SDR sensor scheduling optimization method based on Bayesian reasoning

Abstract

The invention discloses an SDR sensor scheduling optimization method based on Bayesian reasoning, and solves the problem that in the prior art, sensor scheduling with time window constraints cannot be effectively processed. The method comprises the steps that model parameters of a sensor scheduling optimization model during sensor scheduling are determined; wherein the model parameters comprise the state of each sensor in the # imgabs0 # time period, the state of each static target in the # imgabs1 # time period, the state of each dynamic target in the # imgabs2 # time period and the working state sequence of all the sensors; constructing an objective function and constraint conditions of a sensor scheduling optimization model according to the model parameters; solving the target function according to the constraint condition based on Bayesian reasoning to obtain an optimal state sequence of sensor scheduling optimization; rapid convergence can be realized when a sensor scheduling problem with time window constraint is effectively processed, and a solution violating time constraint can be prevented from being generated during crossover variation.

Description

Technical Field

[0001] The present invention relates to the field of wireless communication technologies, and particularly to an SDR sensor scheduling optimization method based on Bayesian inference. Background Art

[0002] In modern wireless communication and Internet of Things systems, the introduction of Software-Defined Radio (SDR) technology provides a new optimization means for target detection and tracking tasks. SDR dynamically configures sensor functions through software, supports real-time optimization and flexible scheduling, thereby enhancing system adaptability and resource utilization efficiency. The implementation of this technology has demonstrated significant advantages in addressing resource constraints and complex environment adaptation issues in traditional methods.

[0003] In target detection and tracking tasks, SDR sensors adjust the turn-on time and frequency to achieve precise monitoring of target positions and states, and maximize the detection time and coverage. However, this process involves complex issues such as multi-target matching and resource allocation, and it is necessary to coordinate sensor resources under multi-dimensional constraints such as time, space, and frequency to ensure the efficient execution of detection tasks. Summary of the Invention

[0004] The present invention provides an SDR sensor scheduling optimization method based on Bayesian inference, which solves the problem in the prior art that the sensor scheduling with time window constraints cannot be effectively processed, realizes fast convergence when effectively processing the sensor scheduling with time window constraints, and can prevent the generation of solutions that violate time constraints during cross-variation.

[0005] The present invention provides an SDR sensor scheduling optimization method based on Bayesian inference, and the method includes: Determine the model parameters of the sensor scheduling optimization model for sensor scheduling; wherein, the model parameters include: the states of each sensor in the time period, the states of each static target in the time period, the states of each dynamic target in the time period, and the working state sequence of all sensors; Construct the objective function and constraint conditions corresponding to the sensor scheduling optimization model according to the model parameters; Under the constraint conditions, solve the objective function based on Bayesian inference to obtain the optimal state sequence of sensor scheduling optimization.

[0006] One or more technical solutions provided in the present invention have at least the following technical effects or advantages: The present invention uses Bayesian inference to predict the on - state of a certain sensor in the next generation based on historical data and new data, realizing an efficient repair strategy and a legal offspring solution generation mechanism. This method can not only assist in improving the daily effective utilization rate of SDR sensors and obtaining the optimal state sequence, but also optimize resource utilization and reduce power loss caused by invalid activation. The present invention uses Bayesian inference, which can not only converge quickly, but also prevent the generation of solutions that violate time constraints during crossover and mutation. This can effectively handle the sensor scheduling problem with time - window constraints. Description of the Drawings

[0007] Figure 1 It is a flowchart of the SDR sensor scheduling optimization method based on Bayesian inference provided by an embodiment of the present invention. Detailed Embodiment

[0008] Next, the technical solutions in the embodiments of the present invention will be clearly and completely described in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, rather than all of the embodiments. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts shall fall within the protection scope of the present invention.

[0009] The present invention provides an SDR sensor scheduling optimization method based on Bayesian inference, as Figure 1 shown, the method includes the following steps S101 to S103.

[0010] S101, determining the model parameters of the sensor scheduling optimization model for sensor scheduling; wherein, the model parameters include: the state of each sensor in the period, the state of each static target in the period, the state of each dynamic target in the period, and the working state sequence of all sensors; Here, the 24 - hour day is divided into time - granularity time periods, that is, , and the time set is represented as ; suppose the sensor set is , , is the total number of sensors; the static target set is , , is the total number of static targets; the dynamic target set is , , is the total number of dynamic targets.

