Search

CN-121984796-A - Remote node dormancy state machine control method and system based on message time sequence analysis

CN121984796ACN 121984796 ACN121984796 ACN 121984796ACN-121984796-A

Abstract

The application discloses a remote node dormancy state machine control method and a remote node dormancy state machine control system based on message time sequence analysis, which relate to the field of node power consumption management control and comprise the steps of obtaining an arrival time stamp and a message identifier of an inbound message, and storing the inbound message according to service class classification; the method comprises the steps of extracting statistical characteristics based on adjacent arrival time interval sequences of messages of all classes, constructing a message arrival probability distribution model and calculating an arrival risk function, carrying out weighted fusion by combining service priorities to obtain a system-level comprehensive risk function and predicting the arrival expected time of the next message, calculating safety margin by combining the current energy state of a node, and selecting a target dormancy state in a multi-stage dormancy state machine. Therefore, the dynamic control of the sleep depth and the wake-up time is realized, and the adaptability and response coordination of the low-power consumption management of the remote node are improved.

Inventors

  • QU BAOCHUN

Assignees

  • 苏州爱雄斯通信技术有限公司

Dates

Publication Date
20260505
Application Date
20260330

Claims (9)

  1. 1. The method for controlling the sleep state machine of the remote node based on the message time sequence analysis is applied to the remote node and is characterized by comprising the following steps: Acquiring arrival time stamps and message identifiers of a plurality of inbound messages received by the remote node on a communication link, classifying the service types of the inbound messages based on the message identifiers, and storing the service types in corresponding service type buffer queues in a classified manner; periodically reading the arrival time stamp in each business category buffer queue, calculating the adjacent arrival time interval sequence of each business category message, and extracting the statistical characteristics of the message interval of each business category based on the adjacent arrival time interval sequence; constructing a message arrival probability distribution model according to the message interval statistical characteristics of each business category, and calculating an arrival risk function corresponding to each business category based on the message arrival probability distribution model, wherein the arrival risk function is used for representing the instantaneous condition intensity of the message of the corresponding business category arriving at a given moment under the condition that the message of the corresponding business category does not arrive at the current moment; Carrying out weighted fusion processing on the arrival risk functions corresponding to each business category according to preset business priority to obtain a system-level comprehensive risk function, and predicting the expected time of arrival of the next message based on the system-level comprehensive risk function; And calculating a safety margin for waking up in advance by combining the current energy state of the remote node, selecting a target sleep state from a preset multi-stage sleep state machine according to the expected time, the safety margin and the system-level comprehensive risk function, controlling the remote node to switch between sleep states of different levels, and operating according to a power consumption level corresponding to the target sleep state before the actual waking up time determined by the expected time and the safety margin.
  2. 2. The method of claim 1, wherein said calculating a sequence of adjacent arrival time intervals for each traffic class message and extracting the traffic class interval statistics based on the sequence of adjacent arrival time intervals comprises: Establishing a time sequence sliding window with preset length for each service class, calculating time difference values between arrival time stamps of adjacent messages in the window in the process of updating the time sequence sliding window, and taking a plurality of time difference values as time interval samples to form the adjacent arrival time interval sequence of the service class; calculating the median of each of the time interval samples in the adjacent arrival time interval sequence, and determining the median as the center arrival interval of the traffic class; Calculating absolute deviations of each time interval sample from the center arrival interval, and extracting the median of all the absolute deviations; performing equivalent proportion conversion on the medians of all the absolute deviations by using a preset scale factor to obtain a dispersion statistic of the business class; And determining the center arrival interval and the dispersion statistic as the message interval statistical characteristic of the service class.
  3. 3. The method according to claim 2, wherein the constructing a message arrival probability distribution model according to the statistical characteristics of the message intervals of each service class, and calculating an arrival risk function corresponding to each service class based on the message arrival probability distribution model, includes: Fitting probability density functions of all business categories by adopting a kernel density estimation method in combination with the center arrival interval and the dispersion statistic of all business categories, and determining corresponding cumulative distribution functions based on the probability density functions; Calculating an arrival risk function corresponding to each business category based on the ratio relation between the probability density function and the unreachable probability of the business category, wherein the unreachable probability is obtained by subtracting a corresponding cumulative distribution function from a value 1; The step of carrying out weighted fusion processing on the arrival risk functions corresponding to the business categories according to the preset business priority to obtain a system-level comprehensive risk function, comprising the following steps: And determining corresponding weight factors according to the preset service priority and the historical misjudgment loss, and carrying out weighted summation processing on the weight factors corresponding to the service categories and the corresponding arrival risk functions to generate a system-level comprehensive risk function, wherein the sum of the weight factors corresponding to the service categories is normalized and constrained to be 1.
