Sensor adaptive sampling rate control method and device, computer equipment and storage medium

Abstract

The embodiment of the invention discloses an adaptive sampling rate control method for a sensor, and the method is characterized in that the method comprises the steps: obtaining a short-term average information age ak-1 and a sampling rate uk-1 of the sensor in a previous control period, and calculating the sampling rate in the current control period according to the interrupt threshold delta of the short-term average information age, the sampling rate uk-1 in the previous control period and the short-term average information age ak-1, and sampling sensor data in the current control period according to the sampling rate in the current control period. According to the self-adaptive sampling rate control method for the sensor, the real-time performance of the sensor for collecting the sensing data is guaranteed under the non-stationary service time condition, and the timeliness of the sensing data in an actual network can be remarkably improved.

Description

Technical Field

[0001] The present invention relates to the field of sensors and signal processing technology, and in particular to a sensor adaptive sampling rate control method, computer equipment and storage medium. Background Art

[0002] Sensors are widely used in fields such as autonomous driving, smart manufacturing, and smart home. In these scenarios, sensors sample a physical process such as temperature, humidity, speed, and position, and encapsulate the sampling results into packets and upload them to the monitor via a wireless link to ensure that the monitor can grasp the latest status of the physical process in real time [1]. Since different packets have different values, outdated and old packets are often of low value and cannot reflect the latest status of sensor measurements. Therefore, traditional performance indicators such as throughput and latency cannot measure the real-time performance of sensor data. For example, maximizing throughput requires increasing the sampling rate of sensors, but this will cause network congestion and make the packets old and outdated. On the contrary, minimizing latency requires reducing the sampling rate of sensors as much as possible, but this will cause the monitor to be unable to obtain the sampling data of the sensor for a long period of time, and it is difficult to effectively grasp its latest status, and the real-time performance of sensor data cannot be guaranteed. To address this problem, the real-time performance of sensor data can be characterized by a new performance indicator: Age-of-Information (AoI). This performance indicator compromises between throughput and latency, indicating the length of time from data generation to the current moment. It can fully describe the real-time performance of sensor data. The smaller the AoI value, the higher the real-time performance of the data.

[0003] Since the data packets generated by sensors need to be queued according to the First-In-First-Out (FIFO) principle before transmission, the sampling rate of the sensor is crucial to ensure the timeliness of its data, that is, the AoI performance indicator. In addition, the adjustment of the sampling rate needs to fully consider the distribution characteristics of the wireless network service time, which represents the time it takes for a data packet to leave the sender queue and arrive at the receiver monitor. For example, when the average service time of the network is low, the sensor should not adopt a too high sampling rate, because this will cause it to generate data packets too frequently, causing queues and network congestion.

[0004] Existing sensor sampling rate control technologies for the Age of Information (AoI) indicator are mainly designed based on the two assumptions that the service time distribution is known or that the service time distribution is stationary. However, in actual communication systems, the service time distribution is often unknown and non-stationary. For example, every time a user joins or leaves the network, the service time distribution changes, and this change is difficult to accurately predict. In systems with unknown and non-stationary service time distribution, the design of sampling rate control is more critical and challenging, but existing technologies cannot solve this problem. Summary of the invention

[0005] The embodiment of the present invention provides a sensor adaptive sampling rate control method, which can significantly improve the timeliness of sensor data in an actual network. The specific technical solution is as follows:

[0006] In a first aspect of the present invention, a method for controlling a sensor's adaptive sampling rate is provided. The method comprises obtaining a short-term average information age a of the sensor in a previous control cycle. k-1 And the sampling rate u k-1 ,

[0007] At least according to the preset interruption threshold δ of the short-term average information age, the sampling rate u in the previous control cycle k-1 , the short-term average information age a k-1 Calculate the sampling rate u in the current control cycle k ,

[0008] The sensor data in the current control cycle is sampled according to the sampling rate in the current control cycle.

[0009] In a possible implementation of the embodiment of the present invention, a k-1 >δ is the sampling rate u in the current control cycle k The calculation of a k-1 <δ, the sampling rate u in the current control cycle k The calculation method is different.

[0010] In a possible implementation manner of the embodiment of the present invention, the interruption threshold δ based on at least the short-term average information age, the sampling rate u in the previous control cycle k-1 , the short-term average information age a k-1 Calculate the sampling rate u in the current control cycle k , specifically including calculating the sampling rate in the current control cycle according to the following:

[0011]

[0012] Among them, v kis the time-varying pseudo partial derivative parameter, λ1 and λ2 are weight parameters, W(.) is the Lambert W function, v k It is calculated as follows:

[0013]

[0014] Where μ is the weight parameter, v0 is the initial value of the time-varying pseudo partial derivative parameter, and θ is the small value judgment threshold set in the sensor estimation calculation, so that

[0015] In a possible implementation of the present invention, the short-term average information age a k-1 Calculated by the following formula:

[0016]

[0017] where Y k,j Y is the system time of the jth data packet successfully received in the kth cycle, calculated by the following formula: k,j =R k,j -G k,j ;

