Micro-energy driven task adaptive matching and scheduling control system

By designing a micro-energy-driven task adaptive matching and scheduling control system in a passive IoT terminal system, using refined task energy consumption models and multi-task scheduling mechanisms, the load problems caused by energy management are solved and efficient Energy utilization and long-term operation.

CN119002432APending Publication Date: 2024-11-22UNIV OF ELECTRONICS SCI & TECH OF CHINA
View PDF 0 Cites 2 Cited by

Patent Information

Application Number
CN202411175802.6
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-08-26
Publication Date
2024-11-22

AI Technical Summary

Technical Problem

Prior art In passive IoT terminal systems, energy perception and local computing of energy management lead to additional capacity load and overhead, making it difficult to achieve fine-grained task planning and long-term operation, especially in scarce environmental energy, and it is difficult to continuously operate.

Method used

A micro-energy-driven task adaptive matching and scheduling control system is designed, including energy state detection circuit, task scheduler, task matching circuit and energy supply control circuit. Through refined task energy consumption model and multi-task scheduling mechanism, dynamic Adjust task priority and scheduling strategy to achieve task matching and scheduling with near-zero power consumption.

Benefits of technology

It improves the energy utilization efficiency of passive systems, reduces the risk of task blocking, and realizes efficient energy management in complex and changing environments, ensuring the long-term and stable operation of the terminal system.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN119002432A_ABST
    Figure CN119002432A_ABST
Patent Text Reader

Abstract

The invention relates to a micro-energy driven task adaptive matching and scheduling control system. The system comprises an energy state detection circuit, a task scheduler, a task matching circuit and an energy supply control circuit, the energy state detection circuit is connected with the energy storage capacitor bank, and is used for detecting capacitor voltage and converting a detection result into a binary value; the task scheduler schedules the tasks in the task group according to a certain sequence and matches the tasks through the task matching circuit, when the tasks are successfully matched, the energy supply control circuit is enabled, after the tasks are successfully executed, the energy supply control circuit is fed back to stop supplying energy, and the task scheduler is informed of successful task execution. In order to solve the problem of task congestion of a terminal system in a complex and changeable environment, a near-zero-power-consumption adaptive task scheduler is designed, so that a task can adaptively update a priority and a scheduling strategy according to the change of environment energy, the energy management of a terminal is decoupled from a high-power-consumption MCU (Microprogrammed Control Unit) and a high-power-consumption ADC (Analog to Digital Converter), and the problem of high overhead is fundamentally solved.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention relates to the technical field of Internet of Things, and in particular to a micro-energy driven task adaptive matching and scheduling control system. Background Art

[0002] The goal of the Internet of Things is to achieve the interconnection of all things. By widely deploying spatially distributed devices with embedded identification, sensing and driving functions, environmental perception and data acquisition can be achieved. In order to adapt to the exponentially growing deployment scale and diversified application scenario requirements of the Internet of Things, more and more IoT terminal systems tend to integrate with environmental energy harvesting technologies, resulting in passive sensing and computing systems (referred to as passive systems). However, the energy consumption requirements of the terminal change dynamically with the task status. In order to ensure the normal operation of the system, fine-grained energy management must be frequently performed. According to a series of operations such as energy detection, task energy consumption calculation and energy allocation included in energy management, the passive system is given additional task load and energy overhead, reaching an average of 51.2% of the collected energy, which makes it difficult to improve the actual energy utilization efficiency.

[0003] At present, the existing technology proposes a dynamic polling scheme to reduce energy management overhead. It can continuously adjust the polling frequency according to the current energy collection efficiency and reduce the energy consumption of energy management by reducing redundant operations. However, the actual reduction of energy management power consumption by this scheme is limited, and it still cannot continue to operate in relatively weak environmental energy such as RF power supply. In addition, other existing works use triggering methods to activate the device only when the energy reaches a threshold, thereby achieving the purpose of reducing energy sampling energy consumption. This scheme simplifies some of the high-power consumption links, but at the same time it also sacrifices management granularity and affects the dynamic planning of task execution.

[0004] In summary, in order to use scarce environmental energy to maintain the long-term operation of passive terminals, it is necessary to dynamically adjust task execution based on the currently collected energy. However, the energy perception and local computing performed by energy management bring additional capacity load and overhead, which becomes the bottleneck of the energy efficiency of passive terminals. Existing low-power energy management technologies cannot achieve fine-grained task planning. Summary of the invention

[0005] The purpose of the present invention is to overcome the shortcomings of the prior art and provide a micro-energy drive task adaptive matching and scheduling control system to solve the deficiencies of the prior art.

[0006] The object of the present invention is achieved by the following technical solutions: a micro-energy driven task adaptive matching and scheduling control system, the system is connected to the gateway communication and connected to the task load, and includes an energy state detection circuit, a task scheduler, a task matching circuit and an energy supply control circuit;

[0007] The energy state detection circuit is connected to the energy storage capacitor group, detects the capacitor voltage and converts the detection result into a binary value, obtains the energy demand of the task team's energy storage capacitor group by pre-configuring the load task demand, maps the demand to the voltage constraint of each capacitor, and saves the voltage constraints of all tasks to the calculation-free task receiving circuit to form a task group; the task scheduler schedules the tasks in the task group in a certain order and matches them through the task matching circuit, when the task is matched successfully, the energy supply control circuit is enabled, and when the task is executed successfully, the feedback energy supply control circuit stops supplying energy and notifies the task scheduler that the task is executed successfully.

