A lightweight sensor scheduling algorithm for remote state estimation systems
Through the lightweight online scheduling strategy, the complexity and flexibility of sensor scheduling in the remote state estimation system are solved, and efficient and secure sensor scheduling is achieved to adapt to different system parameters and channel constraints.
Patent Information
- Application Number
- CN202310185230.9
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2023-02-27
- Publication Date
- 2025-08-22
- Estimated Expiration
- 2043-02-27
AI Technical Summary
In remote state estimation systems, sensor scheduling problems have defects such as poor optimization effects, complex heterogeneous system scheduling algorithms, and insufficient scheduler flexibility. It is especially difficult to achieve efficient and low-complexity sensor scheduling under limited channel constraints.
A lightweight online scheduling strategy is proposed, through sensor data preprocessing, storage medium storage, scheduler calculation of Whittle index vectors and sending modules, combined with information age AoI vector update, to realize efficient scheduling of sensors.
It realizes the next optimal scheduling performance in heterogeneous systems and channel constraints, reduces the computational complexity, improves the system's flexibility and security, adapts to changes in different system parameters, and reduces the impact of malicious attacks.
Smart Images
Figure CN116319873B_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to a lightweight sensor scheduling algorithm, a sensor, a storage medium, a scheduler and a communication module, and belongs to the fields of network scheduling, data sampling and remote state estimation. Background Art
[0002] Compared with traditional wired state estimation, remote state estimation systems have the advantages of a large sensor deployment range, avoiding wiring difficulties, and allowing rapid deployment. In recent years, remote state estimation systems have been widely used and studied.
[0003] The main application scenarios of remote state estimation are as follows. (1) Smart agriculture. By deploying sensors in farmland to accurately monitor various indicators of each farmland, differentiated management of farmland can be achieved, resource utilization can be maximized, and farmland yield can be increased. (2) Drone tracking. At present, the drone industry is highly developed and has entered all aspects of people's lives. The high maneuverability and wide-area flight capabilities of drones have led to their increasing attention in terms of information theft. At present, the identification and interception of anonymous drones are the main ways to prevent information theft. To achieve this goal, sensor networks are often used for monitoring to detect drones in time, and trajectory estimation algorithms are executed to improve interception efficiency. (3) Environmental information collection. Field exploration work often brings danger to workers due to the uncertainty of its working environment. The use of remote state estimation technology can use drones to deploy some sensors to the target area before exploration, collect information in advance, thereby reducing risks and improving the safety factor of workers.
[0004] In an increasing number of remote state estimation applications, the number of sensors is rapidly increasing. Providing a dedicated channel for each sensor is not feasible. Consequently, the problem of sensor scheduling under limited channel constraints has received increasing attention. Common solutions can be categorized as offline and online strategies. Offline strategies schedule sensors by calculating the optimal scheduling interval for each sensor. While this strategy is fast, it results in high estimation errors in remote state estimation. Online strategies, on the other hand, utilize estimation errors to make real-time decisions for sensor scheduling. While this strategy offers high remote state estimation accuracy, it is difficult to scale the system, suffers from high computational complexity, and is relatively challenging to implement in practice.
[0005] Currently, there are some shortcomings in the solution of multi-sensor scheduling problem under limited channel constraints, such as poor optimization effect, complex scheduling algorithm for heterogeneous systems, and insufficient scheduler flexibility. Summary of the Invention
[0006] The present invention addresses the problems existing in existing solutions and provides a lightweight sensor scheduling algorithm, sensor, storage medium, scheduler, and communication module. This technical solution proposes a lightweight online scheduling strategy for the sensor scheduling problem under limited channel constraints in remote state estimation systems.
[0007] The technical solution of the present invention is as follows: a lightweight remote state estimation system sensor scheduling algorithm, the method comprising the following steps:
[0008] Step 1: The sensor collects environmental data and pre-processes it, including formatting the data and packaging it according to a specific transmission protocol;
[0009] Step 2: The storage medium saves the data requested by the sensor;
[0010] Step 3: The scheduler calculates the Whittle index vector based on the maintained system AoI vector and outputs the scheduling decision.
[0011] Step 4: The sending module standardizes and packages the specified data on the local storage medium and sends it to the remote server according to the scheduling decision of the scheduler.
[0012] Step 5: The server sends an ACK signal back to the sending module to confirm whether the data transmission is successful.
[0013] Step 6: The scheduler updates the system AoI vector based on the sending status returned by the sending module.