[0011] Specifically, in step S101, the status of each sensor in the time period is expressed as: ; where represents the longitude of sensor ; represents the latitude of sensor ; represents the detection range of sensor ; represents the set of detectable frequency points of sensor in the th time period; represents the frequency working status of the th frequency point of sensor in the th time period; represents the number of available channels of sensor in the th time period; represents the status of sensor in the th time period.

[0012] Specifically, in step S101, the status of each static target in the time period is expressed as: ; where represents the longitude of static target ; represents the latitude of static target ; represents the radiation radius of static target ; represents the set of radiation frequencies of static target in the th time period; represents the working status of static target in the th time period; represents the status of static target in the th time period, static target set , , , and

[0013] time period is expressed as: ; where represents the longitude of the start time of dynamic target in the th time period; represents dynamic target​ The latitude at the start time of the th time period; Indicates the dynamic target in the th time period; Indicates the dynamic target in the th time period; Indicates the dynamic target in the th time period; the state of the dynamic target set , , is the total number of dynamic targets.

[0014] Specifically, in step S101, the states of each dynamic target in the th time period and the working state sequences of all sensors are expressed as: ; where represents the working state sequence of sensor , and , the sensor set , , is the total number of sensors.

[0015] S102. Construct the objective function and constraint conditions of the sensor scheduling optimization model according to the model parameters; Here, the objective function includes: the first objective function and the second objective function; where The first objective function is expressed as: ; The second objective function is expressed as: ; Where: ; ; ; Where represents the total number of static targets; represents the total number of sensors; represents the total number of time periods, ; represents the intersection area of the detection range of sensor and the radiation range of static target ; represents the radiation range of static target ; represents sensor in the the frequency point of the indicating the static target at the radiation frequency set of the indicating the sensor at the frequency point of the working state of the indicating the static target at the working state of the indicating the sensor at the when the <( frequency point is turned on and the static target matching degree; indicating the dynamic target at the radiation frequency set of the used to indicate whether the dynamic target at the is within the detection range of the sensor ; indicating the dynamic target at the radiation frequency set of the indicating the sensor at the when the frequency point is turned on and the dynamic target matching degree; indicating the dynamic target at the % working state of the indicating the sensor union of the sets of detectable frequency points in each time period.

[0016] When at the working state of the dynamic target When at the working state of the dynamic target When at the the dynamic target is within the detection range of the sensor When at the not in the sensor detection range during a certain period ; When it is, it indicates the working state of the static target at the working state of a certain period is that all frequency points are turned on, When it is, it indicates the working state of the static target at the working state of a certain period is that all frequency points are turned off.

[0017] Here, the constraint conditions include the first constraint condition, the second constraint condition, and the third constraint condition; among them: the first constraint condition includes restricting the turn-on time of each sensor; the second constraint condition includes restricting the number of turned-on frequency points within each time period; the third constraint condition includes restricting the single-day turn-on duration of each sensor.

[0018] Specifically, the first constraint condition is expressed as: ; The second constraint condition is expressed as: ; The third constraint condition is expressed as: ; Among them, represents the frequency point working state of the th frequency point of the sensor in the th time period; represents the start time of the available turn-on period of the sensor ; represents the end time of the available turn-on period of the sensor ; represents the total number of time periods, ; represents the union of the sets of detectable frequency points of the sensor in each time period; represents the number of available channels of the sensor in the th time period; represents the single-day maximum turn-on duration of the sensor ; represents the time granularity.

[0019] S103. Under the constraint conditions, solve the objective function based on Bayesian inference to obtain the optimal state sequence of sensor scheduling optimization.

[0020] Specifically, in step S103, the objective function is solved based on Bayesian inference to obtain the optimal state sequence for sensor scheduling optimization, including: determining the maximum number of iterations , initializing the working state of the first frequency point, and obtaining the set of scheduling schemes for each iteration according to the cyclic update method; and determining the optimal state sequence for sensor scheduling optimization according to the set of scheduling schemes corresponding to the maximum number of iterations.

[0021] The set of scheduling schemes for each iteration is obtained according to the cyclic update method, including:[[]] (1) According to the set of scheduling schemes after the -th iteration , the fitness of each state sequence is calculated using the fitness value calculation formula ; where the fitness is positively correlated with the objective function; where represents the -th scheduling scheme in the set of scheduling schemes , , represents the size of the set of scheduling schemes ; (2) According to the fitness of each state sequence , the binary tournament selection strategy is used to select state sequences and put them into the mating pool to obtain the first set of scheduling schemes after the -th iteration ; (3) According to the state sequences in the first set of scheduling schemes after the -th iteration , the second set of scheduling schemes for the -th iteration is generated using the non-zero rate predicted by Bayesian inference ; Specifically, the implementation process of (3) is as follows: (3.1) Define a unit vector with a size of and a unit vector with a size of , and determine the initial weight value; where the size of the unit vector is expressed as: ; and define an empty matrix with a size of ; where represents the total number of sensors; represents the total number of time periods; represents the number of elements in the union of the sets of detectable frequency points of sensor in each time period.