[0018] G k,j , R k,j denote the generation time and reception time of the jth successfully received data packet in the kth control cycle, respectively, where j=1, 2, ..., N, N is the number of successfully received data packets in the control cycle, and k represents the control cycle index;

[0019] X k,j is the time interval between successfully received packets in the kth control cycle. The time interval between successfully received packets is calculated according to the following formula:

[0020]

[0021] In a possible implementation manner of the embodiment of the present invention, the interruption threshold δ according to the short-term average information age, the sampling rate u in the previous control cycle k-1 , the short-term average information age a k-1 After calculating the sampling rate in the current control cycle, the control method further includes:

[0022] Calculate the upper limit of the sampling rate update based on the packet loss rate, the average value and variance of the service time;

[0023] When the interruption threshold δ according to the short-term average information age and the sampling rate u in the previous control cycle k-1 , the short-term average information age a k-1When it is calculated that the sampling rate in the current control period is greater than the sampling rate update upper limit, the sampling rate in the current control period is set as the sampling rate update upper limit.

[0024] In a possible implementation manner of the embodiment of the present invention, sampling the sensor data in the current control period according to the sampling rate in the current control period specifically includes:

[0025] The sensor data of the current control period is sampled at the sampling rate within the current control period set as the sampling rate update upper limit.

[0026] In a possible implementation manner of the embodiment of the present invention, the step of calculating the sampling rate update upper limit according to the packet loss rate, the average value and the variance of the service time specifically includes calculating the sampling rate update upper limit according to the following steps:

[0027]

[0028] in, They represent the packet loss rate in the kth control cycle, the mean value and variance of the service time, respectively, and are calculated as follows:

[0029]

[0030] Among them, α is the weight parameter.

[0031] The second aspect of the present invention further discloses a sensor adaptive sampling rate control device, the device comprising:

[0032] The acquisition unit is used to obtain the short-term average information age a of the sensor in the previous control cycle k-1 And the sampling rate u k-1 ,

[0033] A calculation unit is used to calculate the average information age according to the preset interruption threshold δ, the sampling rate u in the previous control cycle, and the average information age. k-1 , the short-term average information age a k-1 Calculate the sampling rate u in the current control cycle k ,

[0034] The sampling unit is used to sample the sensor data in the current control period according to the sampling rate in the current control period.

[0035] The third aspect of the present invention also discloses a computer device, which includes a processor and a memory; wherein the memory stores a computer program, and the computer program is suitable for being loaded by the processor and executing the steps of the sensor adaptive sampling rate control method as claimed in any one of the first aspects of the present invention.

[0036] The fourth aspect of the present invention further discloses a computer storage medium on which a computer program is stored. When the computer program is executed by a processor, the sensor adaptive sampling rate control method as described in any one of the first aspects of the present invention is implemented.

[0037] Beneficial effects of the present invention: The sensor adaptive sampling rate control method of the present invention comprises obtaining the short-term average information age a of the sensor in the previous control cycle k-1 And the sampling rate u k-1 , according to the interruption threshold δ of the short-term average information age and the sampling rate u in the previous control cycle k-1 , the short-term average information age a k-1 The sampling rate in the current control cycle is calculated, and the sensor data in the current control cycle is sampled according to the sampling rate in the current control cycle. The technical solution of the present invention is to calculate the average information age a of the sensor in the previous control cycle. k-1 And the sampling rate u k-1 The interruption threshold δ of the short-term average information age is used to calculate the adaptive sampling rate of the current control cycle, which ensures the adaptive adjustment of the sampling rate in units of control cycles. Compared with the existing technology, the real-time performance of sensor data can be guaranteed under non-stationary service time conditions, which significantly improves the timeliness of sensor data in actual networks. BRIEF DESCRIPTION OF THE DRAWINGS

[0038] Figure 1 A schematic flow chart of a sensor adaptive sampling rate control method according to an embodiment of the present invention;

[0039] Figure 2 is a sensor system model in an embodiment of the present invention;

[0040] Figure 3 is the evolution of AoI in each control cycle in the embodiment of the present invention;

[0041] Figure 4a is the average round trip time of each control cycle in two scenarios in the actual system experiment of the embodiment of the present invention

[0042] Figure 4b is the short-term average AoI of the ASC of the embodiment of the present invention and other existing technologies in scenario 1 in the actual system experiment

[0043] Figure 4c The short-term average AoI of the ASC proposed by the present invention and other existing technologies in scenario 2 in the actual system experiment

[0044] Figure 5 The figure is a schematic diagram of the structure of a computer device according to an embodiment of the present invention. DETAILED DESCRIPTION

[0045] The present invention is further described in detail below in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0046] It should be noted that the terms "first", "second", etc. are only used for descriptive purposes and cannot be understood as indicating or implying relative importance or implicitly indicating the number of the indicated technical features. Therefore, the features defined as "first" or "second" may explicitly or implicitly include at least one of the features. In the description of the present invention, the meaning of "plurality" is at least two, such as two, three, etc., unless otherwise clearly and specifically defined.

