A resource scheduling method of a 5G driving networked control system

Abstract

The present application relates to industrial wireless network technology, in particular to a resource scheduling method of 5G driven networked control system, which is suitable for 5G driven networked control system composed of multiple independent discrete linear subsystems and shared 5G network. Considering the packet loss problem of sensing data and control instructions of 5G driven networked control system, the optimal state estimator and LQG control law are given. By analyzing the internal relationship between LQG cost and 5G transmission reliability, the resource scheduling problem is converted into an integer programming problem. Further, a resource scheduling method is proposed, which allocates the optimal transmission resource and scheduling parameter to each subsystem according to the state of each subsystem in each control period, so as to realize the minimization of expected LQG cost.

Description

Technical Field

[0001] This invention relates to industrial wireless network technology, specifically a resource scheduling method for a 5G-driven networked control system. Background Technology

[0002] The emergence of 5G technology has brought new opportunities and challenges to the field of industrial control. Traditional industrial control systems often suffer from slow data transmission speeds, high latency, and limited connection capabilities. These problems will be greatly improved with the support of 5G technology. 5G technology will bring faster, more stable, more reliable, and more intelligent data transmission methods to the industrial control field, helping factories achieve digital transformation and intelligent manufacturing. However, the actual factory radio frequency environment presents adverse factors such as high path loss, strong noise interference, and multipath effects, making it difficult to meet the control performance requirements of the entire 5G-driven networked control system. This has spurred the design of resource scheduling methods that simultaneously meet the reliability and latency requirements of 5G-driven networked control systems.

[0003] To date, no resource scheduling method oriented towards control performance metrics has been developed for 5G-driven networked control systems that experience packet loss in both uplink and downlink transmissions. Summary of the Invention

[0004] To address the challenges of high path loss, strong noise interference, and multipath effects in real-world factory radio frequency environments, this invention proposes a resource scheduling method for a 5G-driven networked control system to ensure the overall controllability requirements of the system. This method is applicable to 5G-driven networked control systems composed of multiple independent discrete linear subsystems and a shared 5G network.

[0005] The technical solution adopted by the present invention to achieve the above objectives is as follows:

[0006] A resource scheduling method for a 5G-driven networked control system includes the following steps:

[0007] 1) Base station acquires channel state H k Based on the channel state, network resource scheduling is performed on the subsystem to obtain the optimal grid allocation. i = 1, 2, ..., m;

[0008] 2) The base station determines the parameters of the optimal modulation and coding scheme.

[0009] 3) The base station will and Broadcast to all subsystems;

[0010] 4) Subsystem i according to and Send sensor data to base station

[0011] 5) Base station update counter l i,k+1 ;

[0012] 6) Base station calculation control input And send it as a control command to subsystem i;

[0013] 7) Subsystem i evolves in state i.

[0014] By dividing the 5G scheduling frame, different transmission resources are obtained. The transmission reliability of different subsystems is determined by the allocated transmission resources and scheduling parameters. Specifically, the 5G scheduling frame is:

[0015] Each frame consists of 10 subframes of the same size, each 1 ms in length. The grid size is determined by a given Numerology parameter μ. When μ is given, each subframe is divided into s = 2 μ There are 1 time slot, each time slot having a length of t. s =1 / 2 μ Each time slot contains 14 OFDM symbols, with the first two symbols used for control overhead. The available network bandwidth Φ is divided into b Physical Resource Blocks (PRBs), and each PRB has a bandwidth of f. b =12f b , where f b =2 μ ×15kHz, f b The subcarrier spacing.

