the inner product operation of the norm; the element in the p-th row and the q-th column of S is defined as represents the observed variable in the uplink control information.

[0020] Optionally, the objective function of the cluster access optimization definition problem is:

[0021]

[0022] wherein f κ,Θ (Λ) represents the objective function; κ represents the smoothing parameter; λ j (Θ -1 / 2 ΛΘ -1 / 2 ) represents the j-th eigenvalue obtained after the eigenvalue decomposition of Θ -1 / 2 ΛΘ -1 / 2 ; Λ represents the diagonal matrix with λ j (Θ -1 / 2 ΛΘ -1 / 2 ) as the diagonal elements; Θ represents the feasible solution.

[0023] Optionally, the augmented form of the feasible solution comprises Θ' and A0':

[0024]

[0025] wherein Θ' represents the augmented form of Θ; A0' represents the augmented form of A0; Θ and A0 are both given feasible solutions; dvec(·) represents the operation of mapping the matrix into a diagonal matrix.

[0026] In a second aspect, the present application provides a large-scale access energy saving optimization device, which comprises:

[0027] The uplink control information obtaining module is configured to send a plurality of response configuration information to the plurality of MTDs, so that each MTD sends uplink control information according to the response configuration information; and the response configuration information is determined according to a task request sent by each MTD.

[0028] The strategic grouping scheme obtaining module is configured to define a clustering access definition optimization problem of the uplink control information according to each uplink control information, and to smooth the clustering access definition optimization problem by using a spectral radius gradient solution method, so as to perform strategic grouping on the plurality of MTDs according to the plurality of uplink control information, and obtain a strategic grouping scheme; each element in a similarity metric matrix in the strategic grouping scheme is used as an optimization objective to minimize; and each element in the similarity metric matrix represents a difference between two observation variables in the uplink control information.

[0029] The large-scale access energy saving optimization implementing module is configured to convert the clustering access definition optimization problem into an eigenvalue maximization problem, and to substitute a feasible solution in an augmented form into the eigenvalue maximization problem to realize large-scale access energy saving optimization by jointly optimizing discontinuous reception cycles and timing parameters corresponding to each strategic grouping scheme; the feasible solution in the augmented form is calculated by introducing an augmented variable; and each element in the association matrix is determined according to a grouping number in the strategic grouping.

[0030] Optionally, the strategic grouping scheme obtaining module smoothes the clustering access definition optimization problem by using the spectral radius gradient solution method, and includes the following steps:

[0031] The clustering access definition optimization problem is smoothed by minimizing a spectral radius gradient of an association matrix in the clustering access definition optimization problem, and setting a trace penalty term in the clustering access definition optimization problem; and the trace penalty term is set according to a cumulative observation number of the uplink control information.

[0032] Optionally, the clustering access definition optimization problem is as follows:

[0033]

[0034] wherein, S represents the similarity metric matrix; ρ represents the trace penalty term; I represents a unit matrix; A represents the association matrix; 1 represents a full one column vector, and A1=1 represents that a sum of elements in each row of A is equal to 1; and <·,·> represents an inner product operation of a norm; an element in the pth row and the qth column in S is defined as represents an observation variable in the uplink control information.

[0035] Optionally, an objective function of the clustering access definition optimization problem is as follows:

[0036]

[0037] where f κ,Θ (A) represents the objective function; k represents a smoothing parameter; l j (Θ -1 / 2 ΛΘ -1 / 2 ) represents the jth eigenvalue obtained after eigenvalue decomposition of Θ -1 / 2 ΛΘ -1 / 2 ; A represents a diagonal matrix with diagonal elements being l j (Θ -1 / 2 ΛΘ -1 / 2 ); and Θ represents a feasible solution.

[0038] Optionally, the augmented form of the feasible solution comprises Θ' and A0':

[0039]

[0040] where Θ' represents the augmented form of Θ; A0' represents the augmented form of A0; Θ and A0 are both given feasible solutions; and dvec(·) represents an operation of mapping a matrix into a diagonal matrix.

[0041] In a third aspect, the present application provides an electronic device, comprising a processor, a communication interface, a memory and a communication bus, wherein the processor, the communication interface and the memory complete mutual communication through the communication bus;

[0042] The memory is used for storing a computer program.

[0043] The processor is used for executing the program stored on the memory, and realizes the method steps of any one of the large-scale access energy saving optimization methods.

[0044] In a fourth aspect, the present application provides a computer readable storage medium, wherein the computer readable storage medium stores a computer program, and the computer program is executed by a processor to realize the method steps of any one of the large-scale access energy saving optimization methods.

[0045] The application provides a large-scale access energy saving optimization method, which converts a cluster access optimization definition problem of uplink control information into a characteristic value maximization problem of an associated matrix determined according to the grouping quantity in a strategic grouping of each element, and substitutes a feasible solution in an augmented form into the characteristic value maximization problem to reduce unnecessary signaling interaction in the case of large-scale MTD access, and further reduce system complexity and operation cost by jointly optimizing discontinuous reception cycles and timing parameters corresponding to each strategic grouping scheme. According to the characteristics that a large number of devices are in a low activity state for a long time in an application scene such as the Internet of Things, the cluster access optimization definition problem is defined to minimize each element in a similarity metric matrix in a strategic grouping scheme as an optimization target to reasonably configure discontinuous reception cycles, reduce unnecessary listening and transmission operations of devices, and thus significantly reduce the energy consumption of terminal devices and prolong the battery life. While ensuring low power consumption, the joint optimization of discontinuous reception cycles and timing parameters corresponding to each strategic grouping scheme can also minimize the conversion time of devices from a sleep state to an active state, that is, shorten the wake-up delay.