[0047] In the above description, the description with reference to the terms "one embodiment", "some embodiments", "example", "specific example", or "some examples" etc. means that the specific features, structures, materials or characteristics described in conjunction with the embodiment or example are included in at least one embodiment or example of the present invention. In this specification, the schematic representations of the above terms do not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials or characteristics described may be combined in any one or more embodiments or examples in a suitable manner. In addition, those skilled in the art may combine and combine the different embodiments or examples described in this specification and the features of the different embodiments or examples, without contradiction.

[0048] In the field of sensor data transmission, there are three main solutions for selecting and adjusting the sampling rate in the existing technology: the first is fixed sampling rate (FS), which does not require a confirmation feedback message for each data packet transmission success or failure, that is, ACK or NACK message, and the sensor always adopts a fixed sampling rate for sampling. When the statistical distribution of the service time is known, the FS technology can minimize the long-term average AoI by solving the optimal sampling rate. However, when the statistical distribution of the service time is unknown and non-stationary, since the FS technology does not dynamically adjust the sampling rate, it cannot effectively reduce the AoI and ensure the timeliness of the sensor data. The second is threshold waiting (TW): Unlike the FS technology, the TW technology requires a confirmation feedback message for each data packet transmission success or failure. Whenever a data packet is transmitted, that is, when the sending sensor receives the feedback message, the technology will wait for a period of time and sample and transmit the next data packet. The optimal waiting time can be determined by the reception time of the feedback message and the current instantaneous AoI of the monitor. TW technology can minimize the long-term average AoI under the condition of stable service time distribution, but it is difficult to cope with non-stationary service time. In addition, the confirmation messages required by TW technology will occupy additional communication resources, affecting the transmission of sensor data and their real-time performance. The third type is the Age Control Protocol (ACP)

[10] : ACP technology also requires confirmation message feedback for each data packet transmission success or failure. ACP estimates the congestion of the current sensor sender queue through the feedback of confirmation messages over a period of time, and selects a higher sampling rate as much as possible without causing congestion, thereby reducing the AoI of sensor data and ensuring its timeliness. ACP technology is relatively aggressive in selecting the sampling rate. Although ACP can achieve extremely low AoI when the service time distribution is stable, the AoI performance will fluctuate greatly when the service time distribution is non-stationary, and it cannot cope with the dynamic changes of the network.

[0049] Since the existing technologies cannot be used in scenarios where the service time distribution is unknown and non-stationary, the present invention focuses on this pain point and mainly solves the following two specific shortcomings.

[0050] Disadvantage 1: It is difficult for existing technologies to guarantee short-term average AoI under non-stationary service time conditions. Existing technologies often focus on minimizing the long-term average AoI, that is, the system average AoI when the number of received data packets tends to infinity. However, when the service time distribution is non-stationary, that is, it is constantly changing, the system should pay more attention to the short-term average AoI rather than the long-term average AoI. The short-term average AoI is defined as the average AoI of the data packets recently received by the receiving end monitor. This is because under non-stationary service time, the short-term average AoI will produce large fluctuations. Therefore, even if the long-term average AoI remains at a low level, the short-term average AoI will still be too high during certain periods of poor service, resulting in system failures. Under non-stationary service time conditions, it is particularly important to guarantee the short-term average AoI of the system during the entire operation period.

[0051] Disadvantage 2: The existing technology lacks a clear model to describe the relationship between the sampling rate and the short-term average AoI. The existing technology is usually based on queuing theory, and characterizes the relationship between the sampling rate and the long-term average AoI under the condition of known service time distribution. For example, when the service time is exponentially distributed, a closed-form expression between the long-term average AoI and the sampling rate is given. However, these expressions will no longer apply when the service time is unknown and non-stationary, and cannot be used to further adjust the sampling rate of the sensor. Therefore, under the condition that the service time is unknown and non-stationary, it is necessary to characterize the relationship between the sensor sampling rate and the short-term average AoI without relying on any prior information assumptions about the service time, so as to further adjust the sampling rate.

[0052] In response to the above technical problems in the field of sensor signal processing, the technical solution of the present invention proposes a sensor adaptive sampling rate control (Adaptive Sampling Control, ASC) technology, which can be effectively applied to actual communication systems with unknown and non-stationary service time. In view of the difficulty of the existing technology in ensuring the short-term average AoI under non-stationary service time conditions, the ASC technology proposed in the present invention adjusts the sampling rate of the sensor in units of control cycle. The present invention does not require any prior knowledge and assumptions about the distribution of service time during design, so that the proposed ASC technology can be used in any actual scenario.