[0016] The scheduling parameters are specifically as follows:

[0017] The control period T0 = τms, τ = 1, 2, ..., 10, meaning that each τ subframes is one control period. In the k-th control period, a Boolean variable matrix Ω is defined. i,k ={0,1} τs×b If Ω i,k (λ,j)=1 indicates that the grid is assigned to subsystem i for transmission, and Ω is equal to 1 when 1≤λ≤τs / 2. i,k (λ,j) corresponds to the grid assigned to the uplink transmission of subsystem i; when τs / 2<λ≤τs, Ω i,k (λ,j) corresponds to the grid allocated to the downlink transmission of subsystem i; if Ω i,k = 0 indicates that subsystem i has no transmission in the k-th control cycle, and each transmission of subsystem i requires a quantities in the horizontal and vertical directions, respectively. i,k and n i,k Adjacent grids, a i,k ≤τs / 2, n i,k≤b, the set of grid assignments that satisfy this condition is denoted as Furthermore, any grid can be assigned to at most one subsystem, i.e., ∑ i Ω i,k ≤11 T ;

[0018] In the k-th control cycle, for subsystem i, its modulation and coding scheme parameters σ i,k The range of values ​​is M = {0, 1, 2, ..., η}, and the channel gain vector of subsystem i is defined. in This represents a positive real vector of dimension b. Let h represent the channel gain of subsystem i on grid PRBj, assuming h ik It remains constant within a control cycle, for a given h ik The transmission success rate q of subsystem i i (σ i,k ,Ω i,k ) by σ i,k and grid assignment results Ω i,k A joint decision.

[0019] In step 1), the optimization objective of network resource scheduling is to calculate and minimize the LQG cost of the entire system, i.e.:

[0020]

[0021] st∑ i Ω i,k ≤11 T

[0022]

[0023] in, Let c be the optimal grid allocation scheme for subsystem i, i = 1, 2, ..., m. k As an intermediate variable, d k As an intermediate variable, θ is an intermediate variable. and The reliability of uplink and downlink transmissions, respectively. A set of grid assignments that meet the set conditions.

[0024] Step 2) specifically refers to:

[0025] In the k-th control cycle, for a fixed-size data packet, the function τ(σ) is defined. i,k B i,k It returns the parameters σ of the given modulation and coding scheme.i,k and bandwidth B i,k The time required to transmit a data packet is determined by the grid allocation set that meets the set conditions. By definition, subsystem i requires bandwidth n to transmit a data packet. i,k f b 'Among them, n i,k f is the number of adjacent grids in the vertical direction assigned to subsystem i. b The bandwidth for each PRB, and the latency not exceeding the number of adjacent grids in the horizontal direction allocated to subsystem i. i,k For each time slot, to achieve the maximum transmission success rate, the optimal σ i,k The selection rules are as follows:

[0026]

[0027] Among them, t s The length of each time slot.

[0028] Step 5) specifically involves:

[0029] Transmission counter l i,k+1 The update in the k-th control cycle is as follows:

[0030]

[0031] Where, γ i,k Let be a Bernoulli random variable representing the flag indicating whether the uplink data packet transmission of subsystem i was successful or failed.

[0032] Step 6) specifically involves:

[0033]

[0034] in, For the system matrix, For the input matrix, Let S be a real number. i,k+1 and U i,k This is a matrix of intermediate variables used in the calculation process. For state estimation in the k-th control cycle, L i,k The feedback matrix, i.e.

[0035] Step 7) specifically involves:

[0036]

[0037]

[0038] y i,k =γ i,k Ci x i,k +ω i,k

[0039] in, For state vectors, To control the input, For the observation vector, For the system matrix, For the input matrix, For the observation matrix, w is a real number i,k and ω i,k All are Gaussian white noise and are uncorrelated, with a mean of 0, and their covariance matrices are Q. i and R i ;

[0040] In the k-th control cycle, the sensor of subsystem i will transmit the sensing data. The information is transmitted wirelessly to the controller at the 5G base station; the controller calculates a state estimate based on the received information. and control input u i,k and through the base station to u i,k The control input is sent to the actuator of subsystem i; the actuator will receive the control input. It acts on subsystem i.