[0046] The application will be further described in detail below with reference to the drawings and the application. BRIEF DESCRIPTION OF DRAWINGS

[0047] Figure 1 is a flowchart of a large-scale access energy saving optimization method provided by an embodiment of the application;

[0048] Figure 2 is a schematic diagram of an mMTC network model of typical uplink and downlink signaling and data interaction provided by an embodiment of the application;

[0049] Fig. 3(a) and Fig. 3(b) are schematic diagrams of energy saving ratios and average wake-up delays under different MTD quantities;

[0050] Fig. 4(a) and Fig. 4(b) are schematic diagrams of energy saving ratios and average wake-up delays under different cumulative observation quantities;

[0051] Figure 5 is a structural schematic diagram of a large-scale access energy saving optimization device provided by an embodiment of the application;

[0052] Figure 6 is a structural schematic diagram of an electronic device provided by an embodiment of the application. DETAILED DESCRIPTION

[0053] The application will be further described in detail below with reference to the drawings and the application.

[0054] In order to solve the technical problems of low energy efficiency, high power consumption, long delay and high operation cost in the prior art large-scale access energy saving optimization method, an embodiment of the application provides a large-scale access energy saving optimization method, as shown inFigure 1 , Figure 1 is a flowchart of a large-scale access energy optimization method provided by an embodiment of the application, and specifically includes the following steps:

[0055] In step S101, multiple response configuration information is sent to multiple MTDs, so that each MTD sends uplink control information according to the response configuration information; the response configuration information is determined according to a task request sent by each MTD.

[0056] In an embodiment of the application, an MTD (Machine Type Device) includes a sensor type device, a smart metering device, an industrial Internet of Things device, and a smart city device, etc. In a 5G large-scale access scenario, these devices are collectively referred to as MTDs.

[0057] Referring to Figure 2 , Figure 2 is a schematic diagram of a mMTC network model of typical uplink and downlink signaling and data interaction provided by an embodiment of the application. A base station performs resource scheduling of machine type devices according to uplink control information transmitted by a physical uplink control channel (PUCCH), and then transmits downlink shared data to activated MTDs through a physical downlink control channel (PDCCH). An activated MTD refers to an MTD whose sent task request is responded to by the base station. Before an MTD transmits uplink control information, a task request needs to be sent to the base station, and the base station will determine response configuration information of uplink control information physical resource element allocation for the MTD according to the task request, and finally send the response configuration information containing the above allocation information to indicate the MTD activation state. The MTD includes activated devices, i.e., activated MTDs, and silent devices, i.e., unactivated MTDs. The silent state refers to a state in which the MTD does not communicate, and the transmission state refers to a state in which the MTD communicates and transmits.

[0058] In an embodiment of the application, after receiving the response configuration information, the MTD sends uplink control information to the base station, and the specific process is as follows:

[0059] The uplink control information first converts the scheduling information bits carried by the four-phase shift keying modulation scheme into a symbol vector b = [b1, b2,..., bK], where K is the number of scheduling information bits, and b1, b2,..., bK are the scheduling information bits. E], where E is the total number of bits after modulation, and each modulation symbol carries 2 bits of information. Within the allocated resource block, specific physical resource elements are further allocated, and each resource element corresponds to the position of a subcarrier on an Orthogonal Frequency Division Multiplexing (OFDM) symbol. The modulated symbol vector b = [b1, b2, ..., b E ] is mapped to the allocated physical resource elements to form a signal vector u=[u1,u2,...,u M ], where M is the number of resource elements allocated for uplink control information, ensuring that each symbol is correctly placed on the designated resource element. Next, symbols are mapped in the frequency and time domains. In the frequency domain, symbols are assigned to specific subcarriers, and in the time domain, symbols are assigned to specific OFDM symbols. This ensures that the uplink control information is distributed across time and frequency to maximize transmission diversity and reliability.

[0060] After the uplink control information is configured through the allocated uplink resource blocks, it is sent back to the base station by the MTD. Figure 2 As shown in (a) of Figure 1, the base station adjusts the resource allocation strategy based on the received uplink control information to optimize network performance. Considering the influence of multipath propagation and fading, the uplink control information y received at frequency f and time t is f,t It can be expressed as:

[0061]

[0062] Among them, α i p t represents the probabilistic activation state of the i-th MTD at time t, where if the i-th MTD is in the active state, then α i (t)=1, otherwise represents the channel frequency response of the i-th MTD at frequency f; N f is the total number of frequency bins corresponding to the discrete Fourier transform size; represents the frequency domain representation of the uplink control information u sent by the i-th MTD; f,t represents the Gaussian white noise at the base station receiving side; i=1,2,...,N p ; N p represents the total number of MTDs; l = 1, 2, ..., L; L represents the length of the channel pulse; j' represents the imaginary unit.

[0063] Step S102, defining a cluster access definition optimization problem for uplink control information according to each uplink control information, and using a spectral radius gradient solution method to smooth the cluster access optimization definition problem, so as to strategically group multiple MTDs according to multiple uplink control information to obtain a strategic grouping scheme; the cluster access definition optimization problem takes minimizing each element in a similarity measure matrix in the strategic grouping scheme as an optimization goal; each element in the similarity measure matrix represents the difference between two observed variables in the uplink control information.