[0053] Embodiment 1 of the present invention discloses a method for controlling a sensor's adaptive sampling rate. Figure 1 As shown, the sensor adaptive sampling rate control method includes:

[0054] Step S1, obtain the short-term average information age a of the sensor in the previous control cycle k-1 And the sampling rate u k-1 ;

[0055] Step S2: according to the interruption threshold δ of the short-term average information age and the sampling rate u in the previous control cycle, k-1 , the short-term average information age a k-1 Calculate the sampling rate within the current control cycle;

[0056] Step S3: sampling the sensor data in the current control period according to the sampling rate in the current control period.

[0057] In the embodiment of the present invention, the average information age a of the sensor in the previous control cycle is k-1 And the sampling rate u k-1 The interruption threshold δ of the short-term average information age is used to calculate the adaptive sampling rate of the current control cycle, which ensures the adaptive adjustment of the sampling rate in units of control cycles. Compared with the existing technology, the real-time performance of sensor data can be guaranteed under non-stationary service time conditions, which significantly improves the timeliness of sensor data in actual networks.

[0058] In one implementation of the embodiment of the present invention, the interruption threshold δ based on at least the short-term average information age, the sampling rate u in the previous control cycle k-1 , the short-term average information age a k-1 Calculate the sampling rate u in the current control cycle k , specifically including calculating the sampling rate in the current control cycle according to the following:

[0059]

[0060] Among them, v k is the time-varying pseudo partial derivative parameter, λ1 and λ2 are weight parameters, W(.) is the Lambert W function, v k It is calculated as follows:

[0061]

[0062] Where μ is the weight parameter, v0 is the initial value of the time-varying pseudo partial derivative parameter, and θ is the small value judgment threshold set in the sensor estimation calculation, so that

[0063] In a possible implementation of the embodiment of the present invention, a k-1 >v, the sampling rate u in the current control cycle k The calculation of a k-1 <δ, the sampling rate u in the current control cycle k The calculation method is different.

[0064] In a possible implementation of the present invention, the short-term average information age a k-1 Calculated by the following formula:

[0065]

[0066] Among them, Y k,k Y is the system time of the jth data packet successfully received in the kth cycle, calculated by the following formula: k,j =R k,j -G k,j ;

[0067] G k,j , R k,j denote the generation time and reception time of the jth successfully received data packet in the kth control cycle, respectively, where j=1, 2, ..., N, N is the number of successfully received data packets in the control cycle, and k represents the control cycle index;

[0068] X k,j is the time interval between successfully received packets in the kth control cycle. The time interval between successfully received packets is calculated according to the following formula:

[0069]

[0070] In a possible implementation manner of the embodiment of the present invention, the interruption threshold δ according to the short-term average information age, the sampling rate u in the previous control cycle k-1 , the short-term average information age a k-1 After calculating the sampling rate in the current control cycle, the control method further includes:

[0071] Calculate the upper limit of the sampling rate update based on the packet loss rate, the average value and variance of the service time;

[0072] When the interruption threshold δ according to the short-term average information age and the sampling rate u in the previous control cycle k-1 , the short-term average information age a k-1 When it is calculated that the sampling rate in the current control period is greater than the sampling rate update upper limit, the sampling rate in the current control period is set as the sampling rate update upper limit.

[0073] In a possible implementation manner of the embodiment of the present invention, sampling the sensor data in the current control period according to the sampling rate in the current control period specifically includes:

[0074] The sensor data of the current control period is sampled at the sampling rate within the current control period set as the sampling rate update upper limit.

[0075] In a possible implementation manner of the embodiment of the present invention, the step of calculating the sampling rate update upper limit according to the packet loss rate, the average value and the variance of the service time specifically includes calculating the sampling rate update upper limit according to the following steps:

[0076]

[0077] in, They represent the packet loss rate in the kth control cycle, the mean value and variance of the service time, respectively, and are calculated as follows:

[0078]

[0079] Among them, α is the weight parameter.

[0080] In a possible implementation manner of the embodiment of the present invention, the preset interruption threshold is related to at least one of the sensor signal change amplitude, sensor system accuracy, and sensor system response time, or the interruption threshold is set to a fixed value, an adaptive value, or a multi-level threshold.

[0081] In a possible implementation of the embodiment of the present invention, a k-1 >δ is the sampling rate u in the current control cycle k The calculation of a k-1 <δ, the sampling rate u in the current control cycle k The calculation method is different.

[0082] In order to illustrate that the sampling rate in the current control cycle obtained based on the above calculation can minimize the long-term average AoI of the system under the constraint of the interruption threshold δ in the short-term average information age in units of control cycles, and thus achieve the real-time performance of sensor data under non-stationary service time conditions, the timeliness of sensor data in the actual network is significantly improved, as described in detail below.

[0083] The present invention calculates the adaptive sampling rate that can minimize the system's short-term average AoI from the unknown and non-stationary service time of the sensor system as follows.