[0008] The energy state detection circuit includes a plurality of comparators with different reference voltages to detect the voltage state of each capacitor in the energy storage capacitor group. The output ends of the plurality of comparators with different reference voltages are connected to a priority encoder. The detection results are encoded into a code of several binary bits by the priority encoder to obtain the encoded voltage states of all capacitors.

[0009] The computation-free task receiving circuit includes a shift register, a register group management unit and a register group; the shift register is used to receive gateway task information, the register group is composed of multiple independent registers, and is used to split and store the control stream sent by the gateway according to the various parts of the task information, and the register group management unit is used to write and erase data of the target register in the controller register group.

[0010] The storage space of the shift register includes four parts: task energy cost, priority, scheduling strategy and task command;

[0011] The task energy cost: by establishing a refined task energy consumption model on the gateway to cover the total power consumption of all components in the task load, the theoretical cost of the task is calculated and sent to the system in a remote manner;

[0012] The priority: specifies the priority of a task when it is scheduled. The higher the priority value, the higher the priority.

[0013] The scheduling strategy: specifies whether the task is executed once or cyclically. When the occupied space value of the scheduling strategy is 1, it indicates that the task is in a one-time execution mode. When the occupied space value of the scheduling strategy is n, it indicates that the task is cyclically executed n times. When the occupied space value of the scheduling strategy is 0, it indicates that the task is in an infinite loop execution mode.

[0014] The task command specifies the target task corresponding to the gateway task information.

[0015] The task energy consumption model includes a task load power model and a voltage stabilization loss model;

[0016] The task load power model includes the frequency-dependent active power consumption P dep , frequency-independent active power consumption P ind And static power consumption P s , that is, the power consumption of the task load is P task =P s +h(P ind +P dep ), the coefficient h indicates whether the task load currently consumes active power;

[0017] The voltage stabilization loss model describes the loss of the voltage stabilization circuit. Assume that the switching power P of the voltage stabilization circuit is K , the input impedance and output impedance are R W1 and R W2 , and the output impedance of the capacitor is R C , the power loss caused by the voltage stabilization circuit at time t is P l (t) = I C (t) 2 ·(R W1 +R C )+I l (t) 2 ·R W2 +P K , I C (t) is the instantaneous current output by the capacitor, I l (t) is the instantaneous current of the task load overhead.

[0018] The task scheduler dynamically sets the priorities of multiple tasks according to the task energy consumption and the system energy state, and matches the appropriate task scheduling order for the current system energy state;

[0019] By dividing the task energy cost into K disjoint and continuous intervals, the i-th interval is expressed as <lb i ,ub i >,lb i Indicates the lower boundary of the interval, ub i represents the upper boundary of the interval. Each interval has a different impact on the task priority. Let the impact of the i-th interval on the task priority be m i , task x t The energy cost of belongs to the i-th interval, then χ t The new priority can be expressed by the formula P t ′=Min{Max{P t +m i , 1}, P max} is calculated, where P t Represents the task x t The old priority, P t ′ represents the task χ tThe new priority, P max Indicates the highest priority of the task, and the lowest priority of the task is 1.

[0020] The success rate of task matching η task To characterize the environmental energy status, the calculation formula is: Among them, Count success Indicates the number of successful task matching, Count failure Indicates the number of task matching failures;

[0021] If the current energy state of the system meets the energy requirements of the task, the task is successfully matched, and Count success Add 1 and update η task If it is not satisfied, the task matching fails and the system continues to store energy. When the task matching fails, in order to avoid task blocking in the system, if the system waits for energy storage time to exceed the upper threshold T wait , then Count failure Add 1 and update η task , otherwise, the system has stored enough energy within the threshold time, Count success Add 1 and update η task , when η task When it increases, it means that the environmental energy conditions are getting better, otherwise, it means that the environmental energy conditions are getting worse.

[0022] The task matching circuit includes numerical comparators with the same number of capacitors as the energy storage capacitor group, the input of each numerical comparator is connected to a capacitor in the energy storage capacitor group, the outputs of all numerical comparators are connected to the first AND gate, and the task energy cost includes N parts, where N is the number of capacitors;

[0023] The coding voltage requirements of the N parts are compared one by one with the coding voltage states of the capacitors through a numerical comparator to obtain N comparison judgment results. Finally, all the comparison judgment results are input into the first AND gate to obtain the task matching judgment result. When the matching judgment output is a high level, it indicates that the current system energy state meets the energy requirements of the target task, otherwise, it indicates that it does not meet the requirements.

[0024] The condition that the current system energy state meets the energy requirement of the target task is: the energy of the energy storage capacitor bank is greater than or equal to the task energy cost; at this time, the two outputs F in each numerical comparator are A>B and F A=B The first OR gate is input, and the output of the first OR gate is used as a comparison result, wherein A represents the binary coded voltage state of the energy storage capacitor, and B represents the binary coded voltage requirement of the component corresponding to the task energy consumption.

[0025] The energy supply control circuit includes a single-pole double-throw S, the input of the single-pole double-throw S is connected to the energy storage capacitor group, one connector of the single-pole double-throw S is connected to an LDO, and the LDO provides stable energy supply for the task, and the other connector is connected to the output of the second AND gate, one of the input ends of the second AND gate is connected to the output end of the second OR gate, and the other input end is connected to the task scheduler; one of the input ends of the second OR gate is connected to the task matching circuit, and the other input end is connected to the output end of the third AND gate; one of the input ends of the third AND gate is connected to the task scheduler, and the other input end is connected to the LDO.