[0064] In the embodiment of the present invention, after receiving the uplink control information, the base station side selects a specific scheduling time slot T using PUCCH format 2 in the typical time slot structure of the uplink control information containing the scheduling request. SR The uplink control information is collected and converted into an equivalent vector, called the uplink control information vector, denoted as y f =(y f,1 ,...y f,TSR ) T , T represents the transpose of the vector. Inspired by the semidefinite programming clustering method, such as Figure 2 As shown in (a), the present invention uses the uplink control information vector y f Reconstruct the cluster quantity criterion in and transform it into an equivalent real variable That is, the observed variables, so that the embodiment of the present invention can dynamically adjust Figure 2 Parameter configuration of the discontinuous reception cycle in (b).

[0065] In one implementation, the cluster access definition optimization problem is to minimize each element in the similarity metric matrix in the strategic grouping scheme as the optimization goal. The cluster access definition optimization problem for defining uplink control information based on each uplink control information in mMTC is:

[0066]

[0067] Where S represents the similarity measure matrix; ρ represents the trace penalty term; I represents the identity matrix; A represents the association matrix; 1 represents the all-one matrix; <·, ·> represents the inner product operation of the norm, that is, Tr((-S-ρI) T A), Tr(·) is the trace of the matrix; S represents the similarity measure matrix, where each element of S is defined as

[0068] In the actual mMTD scenario, the observed variables can be assigned to K independent and unidentified cluster sets, i.e. strategic groupings, where C = {C1,...,C k ,...,C K}. The cluster resource identification assignment for indicating discontinuous reception timer configuration can be characterized by an association matrix A, where p,q∈C k . Since the number of clusters K is unknown, it is necessary to adjust the trace penalty term ρ according to the cumulative observation number N c , i.e. the number of times of changing frequency point f. The trace penalty term ρ is set as ρ = N / N0, where N0 is a constant. The association matrix A is an optimization variable, which can be used to adaptively select the number of clusters and find the optimal cluster at the same time.

[0069] In the embodiments of the present application, based on the cluster access of uplink control information, the discontinuous reception management is further adjusted, as shown in (b) in the figure. Figure 2 The discontinuous reception cycle length includes long and short cycles, which is crucial for energy saving. The uplink control information can dynamically adjust the discontinuous reception cycle length by providing timely scheduling hints. The uplink control information can implicitly affect the setting of discontinuous reception related timers, such as the duration of timers, inactivity timers and discontinuous reception retransmission timers, according to the association matrix. By coordinating these timers with the traffic pattern inferred from the uplink control information, mMTC networks can further optimize power consumption without sacrificing quality of service. The traffic pattern can be inferred by analyzing the uplink control information. The uplink control information contains a lot of information, such as feedback, scheduling request, channel quality indication, etc. Through these information, the base station side can understand the activity and demand of each connected device; periodic traffic pattern is characterized by data transmission occurring at fixed or nearly fixed intervals, such as sensors periodically sending environmental monitoring data or smart meters uploading power consumption information; bursty traffic pattern is irregular data transmission triggered by specific events, such as fire alarm system sending alarm only when detecting fire, smart home devices executing operation according to user instruction; persistent traffic pattern refers to long continuous data stream, which may be accompanied by short intermittent period, such as video monitoring camera continuously uploading video stream or real-time status feedback on industrial automation production line; mixed traffic pattern combines the characteristics of two or more of the above modes, such as vehicle location tracking in intelligent transportation system, which has both periodic location update and bursty data transmission due to special events such as traffic accident.

[0070] In addition, the Medium Access Control Element (MAC CE) will carry the instruction to start the discontinuous reception cycle based on the received uplink control information type, so as to ensure that the device enters the discontinuous reception state at the best time according to the indication of the network control plane.

[0071] Based on (P1), the non-continuous receiving cycle parameter can be further optimized by a smooth eigenvalue maximization method with trace penalty ρ, the uplink control information with scheduling request is the driving signaling of adaptive non-continuous receiving management, through the balance between energy efficiency and communication response, the overall performance of mMTC network can be enhanced.

[0072] In an implementation manner, the clustering access optimization definition problem is smoothed by using a spectral radius gradient solving method, and the smoothing processing includes:

[0073] The clustering access optimization definition problem is smoothed by minimizing the spectral radius gradient of the correlation matrix in the clustering access optimization definition problem, and setting a trace penalty term in the clustering access optimization definition problem.

[0074] Firstly, in order to solve the inconsistency clustering problem of uplink control information in the mMTC high-dimensional environment, a specific gradient solving scheme is proposed for (P1) by minimizing the spectral radius gradient of the correlation matrix A. When the number of clusters K is unknown, the trace penalty term ρ is determined by a data-driven tuning parameter. By replacing the trace constraint in the optimization problem with a properly selected trace penalty term ρ based on a double certificate construction (the double certificate construction is a dual solution construction of the semidefinite programming relaxation), a smooth approximation can be applied to configure the timer start of the MAC CE to activate the discontinuous reception, so as to convert (P1) into a standard semidefinite regular programming:

[0075]

[0076] Wherein, Φ t is a symmetric matrix of (N c N f )×(N c N f ), φ t represents an element in a set of real numbers, and the index t ranges from 1 to T SR . When the smooth approximation is applied, a strict feasible solution Θ of the problem (P2) must be specified as input. Given Θ, the projection of Θ on the boundary of the positive semidefinite cone can be defined as:

[0077]

[0078] Wherein, λ min (Θ -1 / 2 AΘ -1 / 2 ) represents the mapping of Θ -1 / 2 AΘ -1 / 2 to its minimum eigenvalue, corresponding to the intersection of the projection normal from Θ to A and the positive semidefinite cone.