  4. 4. The method of claim 3, wherein predicting the expected time of arrival of the next message based on the system-level integrated risk function includes determining a base predicted time based on the center arrival interval corresponding to each traffic class, and performing checksum correction on the base predicted time using the system-level integrated risk function, and specifically includes: For each service category, acquiring the latest message receiving time of the service category, and adding the latest message receiving time and the corresponding center arrival interval to obtain candidate arrival time of each service category; Selecting the candidate arrival time with earliest corresponding time from the candidate arrival times corresponding to the business categories as a basic prediction time; Calculating a risk evaluation value of the system-level comprehensive risk function at the base prediction moment so as to verify the base prediction moment; when the risk evaluation value is greater than or equal to a preset risk triggering threshold value, determining the basic prediction time as the expected time; And when the risk evaluation value is smaller than the risk triggering threshold value, extending the basic prediction moment backwards along the time axis in a preset search interval, calculating the risk evaluation value corresponding to each extension moment until the risk evaluation value is determined to reach or exceed the target extension moment of the risk triggering threshold value for the first time, and determining the target extension moment as the expected moment of the arrival of the next message.
  5. 5. The method of claim 1, wherein said calculating a safety margin for early wake-up in conjunction with a current energy state of the remote node comprises: Acquiring current residual energy of the remote node as the current energy state, extracting dispersion statistics corresponding to each business category, and carrying out weighted average processing on the dispersion statistics of each business category to calculate and obtain global interval fluctuation characteristics; calculating a safety margin by combining rated energy of the remote node, the current residual energy and the global interval fluctuation characteristic; Extracting the ratio of the current residual energy to the rated energy, and extracting the ratio of the global interval fluctuation characteristic to a preset reference time constant; respectively carrying out index constraint mapping on the two ratios based on a preset non-negative adjustment coefficient, and carrying out continuous multiplication combination on the mapped two ratios and a preset reference safety margin to obtain the safety margin; And, the actual wake-up time is determined by subtracting the safety margin from the desired time.
  6. 6. The method of claim 1, wherein selecting the target sleep state in a preset multi-stage sleep state machine according to the desired time, the safety margin, and the system-level integrated risk function comprises: for each candidate sleep state in a preset multi-stage sleep state machine, respectively extracting static power consumption per unit time, hardware wake-up delay and service missing penalty cost introduced by the hardware wake-up delay corresponding to the candidate sleep state; the system-level comprehensive risk function is used for evaluating the equivalent missed risk index in each candidate sleep state according to the time difference between the current time and the actual wake-up time determined by the expected time and the safety margin and the hardware wake-up delay corresponding to each candidate sleep state; The method comprises the steps of constructing a state selection cost function and respectively calculating the comprehensive cost of each candidate dormant state, wherein the calculation of the comprehensive cost of any candidate dormant state comprises the steps of respectively calculating the hardware maintenance energy consumption cost and the service missed risk cost in the candidate dormant state, and summing the two costs, wherein the hardware maintenance energy consumption cost is the product of the static power consumption of the candidate dormant state in unit time and the time difference, and the service missed risk cost is the product of the service missed penalty cost and the equivalent missed risk index; and selecting the candidate sleep state with the minimum comprehensive cost from all the candidate sleep states as a target sleep state.
  7. 7. The method of claim 4, further comprising a multi-time scale adaptive parameter closed-loop adjustment operation based on operational feedback to dynamically update the weight factor and the risk trigger threshold, comprising: The self-adaptive correction of the weight factor comprises the steps of obtaining the actual arrival time after the remote node is awakened and the latest message of any business category is actually received, subtracting the actual arrival time from the candidate arrival time corresponding to the business category to determine a prediction error, extracting the ratio of the absolute value of the prediction error to the center arrival interval of the business category as a relative error scale, and combining the deviation direction of the prediction error, the self-adaptive learning rate and the importance control factor of the business category to calculate the corresponding weight correction; The method comprises the following steps of carrying out self-adaptive adjustment on a risk triggering threshold, namely counting the actual report missing rate in a current time window, carrying out deviation comparison on the actual report missing rate and a preset target report missing rate to obtain report missing rate deviation, adopting a proportional integral control algorithm to dynamically adjust the risk triggering threshold, wherein the operation of dynamic adjustment specifically comprises the steps of extracting report missing rate deviations of a plurality of historical time windows to calculate accumulated deviation, and respectively carrying out weighted summation on the current report missing rate deviation and the accumulated deviation by adopting a proportional adjustment coefficient and an integral adjustment coefficient to update the risk triggering threshold as dynamic adjustment quantity.