[0026] The present invention has the following advantages:

[0027] 1. Split the task load circuit, divide the power consumption of each component into frequency-related active power consumption, frequency-independent active power consumption and static power consumption, and consider the loss caused by the voltage stabilization circuit (i.e., voltage stabilization loss), and finally establish a task energy consumption model with an accuracy rate of up to 95%.

[0028] 2. To address the problem of terminal system task blocking in complex and changing environments, a near-zero power adaptive task scheduler is designed, which enables tasks to adaptively update their priorities and scheduling strategies according to changes in environmental energy, decoupling the terminal's energy management from high-power MCUs and ADCs, thereby fundamentally solving the high overhead problem. BRIEF DESCRIPTION OF THE DRAWINGS

[0029] Figure 1 It is a structural schematic diagram of the present invention;

[0030] Figure 2 It is a schematic diagram of the task load power model;

[0031] Figure 3 It is a schematic diagram of the energy state detection circuit;

[0032] Figure 4 It is the logic symbol diagram of 74148;

[0033] Figure 5 This is a schematic diagram of the task receiving circuit;

[0034] Figure 6 A schematic diagram for dynamically setting the task energy cost interval;

[0035] Figure 7 This is a schematic diagram of the workflow of the near-zero power adaptive task scheduler;

[0036] Figure 8 Schematic diagram of the task matching circuit;

[0037] Fig. 9 is a schematic diagram of an energy supply control circuit;

[0038] Fig.10 is a schematic diagram of control signal timing situation 1;

[0039] Fig.11 Schematic diagram of control signal timing situation 2. DETAILED DESCRIPTION

[0040] In order to make the purpose, technical scheme and advantages of the embodiments of the present application clearer, the technical scheme in the embodiments of the present application will be clearly and completely described below in conjunction with the drawings in the embodiments of the present application. Obviously, the described embodiments are only a part of the embodiments of the present application, rather than all the embodiments. The components of the embodiments of the present application described and shown in the drawings here can be arranged and designed in various different configurations. Therefore, the detailed description of the embodiments of the present application provided below in conjunction with the drawings is not intended to limit the scope of protection of the application claimed for protection, but merely represents the selected embodiments of the present application. Based on the embodiments of the present application, all other embodiments obtained by those skilled in the art without making creative work belong to the scope of protection of the present application. The present invention is further described below in conjunction with the drawings.

[0041] The present invention specifically relates to a micro-energy driven task adaptive matching and scheduling control system, which realizes task matching with near-zero power consumption. In order to improve the throughput of the passive system, a system energy model and a task energy consumption model are proposed, and the power consumption data of all tasks are stored in a register group, and multiple tasks are scheduled in combination with the system energy status. Specifically, the task set executed by the terminal is fixed, and the task energy consumption is calculated by a refined task energy consumption model, and combined with the system energy model, it is converted into a binary bit string of the energy required for each capacitor in the energy storage capacitor group, and finally stored in the register group. The current system energy is judged by a numerical comparator and an AND gate whether it can execute the task, and whether to enable the LDO to supply energy to the passive system equipment.

[0042] like Figure 1As shown, the system is connected to the gateway communication and to the task load, and includes an energy state detection circuit, a task scheduler, a task matching circuit and an energy supply control circuit; the energy state detection circuit is connected to the energy storage capacitor group, detects the capacitor voltage and converts the detection result into a binary value, obtains the energy demand of the task team energy storage capacitor group by pre-configuring the load task demand, maps the demand to the voltage constraint of each capacitor, and saves the voltage constraints of all tasks to the calculation-free task receiving circuit to form a task group; the task scheduler schedules the tasks in the task group in a certain order and matches them through the task matching circuit, when the task is successfully matched, the energy supply control circuit is enabled, and the energy supply control circuit is responsible for limiting the input voltage of the LDO to prevent the LDO from dissipating too much power due to a large voltage difference, and when the input voltage is lower than the threshold, the LDO is put into a sleep state to reduce the energy consumption of the circuit, and when the task is successfully executed, the feedback energy supply control circuit stops supplying energy and notifies the task scheduler that the task is successfully executed.

[0043] Task energy consumption model:

[0044] In the actual application of the Internet of Things, the terminal usually carries the environmental perception task, and the gateway serves as the data interaction center and implements the control of the terminal. Therefore, when the system starts, the gateway needs to send all task information to the system, including task energy consumption information. Therefore, it is necessary to study the energy consumption of the task load. To this end, the present invention establishes a refined task energy consumption model, and calculates the theoretical value of the energy required for various tasks by decomposing the components of the task load.

[0045] In a task load circuit, in addition to the energy overhead caused by the necessary computing units (MCU, ADC and sensors, etc.), there is also additional overhead caused by voltage regulation. To this end, the present invention studies the specific power consumption composition of different components, considers the voltage regulation loss, and finally establishes a refined task energy consumption model.

[0046] Peripheral power consumption: The core task of the passive system is environmental perception, which mainly obtains environmental data through peripherals (sensors). Therefore, sensors are one of the main energy consumption sources of the system. Different sensors are added to achieve functions according to different perception tasks. The power consumption data of different sensors vary greatly, and specific analysis of the sensors required for different task loads is required.