[0022] (3.2) Save the scheduling sequences in the first scheduling scheme set in the matrix and initialize ; (3.3) Count respectively the number of times that the sensor is in the on state in each scheduling scheme in the matrix at the th time period and the th frequency point in the th scheduling scheme ; (3.4) Determine the frequency that the sensor is in the on state at the th time period and the th frequency point according to the probability calculation formula and the number of times ; (3.5) Calculate the index of the frequency in the matrix and put the frequency into the matrix according to the index; (3.6) Select any column in the matrix , judge whether the probability corresponding to each element in this column satisfies the first judgment condition, and modify the elements in the matrix according to the judgment result to obtain the first modified matrix; (3.7) Determine the corresponding elements in the matrix according to the row and column of the elements in the first modified matrix, and judge whether the elements in the matrix satisfy the second judgment condition, and modify the elements in the first modified matrix according to the judgment result to obtain the second modified matrix; (3.8) Obtain the second scheduling scheme set in the th iteration according to the second modified matrix.

[0023] (4) Merge the second scheduling scheme set and the scheduling scheme set [[ID=Y]]to obtain the third scheduling scheme set ; (5) Perform non-dominated sorting on the third scheduling scheme set , retain the first W fronts after each sorting, and delete the scheduling scheme with the smallest fitness from the Wth front until the size of the third scheduling scheme set is to obtain the scheduling scheme set for the next iteration.

[0024] Exemplarily, in a specific usage embodiment provided by the present invention: Step 1, define the current iteration number as , and initialize ; set the maximum iteration number to ; Randomly initialize the on-state sequence of the th time, so as to generate a scheduling scheme set with a size of under the th iteration; where represents the state sequence corresponding to the th scheduling scheme set under the th iteration, and ; represents the state sequence of the sensor in the nd scheduling scheme set under the th iteration , ; Step 2, define two unit vectors with a size of : and ; initialize the weight ; where ; Step 3, define an empty matrix with a size of ; Step 4, according to the fitness value calculation formula, calculate the fitness value of each state sequence in the scheduling scheme set ; where represents 's fitness value. The fitness value calculation formula is expressed as: ; where represents the first objective value of the scheduling scheme under the th iteration ; represents the second objective value of the scheme under the

[0025] th iteration; where the objective value is calculated according to the first objective function and the second objective function. Step 5, adopt the binary tournament selection strategy to select scheduling schemes and put them into the mating pool, and call it the first scheduling scheme set , where the higher the fitness of a scheduling scheme, the higher the probability of being selected.

[0026] Step 6. Generate a second scheduling scheme set at the th iteration from the scheduling schemes in the first scheduling scheme set ; ; (6.1) Save the scheduling schemes in the first scheduling scheme set in the matrix ; (6.2) Initialize ; (6.3) Count the number of different working states of the sensor in the th time period and the th frequency point for each of the scheduling schemes ; (6.4) Calculate the probability of different working states of the sensor in the th time period and the th frequency point at the th iteration according to the following formula: ; where represents the th element in the th row and th column of the matrix at the th iteration, ; indicates the th element in the unit vector ; ; indicates the th element in the unit vector ;

[0027] (6.5) Let , ; represents the th column of the matrix . Respectively judge whether the corresponding probability of each element satisfies . If it satisfies, then the corresponding scheduling scheme in obeys , and if it does not satisfy, then , where ; ; (6.6) Let , ; (6.7) If , let , and execute step (6.3); If , , let , , and execute step (6.3); If , , , let , , , and execute step (6.3); If , execute step (6.8); (6.8) Let ; (6.9) Select a random column element from the rows of the matrix and go to step (6.10); if , go to step (6.11); represents the -th iteration of the -th element in the -th row of the matrix; (6.10) If , let ; otherwise, let ; represents the -th iteration of the -th element in the -th row of the matrix.

[0028] If , ; otherwise, ; (6.11) Let , ; (6.12) If , execute step (6.13); otherwise, let and execute step (6.9); ](6.13) Let the second scheduling scheme set in ; (6.14) Let .

[0029] Step Seven, the scheduling scheme set and the second set of scheduling schemes Combine them to obtain the combined third set of scheduling schemes .