[0084] See attached Figure 2 is the sensor system model in the embodiment of the present invention. The embodiment of the present invention considers Figure 2In the sensing system shown, the sensor at the transmitting end samples the physical process it observes, and encapsulates the sampled data into data packets and sends them to the monitor at the receiving end via the transmission network. Considering that the sensing device is usually connected to the monitor through the UDP / TCP protocol and adopts a FIFO queue, the present invention assumes that the generated data packets will be stored in a FIFO queue and transmitted on a first-come, first-served basis. In the proposed ASC technology, each control cycle consists of N successfully received data packets. At the end of each control cycle, the receiving end determines the sampling rate that the sending end should adopt in the next control cycle, and feeds back to the sending end through the UDP / TCP protocol. Whether each control cycle ends or not is determined by the counter at the receiving end. Whenever the counter reaches N, the current control cycle ends and the next control cycle starts.

[0085] Since the service time of the transmission network in the actual system is often unknown and non-stationary, the present invention considers a general piecewise short-term stationary service time model when constructing the system model. In this model, the service time distribution can change between different control cycles. In addition, the system model considered by the present invention does not make any prior assumptions on the expression of the service time distribution.

[0086] Since the sensor system considered in the present invention adopts AoI as a performance indicator, the mathematical expression of the performance indicator will be further described below. The present invention uses k = 1, 2, 3, ... to represent the index of each control cycle, u k The sampling rate of each control cycle is One data packet is sampled per second. Figure 2 As shown, G k,j , R k,j They represent the generation time and reception time of the jth successfully received data packet in the kth control cycle, respectively, where j = 1, 2, …, N.

[0087] The present invention is used and Represents the index of the latest received data packet. The instantaneous AoI at time t can be defined as:

[0088] a(t)=tG k(t),j(t) (1)

[0089] In order to further characterize the short-term average AoI in each control cycle, let Y k,j is the system time of the jth data packet successfully received in the kth cycle, which can be expressed as

[0090] Y k,j =R k,j -G k,j (2)

[0091] Let X k,jis the generation time interval between successfully received data packets in the kth control cycle, which can be expressed as:

[0092]

[0093] like Figure 3 As shown, through the area Q k,j The short-term average AoI of the system in the kth control cycle can be obtained, which can be expressed as:

[0094]

[0095] Where Y k,0 =Y k-1,N It is not difficult to find that the receiver monitor can be based on (2)-(4) and only according to the receiving time R of each data packet k,j and the generation time G k,j , calculate the short-term average AoIa for each cycle k Among them, R k,j can be obtained directly at the receiving end, and G k,j Can be encapsulated in the data packet generated by the sender.

[0096] The existing sampling rate control technology mainly minimizes the long-term average AoI of the system under the premise of stable service time. However, the service time of the transmission network is usually non-stationary in actual scenarios. The idea of ​​minimizing the long-term average AoI cannot guarantee the short-term average AoI of the system, resulting in large jitters of the short-term average AoI or even system stagnation. In order to ensure the real-time performance of sensor data collection under non-stationary service time conditions, the present invention reduces the interruption probability of the short-term average AoI through the following problems:

[0097]

[0098] u k ←u1,,u k-1 ,,a1,,a k-1 .(5c)

[0099] Where K is the number of control cycles, is the interrupt judgment function. When a k >δ is equal to 1, otherwise it is 0. δ is the interruption threshold of the short-term average AoI. Objective (5a) aims to minimize the short-term AoI interruption probability for the target threshold δ. In addition, constraint (5b) indicates that the sampling rate u of the current control cycle k It can only be determined by the sampling rate of the previous control cycle and the short-term average AoI, which meets the actual system requirements.

[0100] The present invention proposes a sensor adaptive sampling rate control (Adaptive Sampling Control, ASC) technology. The technology aims to minimize the interruption probability of the system's short-term average AoI, that is, the above-mentioned problem (5), and ensures the real-time performance of the sensor data collected under non-stationary service time conditions. At the end of each control cycle, the ASC technology proposed in the present invention will determine the sampling rate of the sensor in the next control cycle based on the system's historical sampling rate and short-term average AoI. First, it is necessary to establish a dynamic linear data model based on the historical collection data to characterize the mathematical relationship between the sampling rate and the short-term average AoI. Next, the optimal sampling rate for the next control cycle is determined based on the dynamic linear data model. The proposed ASC technology will be specifically explained from the two aspects of dynamic linear data model and sampling rate update, and its deployment in the actual system will be further indicated.

[0101] Dynamic Linear Data Model

[0102] ASC technology calculates the short-term average AoIa in each control cycle k and sampling rate u k The unknown dynamic relationship between them is transformed into the following equivalent dynamic linear data model

[12]

[13]

[0103] Δa k =v k Δu k (6)

[0104] Where Δu k =u k -u k-1 , Δa k =a k -a k-1 , v k is the time-varying pseudo-partial derivative (PPD) parameter. Based on (6), a k and u k The complex relationship between Δu k and Δa k Through the time-varying v k The non-stationary and unknown service time characteristics of the considered sensor system can be expressed as v k Dynamic changes of v k Then we can establish a k and u k The relationship between them is estimated as follows:

[0105]

[0106] In the formula, μ is the weight parameter.