[0047] Microprocessor power consumption: Perception tasks or communication tasks are usually accompanied by calculations, which require the microprocessor to provide computing resources. The power consumption of the microprocessor is mainly composed of dynamic power consumption and static power consumption. Among them, dynamic power consumption comes from circuit activities, such as level flipping, while static power consumption is mainly in the form of leakage current. Dynamic power consumption P dIt can be calculated by the following formula, where α is the activity factor, which is related to the number of level flips, C is the effective switch capacitance, V is the supply voltage, and f is the clock frequency.

[0048] P d =αCV 2 f;

[0049] The static power consumption of the microprocessor is P s Mainly leakage power, its calculation is given by the following formula, where I leak is the leakage current. Static power consumption is usually given in the microprocessor data sheet.

[0050] P s =V·I leak ;

[0051] In order to make the passive system run longer, the task load usually chooses to use a low-power microprocessor. The power consumption of this type of microprocessor is generally in the milliwatt (mW) level, which is suitable for some energy-scarce scenarios.

[0052] Task load power model: The task load of a passive system usually includes components such as microprocessors, peripherals, memory, and clocks, and the overall power consumption of the task load includes not only the active power consumption related to frequency, but also the active power consumption independent of frequency. Therefore, the present invention uses a system-level power model to describe the task load power model. The task load power model divides the power consumption of the load into three types: active power consumption P related to frequency, ... dep , frequency-independent active power consumption P ind And static power consumption P s Then, the power consumption of the task load P task It can be described by the following formula:

[0053] P task =P s +h(P ind +P dep );

[0054] The coefficient h indicates whether the task load currently consumes active power. Specifically, if the task load is in an active state, h=1; otherwise (ie, the task load is in a sleep mode or is turned off), h=0.

[0055] like Figure 2 As shown in Figure 1, it shows the change of power consumption of a task load over time. At the initial moment, the task load is in the off state and does not consume energy. w1 Starting from time, the task load is at frequency f 1 Operation, the overall power consumption is determined by P s , P ind and P dep Composition. To T sAt this moment, due to insufficient system energy, the task load enters a dormant state, and the overall power consumption is only P s As the system stores more and more energy, by the time T w2 At this moment, the task load is at a frequency f 2 Run, and f 2 Less than f 1 , then the task load P dep reduce.

[0056] Voltage stabilization loss: Due to the instability of environmental energy, the passive system cannot meet the task load's demand for smooth voltage, so a voltage stabilization circuit is often used to smooth the voltage. However, the voltage stabilization circuit has energy loss problems and consumes the energy stored in the system. This invention uses a general voltage stabilization loss model to describe the loss of the voltage stabilization circuit. Assume that the passive system uses a relatively efficient switching circuit for voltage stabilization, and its switching power is P K , the input impedance and output impedance are R W1 and R W2 , and the output impedance of the capacitor is R C , then, the power loss P caused by the voltage stabilization circuit at time t l (t) can be expressed by the following formula:

[0057] P l (t) = I C (t) 2 ·(R W1 +R C )+I l (t) 2 ·R W2 +P K ;

[0058] Among them, I C (t) is the instantaneous current output by the capacitor, I l (t) is the instantaneous current of the task load overhead.

[0059] Task total cost model: In addition to the task's own power consumption, the task's energy cost also includes the energy loss caused by voltage regulation. Assume that the task execution time is T and the task's own power consumption is P task (t), the voltage regulation loss is P l (t), the total task cost can be expressed as:

[0060]

[0061] In the actual application of the Internet of Things, the gateway acts as a data exchange center and is responsible for managing the task execution of the passive system. Before starting the task on the passive system, the gateway calculates the energy requirements of all tasks to be executed on the terminal system according to the total task cost model, and encodes the calculation results combined with other task information into gateway instructions, and finally sends the instruction set to the passive system in a remote form to achieve energy distribution.

[0062] Furthermore, if Figure 3 As shown, the normal operation of the digital circuit depends on a reliable and stable power supply, which must be higher than its minimum operating threshold. The passive system can only complete the task planning by detecting the energy of the energy storage capacitor group. The basic idea of ​​the energy state detection circuit is to use multiple comparators with different reference voltages to detect the voltage state of each capacitor with near zero power consumption, thereby obtaining the energy state of the system.

[0063] Specifically, the voltages of the N capacitors in the capacitor bank are detected by m comparators, and the reference voltage is set to V 1 ~V m , and satisfy V 1 <V 2 <... <V m Therefore, when the capacitor meets a certain reference voltage V ref When V ref For the ith capacitor, after the voltage is detected by m comparators, the detection result will be (1, ..., 1, 0, ..., 0). i Bit 1 indicates the first m i The comparator outputs a high level, which is the voltage of capacitor i. Meet the reference voltage Considering that the power consumption information of the task is sent to the system by the gateway through wireless communication and then stored in the register by the system, in order to reduce the communication overhead and save system resources, the present invention uses a priority encoder to encode the detection result to reduce the number of bits.

[0064] The computation-free energy allocation circuit uses a priority encoder to encode information with specific meaning into a number of binary bits. This process is called encoding, and the digital circuit that implements this process is the encoder. Specifically, the function of the encoder is to convert n binary bits into log 2 n binary bits. Table 1 shows the truth table of a three-bit binary encoder that can encode eight inputs into three outputs. 0 , A 1 and A 2 There are three output terminals, and their logical expressions can be expressed as:

[0065]

[0066] Table 1. Truth table of 8-3 line encoder

[0067]

[0068] It can be found that at a certain moment, only one of the eight input terminals of the encoder can be at a low level, and the rest are at a high level. This characteristic does not conform to the capacitor voltage detection result designed in this patent, so the present invention uses a priority encoder to encode the detection result. The priority encoder solves the problem that there can only be one valid signal at the input terminal. Table 2 shows the truth table of the 8-3 line priority encoder, and Figure 4 The logic symbol of the commonly used 8-3 line priority encoder 74148 is shown. Among them, It is the control end, which controls whether the encoder works; It is the expansion terminal. If all inputs are 1, Output 1; while at I 0 ~I 7 middle, I 7 has the highest priority.