[0030] Step eight, for the third set of scheduling schemes Perform non-dominated sorting and retain the first W fronts, where W represents the minimum value that satisfies the sum of the number of all scheduling schemes in the W fronts is greater than or equal to ; Step nine, delete the scheduling scheme with the minimum fitness from the Wth front; Step ten, repeat the process of step nine until the number of all state sequences in the W fronts is ; thus obtaining the scheduling scheme set for the next iteration with a size of in the th iteration .

[0031] The introduction of the Bayesian strategy effectively improves the global search ability of the algorithm. By dynamically updating the posterior probability, it guides the genetic operator to be more inclined to generate high-quality solutions, reduces the risk of falling into local optima, and thus can find a scheduling scheme close to the optimum faster. Secondly, the algorithm only adopts targeted crossover operations and mutation operations during the evolution process. Only adjust the dimensions that do not violate the constraints, avoiding the generation of offspring that violate the constraints, and maintaining the feasibility and stability of the solutions. This strategy not only maintains the diversity of the population but also ensures the efficiency and reliability of the evolution process, thus obtaining a set of approximately optimal scheduling scheme sets.

[0032] The various embodiments in this specification are described in a progressive manner. For the same or similar parts between the various embodiments, reference can be made to each other. The key point of each embodiment is to illustrate the differences from other embodiments. All or part of the present invention can be used in many general or special computer system environments or configurations. For example: personal computers, server computers, handheld devices or portable devices, tablet devices, mobile communication terminals, multi-processor systems, microprocessor-based systems, programmable electronic devices, network PCs, minicomputers, mainframe computers, distributed computing environments including any of the above systems or devices, etc.

[0033] The above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit the present invention; although the present invention has been described in detail with reference to the foregoing embodiments, those of ordinary skill in the art should understand that: they can still modify the technical solutions described in the foregoing embodiments, or perform equivalent replacements on some or all of the technical features; and these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the scope of the technical solutions of the present invention.

Claims

1. An SDR sensor scheduling optimization method based on Bayesian inference, characterized in that Including: Determine the model parameters of the sensor scheduling optimization model for sensor scheduling; wherein, the model parameters include: the states of each sensor in the time period, the states of each static target in the time period, the states of each dynamic target in the time period, and the working state sequences of all sensors; Construct the objective function and constraint conditions corresponding to the sensor scheduling optimization model according to the model parameters; Under the constraint conditions, solve the objective function based on Bayesian inference to obtain the optimal state sequence of sensor scheduling optimization.

2. The SDR sensor scheduling optimization method based on Bayesian inference according to claim 1, characterized in that The states of the respective sensors at the time period are expressed as: ； Among them, represents the longitude of the sensor ; represents the latitude of the sensor ; represents the detection range of the sensor ; represents the set of frequency points that the sensor can enable detection in the th time period; represents the operating state of the th frequency point of the sensor in the th time period; represents the number of available channels of the sensor in the th time period; represents the state of the sensor in the th time period.

3. The SDR sensor scheduling optimization method based on Bayesian inference according to claim 1, characterized in that The states of the respective static targets at the time period are represented as: ； in, Represents a static target longitude; Represents a static target Latitude; Represents a static target Radiation radius; Represents a static target In the The set of radiation frequencies in a time period; Represents a static target In the Working status of each period; Represents a static target In the The status of the time period, , is the total number of static targets.

4. The SDR sensor scheduling optimization method based on Bayesian inference according to claim 1, wherein The states of the respective dynamic targets at the time period are represented as: ； in, Represents a dynamic target In the The longitude of the starting time of each period; Represents a dynamic target In the The latitude of the starting time of each period; Represents a dynamic target In the The set of radiation frequencies in a time period; Represents a dynamic target In the Working status of each period; Represents a dynamic target In the The status of the time period, , is the total number of dynamic targets.

5. The SDR sensor scheduling optimization method based on Bayesian inference according to claim 1, wherein The working state sequences of all sensors are represented as: ； Among them, represents the working state sequence of the sensor , and , , is the total number of sensors; represents the working state of the sensor at the th time period, , is the total number of time periods.

6. The method for optimizing SDR sensor scheduling based on Bayesian inference according to claim 1, wherein The constraint conditions include a first constraint condition, a second constraint condition, and a third constraint condition; where: The first constraint condition is used to constrain the activation time of each sensor; The second constraint condition is used to constrain the number of activated frequency points in each time period; The third constraint condition is used to constrain the daily activation duration of each sensor.