[0014] Furthermore, in step 2, the storage medium stores all the data requested to be sent by the sensor before the scheduler makes a scheduling decision. Before the storage medium stores the data, it also includes:
[0015] Before the sending module starts sending, the system observation variables collected by the sensor are persisted to the storage medium. The storage medium should be equipped with a high-speed read and write storage device, and the storage device should have the ability of high-speed reading and writing and repeated erasure to meet the data storage requirements.
[0016] Furthermore, during system initialization, a small number of characteristic parameters are extracted from the estimated system according to the requirements of the proposed algorithm.
[0017] The characteristic parameters should be able to describe or approximate the iterative update speed of the system's estimated error, including the spectral radius of the state transfer matrix of the system's Kalman filter equation and the spectral radius of the system's noise variance matrix.
[0018] Furthermore, the implementation process of step 3 is as follows:
[0019] Step 3-1, scheduler initialization, for the remote state estimation system described by the Kalman filter, x i(k) represents the state vector of system i at time k, ω i represents the system noise of system i, y i (k) represents the observation vector of system i at time k, v i represents the observation noise of the system, A i C i They represent the system state transfer matrix and the system observation matrix respectively. The transmission parameters of the proposed algorithm are defined as in represents the second norm of the stationary error variance of system i, ||Q i || represents the second norm of the system noise variance.
[0020] x i (k) = A i x i (k-1)+B i ω i ,
[0021] y i (k) = C i x i 9k)+D i v i .
[0022] Step 3-2, information age acquisition, system information age Obey the following update strategy, where s i (t) is the indicator variable for successful sending Indicates that the proposed algorithm decides to send the data of system i and sends it successfully,
[0023]
[0024] Step 3-3, Whittle index calculation. Using system information age Δ i , the Whittle index is calculated using the following equation each time the schedule is executed, where W i (Δ i ) indicates that the AoI of system i is Δ i Whittle index when .
[0025]
[0026] Step 3-4: Whittle index sorting and sensor scheduling set acquisition. Due to system channel constraints, the maximum number of sensors scheduled at a time is M. Use the quick sort algorithm to sort the Whittle index vectors and record the corresponding sensor numbers before sorting. Select the sensors corresponding to the largest M Whittle index values for scheduling.
[0027] Steps 3-5: Information age based on the specific results of network transmission To update, return to step 3-2.
[0028] AoI vector update algorithm.
[0029] This method uses an Age of Information (AoI) vector to describe the cumulative effect of historical transmission successes and failures on system estimation errors. Each element of the AoI vector maintains the AoI of the corresponding sensor, and the AoI vector is updated according to the following rules: if the sensor is scheduled and its data is successfully transmitted, the corresponding index of the AoI vector is updated to 1; otherwise, the corresponding index of the AoI vector is incremented by 1.
[0030] Furthermore, in step 4, the sending module standardizes the sensor data to be sent and saves the sensor data in a storage medium. When sending, the sending module should standardize and package the data in the storage medium.
[0031] Furthermore, in step 5, after obtaining the set of scheduled sensors, the sending module sends the data. The sending module should complete the task of reading and sending the data of the scheduled sensors from the storage medium within one time slot. The sending module should ensure the time invariance of the transmission, that is, the same type of data should have the same probability of successful transmission in different time slots. In the process of allocating channels for each sensor data, it should be ensured that the data of each sensor is sent using only one channel to meet the constraint assumptions of the lightweight algorithm.
[0032] Furthermore, in step 6, after the sending module is finished sending, the scheduler's information age is updated. After the sending module is finished sending, the ACK signal returned by the server is used to determine whether the sending is successful and transmit the result to the scheduler to update the information age vector AoI.
[0033] The scheduler described in the above scheme should include the following units: storage unit, computing unit, and communication unit.
[0034] The storage unit stores the system AoI vector and the characteristic parameters of the estimated system. The storage unit should be modifiable. The AoI vector is updated based on the transmission results after each transmission. The characteristic parameters of the estimated system may change due to parameter calculation methods, system structure, and system scale. Therefore, it is necessary to allow the characteristic parameters of the estimated system to be modifiable to meet requirements.
[0035] The computational unit calculates and sorts the Whittle index vector based on the AoI vector and the characteristic parameters of the estimated system, and retrieves the set of scheduled sensors. To reduce computational complexity, common results of complex operations such as exponential operations can be tabulated for querying, accelerating computation.