[0069] Table 2. Truth table of 8-3 line priority encoder

[0070]

[0071]

[0072] For the voltage detection result (1, ..., 1, 0, ..., 0) of the i-th capacitor, the binary value bin (m i ), which represents the decimal m i The corresponding binary value. i Indicates the number of bits 1. Generally, for the detection results of m comparators, there are a total of m+1 cases, which need to be encoded into log 2 m+1 bits, that is, (m+1)-log 2 (m+1)-line priority encoder.

[0073] The voltage of all capacitors in the capacitor group is detected and encoded, and the encoded voltage states of N capacitors are obtained as shown in Table 3, that is, N*log 2 (m+1) bits. These binary bit stream information constitutes the energy state of the passive system, which is used for subsequent reasonable task planning. At this point, the near-zero power consumption detection of the system energy state is completed.

[0074] Table 3 Passive system energy state coding

[0075]

[0076] Free computing task queue storage: Figure 5 As shown, the shift register is used to receive gateway instructions (i.e., task information), and its storage space includes four parts: task energy cost, priority, scheduling strategy, and task command. Among them, the task energy cost occupies the initial N*log 2 The (m+1)bit space is used to specify the energy cost of the task. The priority occupies the pbit space, which is used to specify the scheduling priority of the task. The scheduling strategy occupies the sbit space, which is used to specify the scheduling strategy of the task. The task command occupies the last tbit space, which is used to schedule system tasks. The shift register adopts a serial input and parallel output structure, which outputs the gateway instructions in parallel to the register group management unit, and then the unit writes the task information into the register group. The following introduces each functional component:

[0077] (1) Register group: The register group consists of multiple independent registers, which can split and store the control flow sent by the gateway according to the parts of the task information. Each square in the figure is a register, and squares of the same color represent the information of the same task. For example, the four parts of task 1 are stored in the first four registers of the register group, followed by the information of task 2, and so on.

[0078] (2) Register group management: The function of this component is to manage the register group. Specifically, it controls the writing and erasing of data in the target register in the register group. When the gateway sends a new instruction to the system, the register group management unit writes the received control flow into four blank registers according to the task information. Similarly, when a task no longer needs to be executed by the system, the register group management unit will also erase the four registers occupied by the task. In addition, this unit is also responsible for executing commands from the task scheduler, such as modifying the priority and scheduling strategy of the target task.

[0079] The specific meanings of the four parts of the gateway instruction are as follows:

[0080] (1) Task energy cost: A task energy consumption model is established on the gateway, and the task energy cost is sent to the system remotely, ultimately realizing computation-free task scheduling for the passive system. The present invention proposes a refined task energy consumption model that covers the total power consumption of all components in the task load circuit, including the voltage regulation loss caused by the voltage regulation circuit, and can effectively calculate the theoretical cost of the task.

[0081] (2) Priority: Task priority specifies the priority of a task when it is scheduled. The higher the priority value set by the present invention, the higher the priority. Specifically, the lowest priority is 1 and the highest is P. max .

[0082] (3) Scheduling strategy: The task scheduling strategy specifies the execution mode of the task, whether it is one-time execution or cyclic execution. Specifically, when the value of the s bit is 1, it indicates that the task is in one-time execution mode; when the value of the s bit is n, it indicates that the task is cyclically executed n times; in particular, when the value of the s bit is 0, it indicates that the task is in infinite loop execution mode.

[0083] (4) Task command: This tbit is used to schedule tasks. It specifies the target task corresponding to the gateway command. At this point, the passive system completes the reception and parsing of gateway commands without the participation of a microprocessor, realizes the calculation-free task group update, minimizes the energy consumption of the system, and provides support for subsequent task scheduling.

[0084] Near-zero power adaptive task scheduler: A task group consists of multiple tasks, each of which has its own independent energy consumption, priority and scheduling strategy. The role of the task scheduler is to select the task suitable for the next execution from the task group and ultimately provide energy for it.

[0085] The present invention proposes a near-zero power adaptive multi-task dynamic scheduling mechanism based on task priority, which dynamically sets the priorities of multiple tasks according to the task energy overhead and the system energy state, matches the appropriate task scheduling order for the current system energy state, and ultimately improves the system's task throughput.

[0086] The initial priority of the task is set by the gateway and sent to the system remotely via wireless communication. Assume that the task group contains the task set ψ = { 1 , χ 2 ,…,χ n}, the energy cost of each task is E 1 ~E n , the initial priorities are P 1 ~P n Consider two tasks χ i and χ j , the task priorities are P i and P j (P i >P j ), the energy costs of the tasks are E i and E j (E i >E j ). Due to the task x i The priority of task x is greater than J , χ i It needs to be scheduled first, but its energy cost is greater, and the system may not be able to store enough energy to execute x i , which leads to the task χ JTherefore, the patent dynamically sets the task priority in combination with the task energy cost, so that the system will not be blocked by tasks.