[0079] Then, for s0 satisfying s0 << S, Θ, the semi-definite programming problem (P2) can be formulated as a specific eigenvalue optimization problem, where the constraint is only a linear equation. If Λ * is the optimal solution of problem (P3):

[0080]

[0081] where <S, Λ> represents the inner product operation between S and the characteristic diagonal matrix Λ to be optimized; s0 represents the inner product threshold between S and the feasible solution; represents the constraint set containing <S, Λ> = s0;

[0082] Then is the optimal solution of problem (P2). In addition, if A * constitutes the optimal solution of problem (P2), then the optimal solution of problem (P3) can be expressed as:

[0083]

[0084] where s * represents the minimum value of <S, A>.

[0085] In an implementation manner, in order to improve the convergence speed, the objective function in problem (P3) can be replaced by a smooth approximation, and finally the objective function of the defined problem of clustering access optimization includes:

[0086]

[0087] where f κ,Θ (Λ) represents the objective function; κ represents a smooth parameter; λ j (Θ -1 / 2 ΛΘ -1 / 2 ) represents the jth eigenvalue obtained after eigenvalue decomposition of the matrix Θ -1 / 2 ΛΘ -1 / 2 ; Λ represents a diagonal matrix whose diagonal elements are eigenvalues λ j (Θ -1 / 2 ΛΘ -1 / 2 ); Θ represents a feasible solution.

[0088] Next, the projection gradient descent is used to solve The number of repetitions is represented by τ, and satisfies the following condition:

[0089]

[0090] where R represents the Euclidean sphere radius, so as to reach a small enough tolerance level ε > 0 within the range of R, that is, the correlation matrix in the uplink control information resource allocation after clustering by smooth approximation is expressed as A τmay be expressed as:

[0091]

[0092] In addition, in order to improve the efficiency of the gradient The processing of projecting the gradient onto the space may be expressed as where the space is composed of all the correlation matrices A satisfying <Φ t , A> = 0 and <S, A> = 0. The above projection matrix extension can be used to process the case of unknown number K of clusters, and the mMTC network can strategically group the MTDs with similar uplink control information transmission modes.

[0093] In step S103, the clustering access optimization definition problem is converted into an eigenvalue maximization problem of the correlation matrix, and a feasible solution in an augmented form is substituted into the eigenvalue maximization problem to realize the large-scale access energy saving optimization by jointly optimizing the discontinuous reception cycle and timing parameters corresponding to each strategic grouping scheme; the feasible solution in the augmented form is calculated by introducing an augmented variable; each element in the correlation matrix is determined according to the number of groups in the strategic grouping.

[0094] In the embodiment of the present application, according to the strategic grouping scheme, the optimal solution of the smoothing problem (P3) can be further extended to the original problem (P1), that is, without knowing K in advance, while maintaining the clustering separation rate that can be achieved when K is known. On this basis, in order to optimize the discontinuous reception cycle and timing parameters accordingly, the embodiment of the present application introduces the following augmented variables, given a strict feasible solution Θ and A0 of (P1), such that <-S, Θ> < <-S, A0> = -s0:

[0095]

[0096] Wherein, Θ' represents the augmented form of Θ; A0' represents the augmented form of A0; dvec(·) represents the operation of mapping a matrix into a diagonal matrix, and the elements on the diagonal are the vectorized form of the original matrix.

[0097] By adopting the smoothing approximation of Θ and A0, the embodiment of the present application can define Λ s The projection of is the solution of the following minimization problem:

[0098] Solving Λ in the space c such that is minimized, where represents the square of the F-norm of the matrix; Λ c represents the characteristic diagonal matrix matching S in the space .s representing the projection space upper feature diagonal matrix Λ c mapping transform; Therefore, discontinuous reception management is equivalent to solving the eigenvalue maximization problem by applying a smooth approximation, where To determine closed-form solution, the present invention utilizes the specific form of Λ s in Θ' and A0' to satisfy the KARUSH-KUHN-TUCKER optimality condition (referring to the necessary condition for the optimal solution of nonlinear programming), where Λ s is projected onto can be represented by the following framework (F1) consisting of the following equations and unknowns ω and ψ:

[0099]

[0100] where, Tr(·) represents the matrix trace operation; ψ represents the hidden parameter of the smooth approximation; ω p represents the explicit parameter of the smooth approximation.

[0101] By solving the framework (F1), the projection matrix This matrix shows how to transform the problem into an eigenvalue maximization problem, thereby reducing the effective dimension of the problem. In parallel with the scenario where K is known, the following augmented variables are needed to derive the projection on

[0102]

[0103] where S' represents the projection of S on ; I' represents the projection of I on ; diag(·) represents the operation of constructing a diagonal matrix; e p and e q are unit vectors, non-zero at the i-th and j-th positions, respectively, and zero at all other positions. e p,q is defined as an (N c T SR ) 2 dimension zero vector, but 1 at the ((p-1)(N c T SR )+q)-th and ((q-1)(N c T SR )+p)-th positions.

[0104] Given After solving the KARUSH-KUHN-TUCKER condition, the following framework (F2) containing the constraint projection is obtained:​

[0105]

[0106] wherein,

[0107] The projection matrix can be expressed by solving the framework (F2) as follows:

[0108]

[0109] wherein, For unknown K, and the association matrix A is located in the orthogonal space, indicating that there is a feasible solution by the dual solution construction, which is optimal under the corresponding cluster separation rate. Based on this, the discontinuous reception timing parameters in A p,q The position is constructed by a dual solution, and the interior point method can be used to solve the relaxation problem of the semi-definite programming, ensuring that the timer parameter related to the trace penalty term p remains feasible and does not depend on any specific clustering prior information. At this point, mMTC network can realize efficient use of resources and energy saving optimization by strategically grouping MTDs with similar uplink control information transmission modes and optimizing the discontinuous reception timer and periodic parameters accordingly.