[0036] After the computation unit determines the scheduled sensor set, the communication unit transmits this set to the communication module to complete the sampling and transmission of sensor data. This communication can be done via wired or wireless communication, but the communication time must be sufficiently short to not affect subsequent data sampling and transmission by the communication module.
[0037] The sending module described in the above solution includes two major functions: data standardization and transmission. Each of these functions will be described below.
[0038] Data standardization is primarily targeted at heterogeneous sensors. Due to its universality, this algorithm allows for variations in sensor data types and sampling accuracy. However, before transmission, sensor data stored in storage media must be further standardized and packaged in the sending module to meet network transmission requirements.
[0039] The sending module should appropriately split overly long data packets. Taking into account channel constraints during transmission, the sending module uses a suitable channel allocation algorithm to allocate a channel for each packet before actual transmission. After transmission, the sending module should check the channel status and the ACK signal returned by the server to determine whether the scheduled sensor data was successfully sent, thereby assisting the scheduler in updating the AoI vector. It should be noted that different network transmission modules require corresponding modifications to meet usage requirements.
[0040] The storage medium described in the above solution should include the following types of circuits: filtering circuits, amplifying circuits, and hardware facilities required for storage.
[0041] The filter circuit filters the data sampled by the sensor to prevent possible noise from affecting the data. The amplifier circuit amplifies the data stored in the medium to facilitate subsequent sampling and processing by the communication module. Storage hardware facilities must meet access speed requirements.
[0042] The sensor in the above scheme should include the following devices: a microprocessor, storage, a sensing unit, an A / D converter, and an auxiliary adjustment device. The microprocessor primarily performs preprocessing of sensor-collected data, such as data smoothing and Kalman filtering. The storage caches data and assists the A / D converter and microprocessor. The sensing unit senses the environment and converts it into an analog signal. The A / D converter further converts the analog signal into a digital signal for further processing by the microprocessor. The auxiliary adjustment device primarily assists the sensing unit and improves its sensing accuracy. The sensor's sampling function requires a sufficiently high sampling rate that meets sampling accuracy without affecting subsequent transmission. An adaptive sampling strategy is employed during the sampling process to improve sampling efficiency. Post-sampling processing strategies, such as, but not limited to, debiasing and filtering, are also employed to mitigate the impact of random noise. The algorithm described in this invention requires that all sensor sensing, A / D conversion, and other operations must be completed before the scheduler executes the scheduling algorithm. Given the periodicity requirements of the filtering algorithm, this requires that the sensor complete sampling and transmission within the filtering cycle. Therefore, the sensor's performance should be fast and stable.
[0043] Compared with the traditional remote state sensor scheduling strategy, this method has the following advantages. (1) Suboptimality of the algorithm. Generally speaking, offline strategies can achieve optimal performance for homogeneous systems, but are difficult to apply to heterogeneous systems or complex constraints. In comparison, this algorithm can achieve suboptimal scheduling performance for heterogeneous systems and channel number constraints. (2) Lightweightness of the algorithm. The operations involved in this algorithm are all scalar operations, and the computational complexity depends on the sorting algorithm. If the sorting algorithm is selected appropriately, the computational complexity of this algorithm can reach O(N). (3) Universality of the algorithm. For different system settings and different system parameters, this algorithm only requires two parameters to perform scheduling. And for system modifications, scheduling algorithm modifications, etc., only minor changes are required to complete. (4) Security of the algorithm. Since this algorithm does not require the state estimation data of the remote state estimation system to run in real time, the security of information transmission in the remote state process is greatly improved. At the same time, this algorithm allows sensors to perform local information encryption, which to a certain extent reduces the impact of malicious attacks on the remote state estimation performance and improves the robustness of the system to attacks. BRIEF DESCRIPTION OF THE DRAWINGS
[0044] Figure 1 It is a structural schematic diagram of the present invention.
[0045] Figure 2 This is a flow chart of the lightweight sensor scheduling algorithm of the present invention.
[0046] Figure 3This is a diagram comparing the remote estimation errors under the constraint that only half of the sensors are scheduled in each scheduling cycle.