[0087] Furthermore, the present invention divides the task energy cost into K intervals, which are mutually disjoint and continuous. The i-th interval is represented as <lb i ,ub i >,lb i Indicates the lower boundary of the interval, ub i represents the upper boundary of the interval. Each interval has a different impact on the task priority. Assume that the impact of the i-th interval on the task priority is m i (e.g. +1, -1, +2, etc.), task χ t The energy cost of belongs to the i-th interval, then χ t The new priority of can be calculated by the following formula. t Represents the task x t The old priority, P t ′ represents the task χ t The new priority, P max Indicates the highest priority of the task, and the lowest priority of the task is 1.

[0088] P′ t =Min{Max{P t +m i , 1}, P max};

[0089] Obviously, the priority of tasks with high energy consumption should be lowered, because it is more difficult for the system to meet their energy requirements, so m i Less than 0. Similarly, the priority of tasks with low energy consumption should be increased because the system can more easily meet their energy requirements, so m i Greater than 0. In order to cope with complex environmental changes, the boundary values ​​of the K energy cost intervals should be set dynamically. For example, when the environmental energy condition is poor, the length of each interval should be shortened; when the environmental energy condition is good, the length of each interval should be extended. Figure 6 The schematic diagram of dynamically setting the interval boundary value is shown. When the environmental energy condition is good (i.e., the input power of the micro-energy source is large), the task χ t The energy cost of task χ may be in the low cost range, while when the environmental energy condition is poor (i.e., the input power of the micro energy source is small), t The energy cost of a task may be in the high-cost range. Therefore, as the environmental energy condition deteriorates, the low-cost task will be transformed into a high-cost task, and its task priority will also be reduced.

[0090] The present invention utilizes the task matching success rate η task To characterize the environmental energy status. task The calculation of is given by the following formula, where Count success Indicates the number of successful task matching, Count failure Indicates the number of task matching failures.

[0091]

[0092] The task scheduler selects the task with the highest current priority and then transmits the energy cost bit stream of the task to the task matching circuit. If the current energy state of the system meets the energy requirements of the task, the task matching is successful, and Count success Add 1 and update η task If not, the task matching fails and the system needs to continue to store energy. When the task matching fails, in order to avoid task blocking in the system, if the system waits for energy storage time to exceed the upper threshold T wait , then Count failure Add 1 and update η task ; Otherwise, the system has stored enough energy within the threshold time, Count success Add 1 and update η task When η task When it increases, it means that the environmental energy condition is getting better; otherwise, it means that the environmental energy condition is getting worse. task The relationship between the length l of each interval can be described by the following formula. The function f is a monotonically increasing function.

[0093] l=f(η task );

[0094] Consider the task scheduling strategy, which stipulates two modes of task execution: one-time execution and cyclic execution. For a one-time execution task, it will be erased from the task group after the task is executed and will not be scheduled again. For a cyclically executed task, if its task priority is high, the task will be continuously scheduled for execution, thereby blocking other tasks. Therefore, when the task scheduler receives the feedback signal that the execution of this type of task has ended, it needs to update its task priority. The specific update operation is given by the following formula. Among them, the function g is a monotonically decreasing function, so the priority of the cyclically executed task will decrease as the number of executions increases.

[0095] P′ t =g(P t );

[0096] The entire workflow of the near-zero power adaptive task scheduler is composed of Figure 7So far, the task scheduler has realized a near-zero power adaptive multi-task dynamic scheduling mechanism that dynamically sets task priorities according to the environmental energy status and task energy consumption, solving the problem that the traditional scheduling mechanism cannot adapt to diverse environments, and minimizing the energy consumption of the task scheduling process, effectively improving the energy utilization efficiency of the passive system.

[0097] Priority scheduling is explained by example: Assume there is a task set<T1,T2,T3> , the initial priority is <3,6,3>, the energy cost is <2,10,1>, the initial length of the task energy consumption interval is 4, a total of 3 intervals, that is, the interval is divided into (<0,4>,<5,9>,<10,14>), and the impact of the 3 intervals on the priority is (+1,0,-1). Schedule task T2 first because it has the highest priority. If the system energy satisfies T2 execution, update Count success is 1, update η task is 1. Assuming that the interval length l is 5 according to the mapping of function f, the energy consumption intervals of the three tasks are updated to (<0,5>,<6,11>,<12,17>). Task T2 is a cyclic task, so assume that its priority is reduced by 1 to 5. Then update all task priorities. The energy consumption of T1 falls into the first interval, and its priority is updated to 3+1=4; the energy consumption of T2 falls into the second interval, and its priority is updated to 5+0=5; the energy consumption of T3 falls into the first interval, and its priority is updated to 3+1=4. Start the next task scheduling, still scheduling T2, but the current system energy does not meet the execution of T2 and the waiting timeout. Update Count failure is 1, update η task is 0.5. Assuming that the interval length l is 3 according to the mapping function f, the energy consumption intervals of the three tasks are updated to (<0,3>,<4,7>,<8,11>). The priority of T2 is reduced by 1 to 4. Then all task priorities are updated. The energy consumption of T1 falls into the first interval, and its priority is updated to 4+1=5; the energy consumption of T2 falls into the third interval, and its priority is updated to 4-1=3; the energy consumption of T3 falls into the first interval, and its priority is updated to 4+1=5. Start the next task scheduling, and schedule T1. And so on.

[0098] Near-zero power task matching circuit: The task scheduler selects the task with the highest current priority from the task group, but the current system energy state may not be able to meet the energy requirements of the task, so it is necessary to judge the matching between the system energy and the task requirements.