[0110] In the embodiment of the application, the cluster access optimization definition problem of uplink control information is converted into an eigenvalue maximization problem of an association matrix determined by the number of groups in each element according to the strategic grouping, and the feasible solution of the augmented form is substituted into the eigenvalue maximization problem to jointly optimize the discontinuous reception period and timing parameters corresponding to each strategic grouping scheme, which can reduce unnecessary signaling interaction in the case of large-scale MTD access, thereby reducing system complexity and operation cost. For the characteristics of a large number of devices in Internet of Things and other application scenarios being in low activity state for a long time, by defining the cluster access optimization definition problem with the minimum value of each element in the similarity metric matrix in the strategic grouping scheme as the optimization objective, the discontinuous reception period is reasonably configured, unnecessary listening and transmission operations of the device are reduced, thereby significantly reducing the energy consumption of the terminal device and prolonging the battery life. While ensuring low power consumption, jointly optimizing the discontinuous reception period and timing parameters corresponding to each strategic grouping scheme can also minimize the transition time from the sleep state to the active state of the device, i.e. shorten the wake-up delay.

[0111] The simulation experiment of the large-scale access energy saving optimization method provided by the embodiment of the application is as follows:

[0112] In the simulation, uplink control information transmission and discontinuous reception configuration parameters in accordance with the 5G standard are adopted. The subcarrier spacing is set to 15 kHz, and the slot length is 14 symbols, in which the scheduling request is mapped to the first symbol of each slot. Therefore, the following settings are adopted for PUCCH format 2, E = 24, M = 128, N f = 12 and T SR = 20. The present application adopts a discontinuous reception long cycle of 2048 milliseconds and a discontinuous reception short cycle of 32 milliseconds, which is consistent with the expected idle period of typical MTD. In addition, in the experiment, a 20 MHz channel with a noise floor of -101 dBm is considered, and the channel impulse response length is 12.

[0113] 1) By modeling the MTD traffic a i (t)p t using a Poisson process, the present application evaluates the event-driven and payload exchange traffic patterns, both of which follow a Poisson distribution and have an average packet arrival rate (denoted by η) of 1 / 300 and 1 / 10, respectively;

[0114] 2) Based on a discrete-time semi-Markov process model, the present application compares the performance of the proposed energy-saving optimization scheme with the typical discontinuous reception mechanism in terms of energy-saving ratio and average wake-up delay.

[0115] Referring to FIGS. 3(a) and 3(b), which are energy-saving ratio and average wake-up delay comparison diagrams under different MTD numbers, the proposed energy-saving optimization scheme is evaluated under event-driven and payload exchange traffic patterns in the present simulation experiment, and different MTD numbers N pThe energy saving ratio and average wake-up delay under different numbers of MTDs. As shown in Fig. 3(a) for the energy saving ratio under different numbers of MTDs, the energy saving performance of the large-scale access energy saving optimization method provided by the embodiment of the application gradually improves and stabilizes with the increase of the total number of MTDs, and is superior to the benchmark scheme under different data packet arrival rates. As the number of data packets arriving per unit time decreases, that is, as η decreases, the MTD will remain in the sleep mode instead of switching to the active state to receive data, thereby saving more energy. In addition, since the focus of optimizing discontinuous reception management is to improve the MTD active state operation, the improvement of the energy saving rate will inevitably cause more sleep switching, which will inevitably cause the average wake-up delay to increase. It has always been a difficulty to seek a balance between energy saving and delay reduction between discontinuous reception cycle switching and selection. As shown in Fig. 3(b) for the average wake-up delay under different numbers of MTDs, the simulation experiment shows that the large-scale access energy saving optimization method proposed under different η values can effectively control the wake-up delay while saving energy. Therefore, it can be seen that with the increase of the number of MTDs, the improvement of the clustering accuracy can improve the adaptive adjustment ability of the discontinuous reception configuration, and to some extent, can also compensate for the increase of the delay.

[0116] Referring to Figs. 4(a) and 4(b), Figs. 4(a) and 4(b) are energy saving ratio and average wake-up delay diagrams under different cumulative observation numbers, and the application draws the relationship diagram of the energy saving ratio and the average wake-up delay and the cumulative observation number N c Fig. 4(a) is the energy saving ratio under different cumulative observation numbers, which shows that the energy saving ratio of the scheme proposed under different η values is overall superior to the benchmark scheme, and the increase of the observation number does not necessarily improve the energy saving efficiency. For a given number of MTDs, a small increase in the observation number has a positive effect on the improvement of the energy saving performance, but too many observations will increase the complexity of the resource allocation after the clustering of the uplink control information, and the optimality of the discontinuous reception management configuration cannot be guaranteed, thereby increasing the energy consumption. Fig. 4(b) is the average wake-up delay under different cumulative observation numbers, which shows that the average wake-up delay of the scheme proposed under different η values can maintain the same level as the benchmark scheme, and can greatly compensate for the delay growth caused by the energy saving improvement. The accumulation of the observation number will cause the average delay to increase, and the reason why too many observations cause the average wake-up delay to decrease slightly is consistent with (a) in Fig. 4, that is, because too many observations cannot guarantee the optimality of the discontinuous reception management configuration, the duration of the retransmission timer increases, which increases the MTD silence time and reduces the overall average wake-up delay of mMTC.