[0047] Figure 4 This is a comparison chart of the computing time required for each scheduling under the constraint that only half of the sensors are scheduled in each scheduling period. DETAILED DESCRIPTION
[0048] The following describes the embodiments of the present invention in detail with reference to the accompanying drawings and examples, so that the present invention can fully understand how to apply technical means to solve technical problems and achieve technical effects, and thus implement the invention accordingly. It should be noted that, as long as no conflict exists, the various embodiments of the present invention and the various features of the embodiments can be combined with each other, and the resulting technical solutions are all within the scope of protection of the present invention.
[0049] Example 1: See Figures 1-4 , a lightweight remote state estimation system sensor scheduling algorithm, the method comprising the following steps:
[0050] Step 1: The sensor collects environmental data and pre-processes it, including formatting the data and packaging it according to a specific transmission protocol;
[0051] Step 2: The storage medium saves the data requested by the sensor;
[0052] Step 3: The scheduler calculates the Whittle index vector based on the maintained system AoI vector and outputs the scheduling decision.
[0053] Step 4: The sending module standardizes and packages the specified data on the local storage medium and sends it to the remote server according to the scheduling decision of the scheduler.
[0054] Step 5: The server sends an ACK signal back to the sending module to confirm whether the data transmission is successful.
[0055] Step 6: The scheduler updates the system AoI vector based on the sending status returned by the sending module.
[0056] In step 2, the storage medium saves all the data requested by the sensor before the scheduler makes a scheduling decision. Before the storage medium saves the data, it also contains:
[0057] Before the sending module starts sending, the system observation variables collected by the sensor are persisted to the storage medium. The storage medium should be equipped with a high-speed read and write storage device, and the storage device should have the ability of high-speed reading and writing and repeated erasure to meet the data storage requirements.
[0058] When the system is initialized, a small number of characteristic parameters are extracted from the estimated system according to the requirements of the proposed algorithm.
[0059] The characteristic parameters should be able to describe or approximate the iterative update speed of the system's estimated error, including the spectral radius of the state transfer matrix of the system's Kalman filter equation and the spectral radius of the system's noise variance matrix.
[0060] The implementation process of step 3 is as follows:
[0061] Step 3-1, scheduler initialization, for the remote state estimation system described by the Kalman filter, x i (k) represents the state vector of system i at time k, ω i represents the system noise of system i, y i (k) represents the observation vector of system i at time k, v i represents the observation noise of the system, A i C i They represent the system state transfer matrix and the system observation matrix respectively. The transmission parameters of the proposed algorithm are defined as in represents the second norm of the stationary error variance of system i, ||Q i || represents the second norm of the system noise variance.
[0062] x i (k) = A i x i (k-1)+B i ω i ,
[0063] y i (k) = C i x i (k)+D i v i .
[0064] Step 3-2, information age acquisition, system information age Obey the following update strategy, where s i (t) is the indicator variable for successful sending Indicates that the proposed algorithm decides to send the data of system i and sends it successfully
[0065]
[0066] Step 3-3, Whittle index calculation. Using system information age Δ i , the Whittle index is calculated using the following equation each time the schedule is executed, where W i (Δ i ) indicates that the AoI of system i is Δ i Whittle index when .
[0067]
[0068] Step 3-4: Whittle index sorting and sensor scheduling set acquisition. Due to system channel constraints, the maximum number of sensors scheduled at a time is M. Use the quick sort algorithm to sort the Whittle index vectors and record the corresponding sensor numbers before sorting. Select the sensors corresponding to the largest M Whittle index values for scheduling.
[0069] Steps 3-5: Information age based on the specific results of network transmission To update, return to step 3-2.
[0070] AoI vector update algorithm.
[0071] This method uses an Age of Information (AoI) vector to describe the cumulative effect of historical transmission successes and failures on system estimation errors. Each element of the AoI vector maintains the AoI of the corresponding sensor, and the AoI vector is updated according to the following rules: if the sensor is scheduled and its data is successfully transmitted, the corresponding index of the AoI vector is updated to 1; otherwise, the corresponding index of the AoI vector is incremented by 1.
[0072] In step 4, the sending module standardizes the sensor data to be sent and saves the sensor data to a storage medium. When sending, the sending module should standardize and package the data in the storage medium.
[0073] In step 5, after obtaining the set of scheduled sensors, the sending module sends the data. The sending module should complete the task of reading and sending the data of the scheduled sensors from the storage medium within one time slot. Among them, the sending module should ensure the time invariance of the transmission, that is, the same type of data should have the same probability of successful transmission in different time slots. In the process of allocating channels for each sensor data, it should be ensured that the data of each sensor is sent using only one channel to meet the constraint assumptions of the lightweight algorithm.