[0107] In order to further improve the estimation accuracy and ensure stability, the proposed ASC technology follows the following reset scheme to deal with the erroneous estimation when the service time suddenly changes:

[0108]

[0109] Where v0 is the initial value of PPD, and θ is a small positive number to ensure Δu k ≠0.

[0110] Sampling rate update

[0111] Using the estimation formula of formula (7), we can establish that a in formula (6) k and u k Next, we will further solve the sampling rate control problem in equation (5) based on this relationship, with the goal of minimizing the short-term average AoI interruption probability.

[0112] In the embodiment of the present invention, when ɑk-1>δ, the sampling rate u in the current control cycle k The calculation of the sampling rate u in the current control cycle when ɑk-1<δ k The calculation method is different.

[0113] The sampling rate for each control cycle can be updated as follows:

[0114]

[0115] Where λ1 and λ2 are weight parameters, and W(.) is the Lambert W function.

[0116] In order to solve the problem that the existing sensor sampling rate calculation and adjustment algorithms cannot be used in unknown and non-stationary service time scenarios, the embodiments of the present invention firstly construct a dynamic linear data model between the sampling rate and the short-term average AoI. This model does not require a priori assumptions about the service time distribution and can be used in any practical system. Based on this dynamic linear data model, the embodiments of the present invention propose an ASC algorithm with the goal of minimizing the interruption probability and provide a dynamic adjustment control scheme for the sampling rate. The technical scheme of the embodiments of the present invention can significantly reduce the interruption probability of the short-term average AoI of the sensor system under the conditions of unknown and non-stationary service time, and the performance improvement compared to the existing technology reaches 100%, providing an efficient sampling control technology for the actual sensor system.

[0117] In order to further illustrate the sensor adaptive sampling rate control method in the embodiment of the present invention and its reduction of the short-term average AoI interruption probability under unknown and non-stationary service time conditions, the following is a detailed description from the perspective of deployment and verification in an actual system.

[0118] The following is a description of the deployment of the adaptive sampling control (ASC) technology in an embodiment of the present invention:

[0119] Common input parameters for all control cycles: interrupt threshold δ, weight coefficients α, μ, λ1, λ2, θ, number of packets per control cycle N

[0120] Input of the kth control cycle:

[0121] 1: Packet reception and generation time R k,j ,G k,j

[0122] The kth control cycle output:

[0123] 2: Sampling rate u k

[0124] 3: null

[0125] 4: Initialization: Δu0, Δa0, v0,

[0126] 5: for each control cycle k=1,2,3,…do

[0127] 6: Update PPD

[0128] 7:

[0129] 8:

[0130] 9: end if

[0131] 10:ifa k-1 >δthen

[0132] 11: Update sampling rate

[0133] 12:else

[0134] 13: Update sampling rate

[0135] 14:end if

[0136] 15:

[0137] 16: will u k Feedback to the sender, inverse mapping u k ←log(u k )

[0138] 17: Receive N packets from the sender

[0139] 18: Calculate the average of the control cycle

[0140] 19: Calculation Update sampling rate upper limit

[0141] 20: end for

[0142] The pseudo code of ASC technology is given above. The pseudo code can be deployed in Figure 1 The receiver monitor of the system shown in the figure. The sensor and the monitor can be connected via the UDP / TCP protocol. In the initialization phase, the sensor can adopt a random sampling rate in the first two control cycles, and the receiver can calculate the initialization parameters required by the algorithm based on the first two control cycles. Next, whenever the receiver receives N packets, the short-term average AoI and the upper limit of the sampling rate of the current control cycle are calculated according to equations (4), (10), (11), (12) and (13), and then the PPD of the next control cycle is calculated according to equations (7) and (8). Finally, the sampling rate of the next control cycle is calculated according to equations (9) and (14) and fed back to the sender. After receiving the update, the sender adopts the new sampling rate, and the cycle continues.

[0143] In order to demonstrate the beneficial effects of the technical solution of the present invention, the short-term average AoI interruption probability of the proposed ASC technology is compared with that of the existing FS, TW and ACP technologies under the condition of unknown service time and non-stationary conditions through actual system experiments. Since the existing FS, TW and ACP technologies have been introduced above, they will not be repeated here.

[0144] The present invention uses three micro hosts to build an experimental platform. These micro hosts are equipped with Ubuntu 20.04 operating system, Intel Core i7-4510U CPU and 4G memory. The data between the micro hosts is transmitted through the Atheros AR9382 wireless network card, adopting the IEEE 802.11 link layer and physical layer protocols. Among them, one micro host is used as an access point (AP) and the other two are used as terminals (STA). In the experiment, AP represents the receiving end monitor, one STA represents the sensor, and the other STA will generate a data stream of random size, representing the unknown impact of the remaining devices in the network on the sensor link. Each data packet generated by the sensor is 25 bytes in size and is transmitted by the UDP transport layer protocol. Finally, each control cycle contains N = 200 data packets, and the weight coefficient of the proposed ASC technology is as follows.