[0117] In the embodiment of the present application, the uplink control information transmission resource scheduling in the large-scale access network and the adaptive adjustment of the discontinuous reception cycle and timer are considered jointly, and a new solution is proposed to reduce the network power consumption while ensuring fast response time. In order to solve the joint optimization based on the uplink control information transmission resource scheduling and the discontinuous reception cycle configuration, a special gradient type solving method is developed through the spectral radius of the similarity metric matrix. Through the projection matrix expansion of the clustered uplink control information, the discontinuous reception cycle timer and the cycle parameter can be closely aligned with the true optimal value in the original space with high probability.

[0118] In the embodiment of the present application, the large-scale access energy saving optimization method based on uplink control information clustering and discontinuous reception management carefully analyzes the uplink control information transmission process and discontinuous reception configuration in the large-scale access network. Compared with the resource allocation optimization method based on the control signaling mechanism and multiple access technology, the proposed method is tailored to the comprehensive uplink and downlink interaction analysis process according to the characteristics of the small number of device activation and sporadic data burst transmission in the large-scale access network, without increasing the processing complexity of the terminal and introducing additional error code risk, reducing power consumption while effectively reducing the wake-up delay. In addition, compared with the discontinuous reception parameter configuration method based on model prediction, the present application processes the uplink control information through the clustering algorithm, which can effectively identify different types of device behavior patterns and make corresponding discontinuous reception configuration adjustment. This signaling interaction driven method not only improves the flexibility of the system, but also enhances its adaptability to external environment changes, such as changes in device behavior and fluctuations in network load. Therefore, even in the complex and variable large-scale access network environment, the large-scale access energy saving optimization method provided in the embodiment of the present application can perform well, with strong robustness and stability.

[0119] Based on the same inventive concept, the embodiment of the present application also provides a structural diagram of a large-scale access energy saving optimization device, as shown in Figure 5 , Figure 5 The structural diagram of the large-scale access energy saving optimization device provided by the embodiment of the present application comprises:

[0120] The uplink control information acquisition module 501 is configured to send a plurality of response configuration information to a plurality of MTDs, so that each MTD sends uplink control information according to the response configuration information; the response configuration information is determined according to the task request sent by each MTD;

[0121] The policy grouping scheme obtaining module 502 is configured to define a cluster access optimization definition problem according to each uplink control information, and to smooth the cluster access optimization definition problem by using a spectral radius gradient solution method, so as to obtain a policy grouping scheme by performing policy grouping on the plurality of MTDs according to the plurality of uplink control information; each element in a similarity metric matrix in the policy grouping scheme is taken as an optimization target in the cluster access optimization definition problem; each element in the similarity metric matrix represents a difference between two observation variables in the uplink control information.

[0122] The large-scale access energy saving optimization implementing module 503 is configured to convert the cluster access optimization definition problem into an eigenvalue maximization problem, and to implement large-scale access energy saving optimization by jointly optimizing discontinuous reception cycles and timing parameters corresponding to each policy grouping scheme by substituting a feasible solution in an augmented form into the eigenvalue maximization problem; the feasible solution in the augmented form is obtained by introducing an augmented variable; each element in the association matrix is determined according to a grouping number in the policy grouping.

[0123] In the embodiment of the present application, the cluster access optimization definition problem of the uplink control information is converted into an eigenvalue maximization problem of an association matrix in which each element is determined according to a grouping number in the policy grouping, and a feasible solution in an augmented form is substituted into the eigenvalue maximization problem to jointly optimize discontinuous reception cycles and timing parameters corresponding to each policy grouping scheme, so that unnecessary signaling interaction can be reduced in the case of large-scale MTD access, and system complexity and operation cost can be reduced. In view of the characteristics that a large number of devices in the application scenarios such as the Internet of Things are in a low activity state for a long time, the cluster access optimization definition problem in which each element in a similarity metric matrix in a policy grouping scheme is taken as an optimization target is defined to reasonably configure discontinuous reception cycles, reduce unnecessary listening and transmission operations of the devices, and thus significantly reduce the energy consumption of the terminal devices and prolong the battery life. While ensuring low power consumption, the joint optimization of discontinuous reception cycles and timing parameters corresponding to each policy grouping scheme can also minimize the transition time of the devices from a sleep state to an active state, that is, shorten the wake-up delay.

[0124] Optionally, the policy grouping scheme obtaining module smoothes the cluster access optimization definition problem by minimizing a spectral radius gradient of an association matrix in the cluster access optimization definition problem and setting a trace penalty term in the cluster access optimization definition problem; the trace penalty term is set according to a cumulative observation number of the uplink control information.

[0125] The policy grouping scheme obtaining module smoothes the cluster access optimization definition problem by minimizing a spectral radius gradient of an association matrix in the cluster access optimization definition problem and setting a trace penalty term in the cluster access optimization definition problem; the trace penalty term is set according to a cumulative observation number of the uplink control information.

[0126] Optionally, the cluster access definition optimization problem is as follows:

[0127]

[0128] Where S represents the similarity measure matrix; ρ represents the trace penalty term; I represents the identity matrix; A represents the incidence matrix; 1 represents the all-one column vector, A1=1 represents that the sum of the elements in each row of A is equal to 1; <·,·> represents the inner product operation of the norm; the element in the pth row and qth column of S is defined as Indicates the observed variable in the uplink control information.