[0074] In step 6, after the sending module is finished sending, the scheduler's information age is updated. After the sending module is finished sending, it is determined whether the sending is successful based on the ACK signal returned by the server and the result is transmitted to the scheduler to update the information age vector AoI.
[0075] The scheduler described in the above scheme should include the following units: storage unit, computing unit, and communication unit.
[0076] The storage unit stores the system AoI vector and the characteristic parameters of the estimated system. The storage unit should be modifiable. The AoI vector is updated based on the transmission results after each transmission. The characteristic parameters of the estimated system may change due to parameter calculation methods, system structure, and system scale. Therefore, it is necessary to allow the characteristic parameters of the estimated system to be modifiable to meet requirements.
[0077] The computational unit calculates and sorts the Whittle index vector based on the AoI vector and the characteristic parameters of the estimated system, and retrieves the set of scheduled sensors. To reduce computational complexity, common results of complex operations such as exponential operations can be tabulated for querying, accelerating computation.
[0078] After the computation unit determines the scheduled sensor set, the communication unit transmits this set to the communication module to complete the sampling and transmission of sensor data. This communication can be done via wired or wireless communication, but the communication time must be sufficiently short to not affect subsequent data sampling and transmission by the communication module.
[0079] The sending module described in the above solution includes two major functions: data standardization and transmission. Each of these functions will be described below.
[0080] Data standardization is primarily targeted at heterogeneous sensors. Due to its universality, this algorithm allows for variations in sensor data types and sampling accuracy. However, before transmission, sensor data stored in storage media must be further standardized and packaged in the sending module to meet network transmission requirements.
[0081] The sending module should appropriately split overly long data packets. Taking into account channel constraints during transmission, the sending module uses a suitable channel allocation algorithm to allocate a channel for each packet before actual transmission. After transmission, the sending module should check the channel status and the ACK signal returned by the server to determine whether the scheduled sensor data was successfully sent, thereby assisting the scheduler in updating the AoI vector. It should be noted that different network transmission modules require corresponding modifications to meet usage requirements.
[0082] The storage medium described in the above solution should include the following types of circuits: filtering circuits, amplifying circuits, and hardware facilities required for storage.
[0083] The filter circuit filters the data sampled by the sensor to prevent possible noise from affecting the data. The amplifier circuit amplifies the data stored in the medium to facilitate subsequent sampling and processing by the communication module. Storage hardware facilities must meet access speed requirements.
[0084] The sensor in the above scheme should include the following devices: a microprocessor, a storage, a sensing unit, an AD converter and an auxiliary adjustment device. Among them, the microprocessor mainly performs pre-processing work on the data collected by the sensor, such as data smoothing, Kalman filtering, etc. The storage performs data caching operations and assists the AD converter and the microprocessor in their work. The sensing unit senses the environment and converts it into an analog signal. The AD converter further converts the analog signal into a digital signal for further processing by the microprocessor. The auxiliary adjustment device mainly assists the sensing unit in its work and improves the sensing accuracy of the sensing unit. The sensor requires that its sampling rate is high enough to complete the sampling function, and it should not affect subsequent transmission while meeting the sampling accuracy. In the sampling process, an adaptive sampling strategy is adopted to improve the efficiency of effective sampling. At the same time, sampling post-processing strategies such as but not limited to debiasing and filtering are adopted to reduce the impact of random noise. The algorithm described in the present invention requires that all sensor perception, AD conversion and other operations have been completed before the scheduler executes the scheduling algorithm. Taking into account the periodic requirements of the filtering algorithm, this requires the sensor to complete the sampling and transmission work within the filtering cycle. Therefore, the performance of the sensor should meet the characteristics of fastness and stability. Specific embodiment 2:
[0086] See also Figure 2 ,The algorithm proposed in this invention is a lightweight sensor scheduling algorithm, and its specific implementation steps are as follows.
[0087] Step 1, scheduler initialization, where x i (k) represents the state vector of system i at time k, ω i represents the system noise of system i, y i (k) represents the observation vector of system i at time k, v i represents the observation noise of the system, A i C i They represent the system state transfer matrix and the system observation matrix respectively. The transmission parameters of the proposed algorithm are defined as in represents the second norm of the stationary error variance of system i, ||Q i || represents the second norm of the system noise variance, and α i β i The parameters are transmitted to the scheduler.