[0041] The present invention has the following beneficial effects and advantages:

[0042] 1. Considering the packet loss problem of sensor data and control commands in 5G-driven networked control systems, we present its optimal state estimator and LQG control. By analyzing the intrinsic relationship between LQG cost and 5G transmission reliability, we transform the resource scheduling problem into an integer programming problem. Furthermore, we propose a resource scheduling method that allocates optimal transmission resources and scheduling parameters to each subsystem based on the state of each subsystem in each control cycle, thereby minimizing the desired LQG cost.

[0043] 2. In the scheduling process, this invention not only considers the LQG control performance, but also uses the optimal LQG controller to compensate for noise and packet loss, thereby achieving better control performance and system capacity. Attached Figure Description

[0044] Figure 1 5G-driven networked control system;

[0045] Figure 2 This is a schematic diagram of a 5G scheduling frame. Detailed Implementation

[0046] The present invention will now be described in further detail with reference to the accompanying drawings and embodiments.

[0047] This invention proposes a resource scheduling method for a 5G-driven networked control system. Considering the packet loss problem of sensor data and control commands in the 5G-driven networked control system, an optimal state estimator and LQG control law are given. By analyzing the intrinsic relationship between LQG cost and 5G transmission reliability, the resource scheduling problem is transformed into an integer programming problem. Furthermore, a resource scheduling method is proposed that allocates optimal transmission resources and scheduling parameters to each subsystem based on the state of each subsystem within each control cycle, thereby minimizing the desired LQG cost.

[0048] For a 5G-driven networked control system consisting of multiple independent discrete linear subsystems and a shared 5G network, the resource scheduling problem can be transformed into an integer programming problem due to the inherent relationship between LQG cost and 5G transmission reliability. Therefore, a resource scheduling method is proposed to minimize the expected LQG cost.

[0049] This invention mainly comprises three parts: modeling of 5G-driven networked control systems, a resource scheduling problem model with the goal of minimizing expected LQG costs, and a resource scheduling method.

[0050] Modeling of 1.5G-driven networked control system

[0051] like Figure 1 As shown, the 5G-driven networked control system is specifically as follows:

[0052] The 5G-driven networked control system consists of m independent discrete linear subsystems and one shared 5G network. The discrete-time state-space equations and observation equations of subsystem i (i = 1, 2, ..., m) are as follows:

[0053]

[0054]

[0055] y i,k =γ i,k C i x i,k +ω i,k

[0056] in For state vectors, To control the input, For the observation vector, For the system matrix, For the input matrix, This is the observation matrix. w i,k and ω i,k All are Gaussian white noise and are uncorrelated, with a mean of 0, and their covariance matrices are Q. i and R iIn the k-th control cycle, the sensor in subsystem i will transmit the sensing data. The information is transmitted wirelessly to the controller at the 5G base station (BS); the controller calculates a state estimate based on the received information. and control input u i,k and through BS to u i,k The control input is sent to the actuator of subsystem i; finally, the actuator receives the control input. It acts on subsystem i.

[0057] Bernoulli random variable γ i,k For uplink transmission (sensing data from sensor to controller) Modeling packet loss during transmission: If the data packet arrives correctly, then y i,k =C i x i,k +ω i,k Otherwise, y i,k =ω i,k Bernoulli random variable ν i,k For downlink transmission (control commands from controller to actuator) i,k Modeling packet loss during transmission: If the data packet arrives correctly, then... Otherwise, the actuator uses a zero-input scheme, i.e. For γ i,k and ν i,k have and and This refers to the reliability of uplink and downlink transmission.

[0058] 2. Resource scheduling problem model with the objective of minimizing expected LQG cost

[0059] The specific 5G transmission reliability is as follows:

[0060] 5G scheduling frames can be divided into different resources. The transmission reliability of different subsystems is determined by the allocated resources and scheduling parameters.