7. The SDR sensor scheduling optimization method based on Bayesian inference according to claim 6, wherein The first constraint condition is expressed as: ； The second constraint condition is expressed as: ； The third constraint condition is expressed as: ； Among them, represents the frequency point operating state of the sensor at the th frequency point in the th time period; represents the start time of the sensor 's turn-on period; represents the end time of the sensor 's turn-on period; ; represents the union of the sets of detectable frequency points of the sensor in each time period; represents the number of available channels of the sensor at the th time period; represents the maximum daily turn-on duration of the sensor ; represents the time granularity.

8. The SDR sensor scheduling optimization method based on Bayesian inference according to claim 1, characterized in that The objective function includes: a first objective function and a second objective function; where, The first objective function is expressed as: ； The second objective function is expressed as: ； in, Indicates the total number of static targets; Indicates the total number of sensors; Indicates the total number of time periods, ; Indicates sensor Detection range and static targets The area of the intersection of the radiation range; Represents a static target Radiation range; Indicates sensor In the The first frequency points; Represents a static target In the The set of radiation frequencies in a time period; Indicates sensor In the The first Frequency working status of each frequency point; Represents a static target In the Working status of each period; Indicates sensor In the Time slots open Frequency point and static target degree of matching; Represents a dynamic target In the The set of radiation frequencies in a time period; Used to represent dynamic targets In the Is the sensor in the time period within the detection range; Represents a dynamic target In the The set of radiation frequencies in a time period; Indicates sensor In the Time slots open Frequency point and dynamic target degree of matching; Represents a dynamic target In the Working status of each period; Indicates sensor The union of the sets of detectable frequency points in each time period.

9. The SDR sensor scheduling optimization method based on Bayesian inference according to claim 1, wherein The solving of the objective function based on Bayesian inference under the constraint conditions to obtain the optimal state sequence of sensor scheduling optimization includes: Determine the maximum number of iterations , initialize the working state of the first frequency point, and obtain the set of scheduling schemes under each round of iteration according to the loop update method; and determine the optimal state sequence of sensor scheduling optimization according to the set of scheduling schemes corresponding to the maximum number of iterations. Obtaining the set of scheduling schemes in each round of iteration according to the cyclic update method, including: According to the set of scheduling schemes after the -th iteration, calculate the fitness corresponding to each state sequence using the fitness value calculation formula ; where the fitness is positively correlated with the objective function; where represents the set of scheduling schemes in the -th scheduling scheme, , represents the size of the set of scheduling schemes ; According to the fitness corresponding to each state sequence , a binary tournament selection strategy is used to select state sequences and put them into the mating pool to obtain the first set of scheduling schemes after the -th iteration ; According to the state sequence in the first scheduling scheme set after the th iteration, generate the second scheduling scheme set in the th iteration by using the non-zero rate predicted by Bayesian inference ; ; Merge the second scheduling plan set and the scheduling plan set to obtain a third scheduling plan set ; For the set of the third scheduling schemes perform non-dominated sorting, and each time after sorting, retain the first W fronts, and delete the scheduling scheme with the smallest fitness from the W-th front until the size of the set of the third scheduling schemes is , and obtain the set of scheduling schemes for the next iteration .

10. The SDR sensor scheduling optimization method based on Bayesian inference according to claim 9, wherein The first scheduling plan set after the ith iteration, using the non-zero rate predicted by Bayesian inference to generate the second scheduling plan set for the ith iteration, including: , including: Define a size of unit vector and a size of unit vector , and determine the initial weight value; where , represents the total number of sensors; represents the total number of time periods; represents the number of elements in the union of the sets of detectable frequency points of sensor in each time period; and define an empty matrix of size ; Save the scheduling sequences in the first scheduling scheme set in the matrix and initialize to make ; ; Separate statistical matrix The sensors in each scheduling scheme In the th time period, the th frequency point in the number of times the working state is on in the scheduling schemes ; According to the said number of times , use the probability calculation formula to determine the sensor at the th time period and the th frequency point, the frequency at which the working state is on; Calculate the index of the frequency in the matrix and put the frequency into the matrix according to the index; Select a matrix Any column in , determine whether the probability corresponding to each element in this column satisfies the first judgment condition. According to the judgment result, modify the elements in the said matrix to obtain a first modified matrix; Determine the corresponding elements in the matrix according to the rows and columns of the elements in the first modification matrix, and determine whether the elements in the matrix meet the second determination condition. Modify the elements in the first modification matrix according to the determination result to obtain a second modification matrix; ​ Obtain the second scheduling plan set at the th iteration according to the second modification matrix .

Images