[0088] x i (k) = A i x i (k-1)+B i ω i ,
[0089] y i (k) = Ci x i (k)+D i v i .
[0090] Step 2: Get the information age. System information age Obey the following update strategy, where s i (t) is the indicator variable for successful sending It means that the proposed algorithm decides to send the data of system i and sends it successfully.
[0091]
[0092] Step 3: Whittle index calculation. Using system information age Δ i , the Whittle index is calculated using the following equation each time the schedule is executed, where W i (Δ i ) indicates that the AoI of system i is Δ i Whittle index when .
[0093]
[0094] Step 4: Sort the Whittle indexes and obtain the sensor scheduling set. Due to system channel constraints, the maximum number of sensors that can be scheduled at any one time is M. Use the quick sort algorithm to sort the Whittle index vectors and record the corresponding sensor numbers before sorting. Select the sensors corresponding to the M largest Whittle index values for scheduling.
[0095] Step 5: Check the information age based on the specific results of network transmission. Updates.
[0096] The following is a calculation of a specific example to further illustrate the process of this algorithm. For a remote state estimation system, the state transfer equation matrix is as follows
[0097]
[0098]
[0099]
[0100] For AoI = [1, 1], the Whittle index vector is W = [4.003 5.176], so sensor 2 is scheduled; after the transmission is successful, the AoI vector is updated to AoI = [2, 1].
[0101] When scheduling the second time slot, the currently maintained AoI vector AoI = [2, 1] is used to calculate the Whittle index vector W = [21.045.176]. At this time, the decision is made to schedule sensor 1. After the transmission is successful, the AoI vector is updated to AoI = [1, 2].
[0102] Example 3:
[0103] Example 3 of the present invention considers the performance of the algorithm in different system scales when only half of the sensors are scheduled in each scheduling cycle. Figure 2 is the average error of remote state estimation, Figure 3 is the average decision calculation time. Considering that some scheduling algorithms grow faster with the calculation time, Figure 3 The vertical axis uses logarithmic coordinates.
[0104] Figure 2 It shows that compared with simple scheduling algorithms such as greedy strategies, our proposed method (Lightweight AoIWhittle's index policy) shows obvious advantages when the system scale is large. Figure 3 It shows that the computation time of our proposed method hardly changes as the system scale increases.
[0105] From the above experimental results, it can be seen that the method we proposed shows good scheduling effects under different system scales, and the scheduling calculation time does not show obvious changes with the increase of system scale, which shows that the algorithm we proposed has the characteristics of low computational complexity and good scheduling effect.
[0106] Although the embodiments disclosed herein are as described above, the contents described herein are merely embodiments for facilitating understanding of the present invention and are not intended to limit the present invention. Any person skilled in the art may make any modifications and variations in the form and details of the embodiments without departing from the spirit and scope of the present invention. However, the scope of patent protection of the present invention shall remain subject to the scope defined by the appended claims.
Claims
1. A lightweight sensor scheduling method for a remote state estimation system, characterized in that: The method comprises the following steps: Step 1: Sensors collect environmental data and preprocess them; Step 2: The storage medium saves the data requested by the sensor; Step 3: The scheduler calculates the Whittle index vector based on the maintained system AoI vector and outputs the scheduling decision. Step 4: The sending module standardizes and packages the specified data on the local storage medium and sends it to the remote server according to the scheduling decision of the scheduler. Step 5: The server sends an ACK signal back to the sending module to confirm whether the data transmission is successful. Step 6: The scheduler updates the system AoI vector based on the sending status returned by the sending module; The implementation process of step 3 is as follows: Step 3-1, scheduler initialization, assuming that the system contains N subsystems, for the remote state estimation system described by the Kalman filter, as shown in the following formula, where x i (k) represents the state vector of system i at time k, ω i represents the system noise of system i, y i (k) represents the observation vector of system i at time k, v i represents the observation noise of the system, A i C i They represent the system state transfer matrix and the system observation matrix respectively. The transmission parameters of the proposed algorithm are defined as α i =||A i ||, in represents the second norm of the stationary error variance of system i, ||Q i || represents the second norm of the system noise variance, x i (k)=A i x i (k-1)+B i oh i , y i (k)=C i x i (k)+D i v i . Step 3-2, information age acquisition, system information age Obey the following update strategy, where s i (t) is the indicator variable for successful sending γ i =1 means that the proposed algorithm decides to send the data of system i and sends it successfully, Step 3-3, Whittle index calculation, using system information age Δ i , the Whittle index is calculated using the following equation each time the schedule is executed, where W i (Δ i ) indicates that the AoI of system i is Δ i Whittle index when Step 3-4: Whittle index sorting and sensor scheduling set acquisition. Due to system channel constraints, the maximum number of sensors scheduled at each time is M. Use the quick sort algorithm to sort the Whittle index vectors and record the corresponding sensor numbers before sorting. Select the sensors corresponding to the largest M Whittle index values for scheduling; Steps 3-5: Information age based on the specific results of network transmission To update, return to step 3-2.