[0129] Optionally, the objective function of the cluster access optimization definition problem is:

[0130]

[0131] Among them, f κ,Θ (Λ) represents the objective function; κ represents the smoothing parameter; λ j (Θ -1 / 2 ΛΘ -1 / 2 ) represents the pair Θ -1 / 2 ΛΘ -1 / 2 The jth eigenvalue obtained after eigenvalue decomposition; Λ indicates that the diagonal element is λ j (Θ -1 / 2 ΛΘ -1 / 2 ) is a diagonal matrix; Θ represents a feasible solution.

[0132] Optionally, the augmented form of feasible solutions includes Θ′ and A0′:

[0133]

[0134] Where Θ′ represents the augmented form of Θ; A0′ represents the augmented form of A0; Θ and A0 are both given feasible solutions; dvec(·) represents the operation of mapping the matrix into a diagonal matrix.

[0135] The embodiment of the present invention further provides an electronic device, such as Figure 6 As shown, it includes a processor 601, a communication interface 602, a memory 603 and a communication bus 604, wherein the processor 601, the communication interface 602, and the memory 603 communicate with each other through the communication bus 604.

[0136] Memory 603, used for storing computer programs;

[0137] The processor 601 is configured to implement the method steps described in any of the above-mentioned large-scale access energy-saving optimization methods when executing the program stored in the memory 603.

[0138] The communication bus mentioned in the above electronic device can be a Peripheral Component Interconnect (PCI) bus or an Extended Industry Standard Architecture (EISA) bus, etc. The communication bus can be divided into an address bus, a data bus, a control bus, etc. For the convenience of representation, only one thick line is used to represent in the figure, but it does not represent that there is only one bus or only one type of bus.

[0139] The communication interface is used for communication between the above electronic device and other devices.

[0140] The memory can include a Random Access Memory (RAM) and can also include a Non-Volatile Memory (NVM), for example, at least one disk memory. Optionally, the memory can also be at least one storage device located away from the above-mentioned processor.

[0141] The above-mentioned processor can be a general-purpose processor, including a Central Processing Unit (CPU), a Network Processor (NP), etc. It can also be a Digital Signal Processing (DSP), an Application Specific Integrated Circuit (ASIC), a Field-Programmable Gate Array (FPGA) or other programmable logic device, a discrete gate or transistor logic device, a discrete hardware component.

[0142] The application further provides a computer readable storage medium. The computer readable storage medium stores a computer program. When the computer program is executed by a processor, the method steps of any one of the above-mentioned large-scale access energy saving optimization methods are implemented.

[0143] Optionally, the computer readable storage medium can be a Non-Volatile Memory (NVM), for example, at least one disk memory.

[0144] Optionally, the computer readable storage medium can also be at least one storage device located away from the above-mentioned processor.

[0145] In yet another embodiment of the present application, a computer program product containing instructions which, when executed on a computer, cause the computer to perform the method steps of any of the above described methods for large scale access energy saving optimization is also provided.

[0146] It should be noted that the terms "first", "second", and so on do not necessarily indicate a specific order or sequence, but are used to distinguish similar objects. It should be understood that the data used in this way can be interchanged under appropriate circumstances, so that the embodiments of the application described herein can be implemented in an order other than that illustrated or described herein. The embodiments described in the following exemplary embodiments do not represent all embodiments consistent with the present application. Rather, they are merely examples of apparatuses and methods consistent with some aspects of the present application.

[0147] In the description of the specification, the description of the terms "one embodiment", "some embodiments", "an example", "a specific example", or "some examples" means that the specific features or characteristics described in connection with the embodiment or example are included in at least one embodiment or example of the present application. In the description of the specification, the illustrative description of the above terms does not necessarily refer to the same embodiment or example. Moreover, the specific features or characteristics described can be combined in any one or more embodiments or examples in a suitable manner. In addition, those skilled in the art can combine and combine different embodiments or examples described in the specification.

[0148] Although the present application is described herein in connection with various embodiments, other variations of the disclosed embodiments can be understood and implemented by those skilled in the art from the description of the drawings and the disclosure. In the description of the present application, the word "comprising" does not exclude other components or steps, "one" or "an" does not exclude a plurality, and "plurality" means two or more, unless otherwise explicitly specified. In addition, some measures are described in different embodiments, but this does not mean that these measures cannot be combined to produce good results.

[0149] The method provided by the embodiments of the present application can be applied to electronic devices. Specifically, the electronic device can be: a desktop computer, a portable computer, a smart mobile terminal, a server, etc. Herein, any electronic device that can implement the present application is within the protection scope of the present application.

[0150] For device / electronic device / storage medium embodiments, because they are basically similar to method embodiments, the description is relatively simple, and the relevant part can be referred to the part of the method embodiment.

[0151] It should be noted that the device, electronic equipment and storage medium of the embodiments of the present application are respectively the device, electronic equipment and storage medium of the above-mentioned one large-scale access energy saving optimization method, and all embodiments of the above-mentioned one large-scale access energy saving optimization method are applicable to the device, electronic equipment and storage medium, and can achieve the same or similar beneficial effects.

[0152] The above is a further detailed description of the present application in combination with specific preferred embodiments, and the specific implementation of the present application cannot be limited to these descriptions. For ordinary skilled persons in the technical field to which the present application belongs, without departing from the concept of the present application, a number of simple deductions or substitutions can be made, which should be considered as falling within the protection scope of the present application.