[0099] like Figure 8As shown in the figure, from the overall logic point of view, the task energy cost includes N parts (N is the number of energy storage capacitors), where the i-th part describes the coded voltage requirement for the i-th energy storage capacitor. The coded voltage requirements of these N parts are compared one by one with the coded voltage state of the corresponding energy storage capacitor through a numerical comparator to obtain N comparison judgment results 1 or 0, and finally all the comparison judgment results are input into the AND gate to obtain the task matching judgment result 1 or 0. When the matching judgment output is high level 1, it means that the current system energy state meets the energy requirement of the target task, otherwise, it means that it does not meet the energy requirement.

[0100] Considering that the condition that the system energy state meets the task energy requirement is that the energy of the energy storage capacitor group is greater than or equal to the task energy consumption. Therefore, the present invention converts the two outputs F A>B and F A=B Input an OR gate, and use the output of the OR gate as the comparison result. Among them, A represents the binary coded voltage state of the energy storage capacitor, and B represents the binary coded voltage requirement of the corresponding component of the task energy cost. Based on the above design, the need to use microprocessors and ADCs to calculate the energy state of passive systems in the traditional way is avoided, and calculation-free near-zero power task matching is achieved, minimizing system costs.

[0101] Energy supply control circuit: Its ultimate goal is to provide stable energy supply for the mission load. When the system energy state meets the mission energy requirement, the passive system powers the mission load.

[0102] like Fig. 9 As shown, when the matching judgment result output by the calculation-free task matching circuit is high, it indicates that the task scheduled by the task scheduler can be successfully executed by the system, and then the switch S is turned on, thereby providing stable energy supply for the task through the LDO. The advantage of doing this is that it can avoid the task being interrupted due to insufficient energy in the energy storage capacitor group. However, as the system supplies energy to the task, the energy of the energy storage capacitor group gradually decreases, which may cause the matching judgment to become a low level, thereby interrupting the energy supply.

[0103] To solve this problem, the present invention uses an OR gate and two AND gates to form a simple control circuit. In the default state, the control signal T 1 , T 2 , T 4 , T 5 and T 6 are both in low level state, switch S is disconnected, and T 3 Since the pull-up resistor R l By default, it is in high level state. 1 When it is high level, T 5 Set high, switch S is turned on, and the energy storage capacitor bank starts to supply energy for the task through LDO. 2As the system supplies energy for the task, the system energy state will change in two ways: (1) decreasing but still meeting the task match; (2) decreasing to not meeting the task match. For case 1, T 5 Always set high, then T 6 When the task is completed, T 3 Setting it low makes T 6 is also set low, thus terminating the energy supply. For case 2, since T 4 Always set high, so T 5 Also set high, then T 6 Finally, when the task is successfully completed, by setting T 3 Set low so that T 6 Also set low, so that the switch S is disconnected, stop supplying energy to the task load, and reduce the energy consumption of the circuit. In particular, when the task is completed, T 3 The low signal is input to the task scheduler to notify it to modify the task group. For the above two cases, the control signal timing of the energy supply control circuit is as follows Figure 10-11 As shown in the figure, adaptive energy supply is realized when the task matching is successful, and adaptive energy supply is stopped when the task execution is completed, thereby reducing the system overhead to the greatest extent.

[0104] The above is only a preferred embodiment of the present invention. It should be understood that the present invention is not limited to the form disclosed herein, and should not be regarded as excluding other embodiments, but can be used for various other combinations, modifications and improvements, and can be modified within the scope of the concept described herein through the above teachings or the technology or knowledge of the relevant field. The changes and modifications made by those skilled in the art do not deviate from the spirit and scope of the present invention, and should be within the scope of protection of the claims attached to the present invention.

Claims

1. A micro-energy drive task adaptive matching and scheduling control system, characterized by: The system is connected to the gateway in communication and to the task load, and includes an energy state detection circuit, a task scheduler, a task matching circuit and an energy supply control circuit; The energy state detection circuit is connected to the energy storage capacitor group, detects the capacitor voltage and converts the detection result into a binary value, obtains the energy demand of the task team's energy storage capacitor group by pre-configuring the load task demand, maps the demand to the voltage constraint of each capacitor, and saves the voltage constraints of all tasks to the calculation-free task receiving circuit to form a task group; the task scheduler schedules the tasks in the task group in a certain order and matches them through the task matching circuit, when the task is matched successfully, the energy supply control circuit is enabled, and when the task is executed successfully, the feedback energy supply control circuit stops supplying energy and notifies the task scheduler that the task is executed successfully.

2. A micro-energy drive task adaptive matching and scheduling control system according to claim 1, characterized in that: The energy state detection circuit includes a plurality of comparators with different reference voltages to detect the voltage state of each capacitor in the energy storage capacitor group. The output ends of the plurality of comparators with different reference voltages are connected to a priority encoder. The detection results are encoded into a code of several binary bits by the priority encoder to obtain the encoded voltage states of all capacitors.

3. The micro-energy drive task adaptive matching and scheduling control system according to claim 1 is characterized by: The computation-free task receiving circuit includes a shift register, a register group management unit and a register group; the shift register is used to receive gateway task information, the register group is composed of multiple independent registers, and is used to split and store the control stream sent by the gateway according to the various parts of the task information, and the register group management unit is used to write and erase data of the target register in the controller register group.