2. The lightweight remote state estimation system sensor scheduling method according to claim 1 is characterized in that: In step 1, the sensor periodically performs rapid data acquisition. The sensor includes a sensing unit, an AD converter, a microprocessor, a battery module, a buffer, and an auxiliary adjustment device.
3. The lightweight remote state estimation system sensor scheduling method according to claim 1, wherein the hardware feature is that in step 2, the storage medium stores all data requested to be sent by the sensor before the scheduler makes a scheduling decision, and the storage medium further includes: Before the sending module starts sending, the system observation variables collected by the sensor are persisted to the storage medium. The storage medium should be equipped with a high-speed read and write storage device, and the storage device should have the ability of high-speed reading and writing and repeated erasure to meet the data storage requirements.
4. The lightweight remote state estimation system sensor scheduling method according to claim 1, wherein the algorithm is characterized in that a small number of characteristic parameters of the estimated system are extracted during system initialization. The characteristic parameters should be able to describe or approximate the iterative update speed of the system's estimated error, including the spectral radius of the state transfer matrix of the system's Kalman filter equation and the spectral radius of the system's noise variance matrix.
5. The lightweight remote state estimation system sensor scheduling method according to claim 1, characterized in that: In step 4, the sending module standardizes the sensor data to be sent and saves the sensor data to a storage medium. When sending, the sending module should standardize and package the data in the storage medium.
6. The lightweight remote state estimation system sensor scheduling method according to claim 1, characterized in that: In step 5, after obtaining the set of scheduled sensors, the sending module sends the data. The sending module should complete the task of reading and sending the data of the scheduled sensors from the storage medium within one time slot. Among them, the sending module should ensure the time invariance of the transmission, that is, the same type of data should have the same probability of successful transmission in different time slots. In the process of allocating channels for each sensor data, it should be ensured that the data of each sensor is sent using only one channel to meet the constraint assumptions of the lightweight algorithm.
7. The lightweight remote state estimation system sensor scheduling method according to claim 1, characterized in that: In step 6, after the sending module completes sending, the scheduler updates the information age vector AoI. After the sending module completes sending, it determines whether the sending is successful based on the ACK signal returned by the server and transmits the result to the scheduler to update the information age vector AoI.
8. The lightweight remote state estimation system sensor scheduling method according to claim 1, characterized in that: In step 3, the scheduler includes a storage unit, a computing unit, and a communication unit. The storage unit stores the system AoI vector and the characteristic parameters of the estimated system. The storage unit should be modifiable. The AoI vector is updated according to the transmission result after each transmission. The characteristic parameters of the estimated system may change due to the parameter calculation method, system structure and system scale. Therefore, the characteristic parameters of the estimated system can be modified to meet the needs. The computing unit calculates and sorts the Whittle index vector based on the AoI vector and the characteristic parameters of the estimated system, and retrieves the set of scheduled sensors. To reduce computational complexity, common results of exponential and complex operations are tabulated for query to accelerate computation. After the computing unit obtains the scheduled sensor set, the communication unit transmits the set to the communication module to complete the sampling and sending of the sensor data.
9. The lightweight remote state estimation system sensor scheduling method according to claim 1, characterized in that: In step 5, the sending module includes two major functions: data standardization and transmission. Data standardization is mainly for heterogeneous sensors, allowing different sensor data types and sampling accuracy. However, before sending, according to network transmission requirements, the sensor data stored in the storage medium needs to be further standardized and packaged in the sending module to meet transmission requirements. The sending module should reasonably split overly long data packets. Considering the channel constraints during the sending process, the sending module uses an appropriate channel allocation algorithm to allocate a channel for each packet before actual transmission. After the sending is completed, the sending module should detect the channel status and the ACK signal returned by the server to determine whether the scheduled sensor data has been sent successfully, so as to assist the scheduler in updating the AoI vector.