[0145]

[0146] This experiment considers two different scenarios. In scenario 1, another STA used to generate data flow provides 0.1Mbps traffic, which represents a scenario with light load and basically stable service time distribution. In scenario 2, the traffic generated by another STA varies randomly from 0 to 0.8Mbps, which represents a scenario with dynamic changes in network load and non-stable service time distribution. In order to better reflect the difference in service time in the two considered scenarios, Figure 3 The average round trip time for each control cycle is shown in . Since the average service time cannot be directly measured in the actual system, we measure the average round trip time in the experiment, which is highly correlated with the average service time. It can be seen that in scenario 1, the average round trip time of different control cycles is almost 5ms, while in scenario 2, the average round trip time changes dynamically between 10ms and 20ms.

[0147] Figure 4b The short-term average AoI of each control cycle in scenario 1 is shown. Due to the low average RTT in scenario 1, a relatively strict interruption threshold δ = 10ms is adopted in this experiment. As shown in Figure 4, around the 80th, 200th, and 300th control cycles, the short-term average AoI of the ASC technology fails to reach the expected interruption threshold. This is because the RTT in these time periods is high, such as Figure 3 As shown in the figure, the short-term average AoI cannot reach 10ms. In addition, since the FS technology adopts a fixed sampling rate, when the service time distribution changes, that is, when the average RTT changes, its short-term average AoI deviates significantly from the expected interruption threshold. Finally, although the TW and ACP technologies can achieve extremely low short-term average AoI when the average RTT is low, their short-term average AoI increases significantly when the average RTT increases, far exceeding the target interruption threshold. This is because, on the one hand, the TW and ACP technologies rely on the confirmation message feedback of each data packet, which generates a high overhead in the actual system. On the other hand, ACP will aggressively select a higher sampling rate. Although such a strategy can achieve a very small short-term average AoI under stable service time conditions, it is very easy to cause network congestion when the service time changes.

[0148] Figure 4c The short-term average AoI of each control cycle in scenario 2 is shown. Since the average RTT of each control cycle in scenario 2 is high and changes dynamically, δ = 30ms is adopted as the interruption threshold in the experiment. Figure 5It can be seen that the ASC technology proposed in the present invention can effectively control the short-term average AoI of the sensor below the interruption threshold under non-stationary and unknown service time distribution conditions. However, the short-term average AoI of the existing FS, TW and ACP technologies deviates significantly from the interruption threshold required by the system when the service time is non-stationary, and the real-time performance of the sensor data cannot be guaranteed. Finally, Table 3 gives the short-term average AoI interruption probability of different sampling technologies in the experiment. Compared with the prior art, the ASC technology proposed in the present invention can reduce the interruption probability by about 100% and significantly improve the timeliness of sensor data in the actual network. The specific short-term average AoI interruption probabilities of different sampling technologies are shown as follows:

[0149]

[0150] Another embodiment of the present invention further discloses a sensor adaptive sampling rate control device, characterized in that the device comprises:

[0151] The acquisition unit is used to obtain the short-term average information age and sampling rate of the sensor in the previous control cycle.

[0152] A calculation unit, configured to calculate the sampling rate u in the current control cycle according to at least a preset interruption threshold of the short-term average information age, the sampling rate in the previous control cycle, and the short-term average information age. k ,

[0153] The sampling unit is used to sample the sensor data in the current control period according to the sampling rate in the current control period.

[0154] Another embodiment of the present invention further discloses a computer storage medium having a computer program stored thereon. When the computer program is executed by a processor, the sensor adaptive sampling rate control method as described in any one of the first aspects of the present invention is implemented.

[0155] Embodiment 4

[0156] In one embodiment, a computer device is provided. The computer device may be a server, and its internal structure diagram may be as follows: Figure 5As shown. The computer device includes a processor, a memory, a network interface and a database connected via a system bus. Among them, the processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, a computer program and a database. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. When the computer program is executed by the processor, it implements the sensor adaptive sampling rate control method of any implementation method of the above-mentioned embodiment. Since any implementation method of each specific step is described in detail in the above-mentioned embodiment and each implementation method, it will not be repeated here.

[0157] Those skilled in the art will understand that Figure 5 The structure shown in the figure is only a block diagram of a part of the structure related to the solution of the present application, and does not constitute a limitation on the computer device to which the solution of the present application is applied. The specific computer device may include more or fewer components than those shown in the figure, or combine certain components, or have a different arrangement of components.

[0158] The sensor adaptive sampling rate control method of the present invention, as well as the computer device and computer storage medium of the above-mentioned embodiments, all calculate the adaptive sampling rate of the current control cycle through the average information age of the sensor in the previous control cycle and the sampling rate and the interruption threshold of the short-term average information age, thereby ensuring the adaptive adjustment of the sampling rate in units of control cycles. Compared with the prior art, the real-time performance of sensor sensing data can be guaranteed under non-stationary service time conditions, which significantly improves the timeliness of sensor data in actual networks.