[0061] The resource scheduling problem model with the objective of minimizing the expected LQG cost is as follows:

[0062] First, let's introduce the 5G scheduling frame and scheduling parameters:

[0063] like Figure 2 As shown, the 5G scheduling frame is specifically as follows:

[0064] For 5G scheduling frames, each frame consists of 10 subframes of the same size, each 1ms in length. The grid size is determined by a given Numerology parameter μ. When μ (μ≥1) is given, each subframe is divided into s=2 μ There are 1 time slot, each time slot having a length of t. s =1 / 2 μ ms. Each time slot contains 14 OFDM symbols, with the first two symbols used for control overhead. Assume the available network bandwidth Φ can be divided into b Physical Resource Blocks (PRBs), each PRB having a bandwidth of f. b =12f b , where f b =2 μ ×15kHz is the subcarrier spacing. Figure 2 Each subframe is divided into two time slots: time slot 1 is used for uplink transmission, and time slot 2 is used for downlink transmission. The number of transmission subsystems is m = 3 (three different grid patterns), and the same number of PRBs are used in both uplink and downlink transmissions within the same subsystem.

[0065] The specific scheduling parameters are as follows:

[0066] The control period T0 = τms (τ = 1, 2, ..., 10), meaning that each τ subframe length is one control period. In the k-th control period, a Boolean variable matrix Ω is defined. i,k ={0,1} τs×b Ω i,k (λ,j) = 1, indicating that the grid (time slot λ, PRBj) is allocated to subsystem i for transmission. When 1 ≤ λ ≤ τs / 2, Ω i,k (λ,j) corresponds to the grid assigned to the uplink transmission of subsystem i; when τs / 2<λ≤τs, Ω i,k (λ,j) corresponds to the grid allocated to the downlink transmission of subsystem i; if Ω i,k = 0 indicates that subsystem i has no transmission in the k-th control cycle. Assume that each transmission of subsystem i requires a quantity in the horizontal and vertical directions, respectively. i,k (a i,k ≤τs / 2) and n i,k (n i,k For adjacent grids of ≤b), the set of grid assignments that satisfy this condition is denoted as . Furthermore, any grid can be assigned to at most one subsystem, i.e., ∑ i Ω i,k ≤11 T .

[0067] In the k-th control cycle, for subsystem i, its modulation and coding scheme (MCS) parameter σ i,kThe range of values ​​is M = {0, 1, 2, ..., η}. Define the channel gain vector of subsystem i. in Let h represent the channel gain of subsystem i on PRBj. Assume h... i,k It remains constant within a control cycle, for a given h i,k The transmission success rate q of subsystem i i (σ i,k ,Ω i,k ) by MCS parameter σ i,k and grid assignment results Ω i,k A joint decision.

[0068] In the k-th control cycle, for a fixed-size data packet, the function τ(σ) is defined. i,k B i,k ), which returns a given σ i,k and bandwidth B i,k The time required to transmit a data packet. According to By definition, subsystem i requires bandwidth n to transmit a data packet. i,k f b And the delay does not exceed a i,k Each time slot. To achieve the highest transmission success rate, the optimal σ... ik The selection rules are as follows:

[0069]

[0070] Secondly, the LQG performance metrics are given:

[0071] Define the available information set of the controller in the k-th control cycle as follows: (i = 1, 2, ..., m), where

[0072] For cases where packet loss occurs in both uplink and downlink transmissions, the optimal estimator is as follows:

[0073]

[0074]

[0075]

[0076]

[0077] P i,k+1|k+1 =P i,k+1|k -γ i,k+1 K i,k+1 C i P i,k+1|k

[0078]

[0079] Based on the finite-time LQG control method, the value function V is defined. i,k (x i,k )as follows:

[0080]

[0081]

[0082] Where N is the length of the time domain being considered. It is a non-negative definite matrix. It is a positive definite matrix.

[0083] The optimal feedback control is:

[0084]

[0085] Substituting the optimal feedback into the function, we get: Where S i,k and c i,k They are respectively:

[0086]

[0087]

[0088] The initial values ​​are S i,N =W i,N c i,N =0.