Claims

1. A large-scale access energy-saving optimization method, characterized in that: The large-scale access energy-saving optimization method includes: Sending multiple response configuration information to multiple MTDs so that each MTD sends uplink control information according to the response configuration information; the response configuration information is determined according to the task request sent by each MTD; A cluster access definition optimization problem for uplink control information is defined based on each uplink control information, and the cluster access optimization definition problem is smoothed using a spectral radius gradient solution method to strategically group the multiple MTDs based on the multiple uplink control information to obtain a strategic grouping scheme; the cluster access definition optimization problem has the optimization goal of minimizing each element in a similarity metric matrix in the strategic grouping scheme; each element in the similarity metric matrix represents a difference between two observed variables in the uplink control information; The cluster access optimization definition problem is converted into an eigenvalue maximization problem of an association matrix, and an augmented form feasible solution is substituted into the eigenvalue maximization problem to achieve large-scale access energy-saving optimization by jointly optimizing the discontinuous reception period and timing parameters corresponding to each strategic grouping scheme; the augmented form feasible solution is obtained by introducing augmented variables; each element in the association matrix is ​​determined according to the number of groups in the strategic grouping.

2. The large-scale access energy-saving optimization method according to claim 1, characterized in that: The cluster access optimization definition problem is smoothed using the spectral radius gradient solution method, including: The cluster access optimization definition problem is smoothed by minimizing the spectral radius gradient of the association matrix in the cluster access optimization definition problem and setting a trace penalty term in the cluster access optimization definition problem; the trace penalty term is set according to the cumulative number of observations of the uplink control information.

3. The large-scale access energy-saving optimization method according to claim 1, characterized in that: The cluster access definition optimization problem is: Where S represents the similarity measure matrix; ρ represents the trace penalty term; I represents the identity matrix; A represents the incidence matrix; 1 represents a column vector of all 1s, A1=1 represents that the sum of the elements in each row of A is equal to 1; <·,·> represents the inner product operation of the norm; the element in the pth row and qth column of S is defined as Indicates the observed variable in the uplink control information.

4. The large-scale access energy-saving optimization method according to claim 1, characterized in that: The objective function of the cluster access optimization problem is: Among them, f κ,Θ (Λ) represents the objective function; κ represents the smoothing parameter; λ j (Θ -1 / 2 ΛΘ -1 / 2 ) represents the pair Θ -1 / 2 ΛΘ -1 / 2 The jth eigenvalue obtained after eigenvalue decomposition; Λ indicates that the diagonal element is λ j (Θ -1 / 2 ΛΘ -1 / 2 ) is a diagonal matrix; Θ represents a feasible solution.

5. The large-scale access energy-saving optimization method according to claim 1, characterized in that: The feasible solutions of the augmented form include Θ′ and A0′: Where Θ′ represents the augmented form of Θ; A0′ represents the augmented form of A0; Θ and A0 are both given feasible solutions; dvec(·) represents the operation of mapping the matrix into a diagonal matrix.

6. A large-scale access energy-saving optimization device, characterized in that: The large-scale access energy-saving optimization device includes: An uplink control information acquisition module is configured to send a plurality of response configuration information to a plurality of MTDs, so that each MTD sends uplink control information according to the response configuration information; the response configuration information is determined according to the task request sent by each MTD; A strategic grouping scheme acquisition module is configured to define a cluster access definition optimization problem for uplink control information based on each uplink control information, and to smooth the cluster access optimization definition problem using a spectral radius gradient solution method to strategically group the multiple MTDs based on the multiple uplink control information to obtain a strategic grouping scheme; the cluster access definition optimization problem has as its optimization objective minimizing each element in a similarity metric matrix in the strategic grouping scheme; each element in the similarity metric matrix represents a difference between two observed variables in the uplink control information; A large-scale access energy-saving optimization implementation module is used to transform the cluster access optimization definition problem into an eigenvalue maximization problem of an association matrix, and substitute an augmented form of a feasible solution into the eigenvalue maximization problem to achieve large-scale access energy-saving optimization by jointly optimizing the discontinuous reception period and timing parameters corresponding to each strategic grouping scheme; the augmented form of the feasible solution is obtained by introducing augmented variables; each element in the association matrix is ​​determined according to the number of groups in the strategic grouping.

7. The large-scale access energy-saving optimization device according to claim 6, characterized in that: The strategic grouping solution acquisition module uses a spectral radius gradient solution method to smoothly process the cluster access optimization definition problem, including: The cluster access optimization definition problem is smoothed by minimizing the spectral radius gradient of the association matrix in the cluster access optimization definition problem and setting a trace penalty term in the cluster access optimization definition problem; the trace penalty term is set according to the cumulative number of observations of the uplink control information.

8. The large-scale access energy-saving optimization device according to claim 6, characterized in that: The cluster access definition optimization problem is: Where S represents the similarity measure matrix; ρ represents the trace penalty term; I represents the identity matrix; A represents the incidence matrix; 1 represents a column vector of all 1s, A1=1 represents that the sum of the elements in each row of A is equal to 1; <·,·> represents the inner product operation of the norm; the element in the pth row and qth column of S is defined as Indicates the observed variable in the uplink control information.

9. An electronic device, characterized in that: It includes a processor, a communication interface, a memory and a communication bus, wherein the processor, the communication interface and the memory communicate with each other via the communication bus; Memory for storing computer programs; The processor is configured to implement the large-scale access energy-saving optimization method according to any one of claims 1 to 5 when executing a program stored in the memory.

10. A computer-readable storage medium, characterized in that The computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the large-scale access energy-saving optimization method according to any one of claims 1 to 5 is implemented.

Images