[0159] The device embodiments described above are merely illustrative, wherein the units described as separate components may or may not be physically separated, and the components displayed as units may or may not be physical units, that is, they may be located in one place, or they may be distributed on multiple network units. Some or all of the modules may be selected according to actual needs to achieve the purpose of the scheme of this embodiment. Ordinary technicians in this field can understand and implement it without paying creative labor.

[0160] Through the description of the above implementation methods, those skilled in the art can clearly understand that each implementation method can be implemented by means of software plus a necessary general hardware platform, and of course, it can also be implemented by hardware. Based on this understanding, the above technical solution is essentially or the part that contributes to the prior art can be embodied in the form of a software product, and the computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, a disk, an optical disk, etc., including a number of instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to execute the methods described in each embodiment or some parts of the embodiments.

[0161] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present invention.

Claims

1. A sensor adaptive sampling rate control method, characterized in that: The method comprises, Get the short-term average information age a of the sensor in the previous control cycle k-1 And the sampling rate u k-1 , At least according to the preset interruption threshold δ of the short-term average information age, the sampling rate u in the previous control cycle k-1 , the short-term average information age a k-1 Calculate the sampling rate u in the current control cycle k , The sensor data in the current control cycle is sampled according to the sampling rate in the current control cycle.

2. A sensor adaptive sampling rate control method as claimed in claim 1, characterized in that: a k-1 >δ, the sampling rate u in the current control cycle k The calculation of a k-1 <δ The sampling rate u in the current control cycle k The calculation method is different.

3. A sensor adaptive sampling rate control method as claimed in claim 2, characterized in that: The interruption threshold δ based on at least the short-term average information age, the sampling rate u in the previous control cycle k-1 , the short-term average information age a k-1 Calculate the sampling rate u in the current control cycle k , specifically including calculating the sampling rate in the current control cycle according to the following: Among them, v k is the time-varying pseudo partial derivative parameter, λ1 and λ2 are weight parameters, W(.) is the Lambert W function, v k It is calculated as follows: Where μ is the weight parameter, v0 is the initial value of the time-varying pseudo partial derivative parameter, and θ is the small value judgment threshold set in the sensor estimation calculation, so that Δu k ≠0.

4. The sensor adaptive sampling rate control method according to claim 1, characterized in that: The short-term average information age a k-1 Calculated by the following formula: Among them, Y k，j Y is the system time of the jth data packet successfully received in the kth cycle, calculated by the following formula: k，j =R k，j -G k，j ; G k，j , R k，j denote the generation time and reception time of the jth successfully received data packet in the kth control cycle, respectively, where j=1, 2, ..., N, N is the number of successfully received data packets in the control cycle, and k represents the control cycle index; X k，j is the time interval between successfully received data packets in the kth control cycle, and is calculated according to the following formula:

5. The sensor adaptive sampling rate control method according to claim 1, characterized in that: In the interruption threshold δ according to the short-term average information age, the sampling rate u in the previous control cycle k-1 , the short-term average information age a k-1 After calculating the sampling rate in the current control cycle, the control method further includes: Calculate the upper limit of the sampling rate update based on the packet loss rate, the average value and variance of the service time; When the interruption threshold δ according to the short-term average information age and the sampling rate u in the previous control cycle k-1 , the short-term average information age a k-1 When it is calculated that the sampling rate in the current control period is greater than the sampling rate update upper limit, the sampling rate in the current control period is set as the sampling rate update upper limit.

6. The sensor adaptive sampling rate control method according to claim 5, characterized in that: The sampling of sensor data in the current control cycle according to the sampling rate in the current control cycle specifically includes: The sensor data of the current control period is sampled at the sampling rate within the current control period set as the sampling rate update upper limit.

7. The sensor adaptive sampling rate control method according to claim 5, characterized in that: The method of calculating the sampling rate update upper limit according to the packet loss rate, the average value and the variance of the service time specifically includes calculating the sampling rate update upper limit according to the following steps: in, They represent the packet loss rate in the kth control cycle, the mean value and variance of the service time, respectively, and are calculated as follows: Among them, α is the weight parameter.

8. A sensor adaptive sampling rate control device, characterized in that: The device comprises: The acquisition unit is used to obtain the short-term average information age a of the sensor in the previous control cycle k-1 And the sampling rate u k-1 , A calculation unit is used to calculate the average information age according to the preset interruption threshold δ, the sampling rate u in the previous control cycle, and the average information age. k-1 , the short-term average information age a k-1 Calculate the sampling rate u in the current control cycle k , The sampling unit is used to sample the sensor data in the current control period according to the sampling rate in the current control period.

9. A computer device, characterized in that: The computer device comprises a processor and a memory; wherein the memory stores a computer program, and the computer program is suitable for being loaded by the processor and executing the steps of the sensor adaptive sampling rate control method according to any one of claims 1 to 8.

10. A computer storage medium, characterized in that: A computer program is stored thereon, and when the computer program is executed by a processor, the sensor adaptive sampling rate control method according to any one of claims 1 to 8 is implemented.

Images