[0089] Transmission counter l i,k+1 The update in the k-th control cycle is as follows:

[0090]

[0091] The controller can obtain the state prediction value of subsystem i in the k-th control cycle:

[0092]

[0093] definition The expected LQG (Linear Quadratic Gaussian) cost of subsystem i in the kth control cycle is:

[0094]

[0095] Define the channel gain matrix of all subsystems as follows: The value vector of MCS is The grid assignment result is Ω k ={Ω 1,k,...,Ω m,k}. In H k Given the conditions, the uplink and downlink transmission success rate vectors of all subsystems are: and Define the state prediction matrix of all subsystems as follows Then the sum of the expected LQG costs of the m subsystems in the kth control cycle is:

[0096]

[0097] Finally, the resource scheduling problem is transformed into an integer programming problem:

[0098] The optimization objective of network resource scheduling is to minimize the LQG cost of the entire system, that is:

[0099]

[0100] st∑ i Ω i,k ≤11 T

[0101]

[0102] definition The resource scheduling problem described above is equivalent to an integer programming problem:

[0103]

[0104] in

[0105] Problem (5) is an integer programming problem, which is generally difficult to solve. To simplify the problem's solution complexity, it is required that the uplink and downlink transmissions of each subsystem can only be performed within one time slot (i.e., a). i,k =1), and the uplink and downlink transmission allocations for all subsystems are fixed and the same number of grids ρ (i.e., n). i,k =ρ), and through this processing, the Hungarian method can effectively solve this problem. The maximum number of subsystems that can be scheduled within a control cycle is in This is a floor function. When m > m0, all subsystems cannot be scheduled within a single control cycle.

[0106] 3. Resource scheduling methods

[0107] The implementation process of the resource scheduling method is as follows:

[0108] 1) Collect channel state H k Then, the optimal grid assignment is obtained by solving problem (5) in the BS algorithm.

[0109] 2) BS is determined by (2). (i = 1, 2, ..., m);

[0110] 3)BS will and Broadcast to all subsystems;

[0111] 4) Subsystem i according to and Send sensor data to BS (i = 1, 2, ..., m);

[0112] 5) The controller updates the counter l according to (4). i,k+1 (i = 1, 2, ..., m);

[0113] 6) The controller calculates the control input based on (3). (i = 1, 2, ..., m);

[0114] 7) BS according to and Send control commands (i = 1, 2, ..., m) to subsystem i;

[0115] 8) The state of subsystem i evolves according to (1) (i = 1, 2, ..., m).

Claims

1. A resource scheduling method for a 5G-driven networked control system, characterized in that, Includes the following steps: 1) Base station acquires channel status Based on the channel state, network resource scheduling is performed on the subsystem to obtain the optimal grid allocation. , i =1, 2,..., m , among which, if , indicating grid slot PRB j Assigned to subsystem i To transmit, if , indicating subsystem i In the k No transmission during the control cycle; 2) The base station determines the parameters of the optimal modulation and coding scheme. ; 3) The base station will and Broadcast to all subsystems; 4) Subsystem i according to and Send sensor data to base station ; 5) Base station update counter ; 6) Base station calculation control input And send it as a control command to the subsystem. i ; 7) Subsystem i The state evolves; In step 1), the optimization objective of network resource scheduling is to calculate and minimize the LQG cost of the entire system. in, For subsystem i The optimal grid allocation scheme, i =1, 2,..., m , , representing the downlink transmission success rate vector, As an intermediate variable, , As an intermediate variable, , , , Let be the predicted state value of subsystem i in the kth control cycle. For Bernoulli random variables, process variables Value function variable , It is a positive definite matrix. It is a non-negative definite matrix. As an intermediate variable, Bernoulli random variable This indicates that the uplink data packet was lost. This indicates that the uplink data packet arrived correctly; Bernoulli random variable. This indicates that downlink data packets were lost. This indicates that the downlink data packet arrived correctly. and have and , and The reliability of uplink and downlink transmissions, respectively. A set of grid assignments that meet the set conditions. , The optimal estimator is: The controller in k The available information set for each control cycle is , , , ; Step 6) specifically involves: in, For the system matrix, For the input matrix, For real numbers, and This is a matrix of intermediate variables used in the calculation process. In the first k State estimation for each control cycle The feedback matrix, i.e. ; Step 7) specifically involves: in, For state vectors, To control the input, For the observation vector, For the observation matrix, For real numbers, and All are Gaussian white noise and are uncorrelated, with a mean of 0. The covariance matrices are respectively and ; In the k One control cycle, subsystem i The sensor will sense the data The information is transmitted wirelessly to the controller at the 5G base station; the controller calculates a state estimate based on the received information. and control input and through base stations Send to subsystem i The actuator; the actuator receives control input Acting on subsystems i .