4. A micro-energy drive task adaptive matching and scheduling control system according to claim 3, characterized in that: The storage space of the shift register includes four parts: task energy cost, priority, scheduling strategy and task command; The task energy cost: by establishing a refined task energy consumption model on the gateway to cover the total power consumption of all components in the task load, the theoretical cost of the task is calculated and sent to the system in a remote manner; The priority: specifies the priority of a task when it is scheduled. The higher the priority value, the higher the priority. The scheduling strategy: specifies whether the task is executed once or cyclically. When the occupied space value of the scheduling strategy is 1, it indicates that the task is in a one-time execution mode. When the occupied space value of the scheduling strategy is n, it indicates that the task is cyclically executed n times. When the occupied space value of the scheduling strategy is 0, it indicates that the task is in an infinite loop execution mode. The task command specifies the target task corresponding to the gateway task information.

5. A micro-energy drive task adaptive matching and scheduling control system according to claim 4, characterized in that: The task energy consumption model includes a task load power model and a voltage stabilization loss model; The task load power model includes the frequency-dependent active power consumption P dep , frequency-independent active power consumption P ind And static power consumption P s , that is, the power consumption of the task load is P task =P s +h(P ind +P dep ), the coefficient h indicates whether the task load currently consumes active power; The voltage stabilization loss model describes the loss of the voltage stabilization circuit. Assume that the switching power P of the voltage stabilization circuit is K , the input impedance and output impedance are R W1 and R W2 , and the output impedance of the capacitor is R C , the power loss caused by the voltage stabilization circuit at time t is P l (t) = I C (t) 2 ·(R W1 +R C )+I l (t) 2 ·R W2 +P K , I C (t) is the instantaneous current output by the capacitor, I l (t) is the instantaneous current of the task load overhead.

6. A micro-energy drive task adaptive matching and scheduling control system according to claim 4, characterized in that: The task scheduler dynamically sets the priorities of multiple tasks according to the task energy cost and the system energy state, and matches the appropriate task scheduling order for the current system energy state; By dividing the task energy cost into K disjoint and continuous intervals, the i-th interval is expressed as <lb i ,ub i >,lb i Indicates the lower boundary of the interval, ub i represents the upper boundary of the interval. Each interval has a different impact on the task priority. Let the impact of the i-th interval on the task priority be m i , task x t The energy cost of belongs to the i-th interval, then χ t The new priority can be expressed by the formula P t ′=Min{Max{P t +m i ,t},P max } is calculated, where P t Represents the task x t The old priority, P t ′ represents the task χ t The new priority, P max Indicates the highest priority of the task, and the lowest priority of the task is 1.

7. The micro-energy drive task adaptive matching and scheduling control system according to claim 6, characterized in that: The success rate of task matching η task To characterize the environmental energy status, the calculation formula is: Among them, Count success Indicates the number of successful task matching, Count failure Indicates the number of task matching failures; If the current energy state of the system meets the energy requirements of the task, the task is successfully matched, and Count success Add 1 and update η task If it is not satisfied, the task matching fails and the system continues to store energy. When the task matching fails, in order to avoid task blocking in the system, if the system waits for energy storage time to exceed the upper threshold T wait , then Count failure Add 1 and update η task , otherwise, the system has stored enough energy within the threshold time, Count success Add 1 and update η task , when η task When it increases, it means that the environmental energy conditions are getting better, otherwise, it means that the environmental energy conditions are getting worse.

8. The micro-energy drive task adaptive matching and scheduling control system according to claim 4, characterized in that: The task matching circuit includes numerical comparators with the same number of capacitors as the energy storage capacitor group, the input of each numerical comparator is connected to a capacitor in the energy storage capacitor group, the outputs of all numerical comparators are connected to the first AND gate, and the task energy cost includes N parts, where N is the number of capacitors; The coding voltage requirements of the N parts are compared one by one with the coding voltage states of the capacitors through a numerical comparator to obtain N comparison judgment results. Finally, all the comparison judgment results are input into the first AND gate to obtain the task matching judgment result. When the matching judgment output is a high level, it indicates that the current system energy state meets the energy requirements of the target task, otherwise, it indicates that it does not meet the requirements.

9. A micro-energy drive task adaptive matching and scheduling control system according to claim 8, characterized in that: The condition that the current system energy state meets the energy requirement of the target task is: the energy of the energy storage capacitor bank is greater than or equal to the task energy cost; at this time, the two outputs F in each numerical comparator are A>B and F A=B The first OR gate is input, and the output of the first OR gate is used as a comparison result, wherein A represents the binary coded voltage state of the energy storage capacitor, and B represents the binary coded voltage requirement of the component corresponding to the task energy consumption.

10. The micro-energy drive task adaptive matching and scheduling control system according to claim 1, characterized in that: The energy supply control circuit includes a single-pole double-throw S, the input of the single-pole double-throw S is connected to the energy storage capacitor group, one connector of the single-pole double-throw S is connected to an LDO, and the LDO provides stable energy supply for the task, and the other connector is connected to the output of the second AND gate, one of the input ends of the second AND gate is connected to the output end of the second OR gate, and the other input end is connected to the task scheduler; one of the input ends of the second OR gate is connected to the task matching circuit, and the other input end is connected to the output end of the third AND gate; one of the input ends of the third AND gate is connected to the task scheduler, and the other input end is connected to the LDO.

Citation Information

Cited By

  • Micro-power-consumption redundancy-eliminating task hierarchical general execution circuit

    CN119110002A

  • A micro-power redundant task layered universal execution circuit

    CN119110002B