2. The resource scheduling method for a 5G-driven networked control system according to claim 1, characterized in that, By dividing the 5G scheduling frame, different transmission resources are obtained. The transmission reliability of different subsystems is determined by the allocated transmission resources and scheduling parameters. Specifically, the 5G scheduling frame is: Each frame consists of 10 subframes of the same size, each 1ms in length, with the grid size determined by a given Numerology parameter. Decision, when Given a time frame, each subframe is divided into... There are 1 time slot, and the length of each time slot is 1. ms, each time slot contains 14 OFDM symbols, of which the first two symbols are used for control overhead, and the available network bandwidth. Divided into b There are 10 physical resource blocks (PRBs), and the bandwidth of each PRB is 10 ... ,in kHz, The subcarrier spacing.

3. The resource scheduling method for a 5G-driven networked control system according to claim 2, characterized in that, The scheduling parameters are specifically as follows: Control cycle , That is, each Each subframe has a length of one control cycle, and in the... k Each control cycle defines a Boolean variable matrix. ,like This indicates that the grid is assigned to the subsystem. i Transmission is performed when hour, Corresponding allocation to subsystems i Uplink transmission mesh; when hour, Corresponding allocation to subsystems i The downlink transmission grid; if , indicating subsystem i In the k No transmission during a control cycle, subsystem i Each transmission requires a number of [number] units in the horizontal and vertical directions, respectively. and Adjacent grids, , The set of grid assignments that satisfy this condition is denoted as Furthermore, any grid can be assigned to at most one subsystem, i.e. ; In the k Each control cycle, for the subsystem i Its modulation and coding scheme parameters The range of values ​​is Define subsystem i Channel gain vector ,in The dimension is b A positive real number vector Representation Subsystem i In grid PRB j Channel gain on, assuming It remains constant within a control cycle, for a given Subsystem i Transmission success rate Depend on and grid assignment results A joint decision.

4. The resource scheduling method for a 5G-driven networked control system according to claim 1, characterized in that, Step 2) specifically refers to: In the k For a fixed-size data packet, define a function for each control cycle. It returns the parameters of a given modulation and coding scheme. and bandwidth The time required to transmit a data packet is determined by the grid allocation set that meets the set conditions. Definition of subsystem i Transmitting a data packet requires bandwidth. in, To allocate to subsystems i The number of adjacent grid cells in the vertical direction, The bandwidth for each PRB, and the latency does not exceed the bandwidth allocated to the subsystem. i Number of adjacent grids in the horizontal direction a i,k To achieve the highest transmission success rate, the optimal time slot is [number of slots]. The selection rules are as follows: in, The length of each time slot.

5. The resource scheduling method for a 5G-driven networked control system according to claim 1, characterized in that, Step 5) specifically involves: Transmission counter In the Each control cycle is updated to: in, Subsystem for representing Bernoulli random variables i A flag indicating whether the uplink data packet transmission was successful or failed.

Images