  8. 8. The method according to claim 1, characterized in that it further comprises an extreme boundary protection operation, comprising in particular: Monitoring the instantaneous channel occupancy rate of a communication link and the total message arrival frequency in real time; Triggering bypass protection logic when detecting that the instantaneous channel occupancy rate continuously exceeds a preset occupancy rate saturation threshold value in a preset detection window or the total message arrival frequency continuously exceeds a preset frequency saturation threshold value in the preset detection window, wherein the bypass protection logic is used for forcedly suspending calculation of a system-level comprehensive risk function and sleep state switching control and locking the remote node to be in a full-active state; And when the instantaneous channel occupancy rate is monitored to be continuously kept below the occupancy rate saturation threshold value in a preset recovery window and the total message arrival frequency is monitored to be continuously kept below the frequency saturation threshold value in the preset recovery window, the bypass protection logic is released, and the calculation of the system-level comprehensive risk function and the sleep state switching control are resumed.
  9. 9. A remote node sleep state machine control system based on message timing analysis, the system comprising: The message classification unit is used for acquiring arrival time stamps and message identifiers of a plurality of inbound messages received by the remote node on a communication link, classifying the service types of the inbound messages based on the message identifiers, and storing the service types in corresponding service type buffer queues in a classified manner; The interval analysis unit is used for periodically reading the arrival time stamp in each business category buffer queue, calculating the adjacent arrival time interval sequence of each business category message, and extracting the message interval statistical characteristic of each business category based on the adjacent arrival time interval sequence; The system comprises a risk calculation unit, a message arrival probability distribution model, an arrival risk function and a message processing unit, wherein the risk calculation unit is used for constructing a message arrival probability distribution model according to the message interval statistical characteristics of each business category and calculating an arrival risk function corresponding to each business category based on the message arrival probability distribution model, and the arrival risk function is used for representing the instantaneous condition intensity of a message of the corresponding business category, which arrives at a given moment under the condition that the message of the corresponding business category does not arrive at the current moment; The fusion prediction unit is used for carrying out weighted fusion processing on the arrival risk functions corresponding to the business categories according to the preset business priority to obtain a system-level comprehensive risk function, and predicting the expected time of arrival of the next message based on the system-level comprehensive risk function; And the state control unit is used for calculating a safety margin for waking up in advance according to the current energy state of the remote node, selecting a target sleep state from a preset multi-stage sleep state machine according to the expected time, the safety margin and the system-level comprehensive risk function, controlling the remote node to switch between sleep states of different levels, and operating according to a power consumption level corresponding to the target sleep state before the actual waking up time determined by the expected time and the safety margin.

Description

Remote node dormancy state machine control method and system based on message time sequence analysis Technical Field The application relates to the technical field of node power consumption management control, in particular to a remote node dormancy state machine control method and system based on message time sequence analysis. Background With the continuous development of the internet of things, the vehicle-mounted network and the industrial interconnection system, a large number of remote embedded nodes are widely deployed in complex application environments and generally depend on long-term operation of batteries or a limited power supply system. Under such an application scenario, the power consumption control capability of the node is related to the equipment endurance level, and further affects the running stability, maintenance cost and overall reliability of the system, so that low power consumption management for the remote node has become an important issue of concern in the related technical field. Aiming at the node power consumption optimization requirement, a low-power consumption control mode of a processor or a functional module level is generally adopted at present, for example, a scheme of frequency adjustment, voltage adjustment, module dormancy, periodic monitoring, periodic dormancy, wakeup scheduling and the like based on a preset rule, and the like at the network side are adopted to reduce the energy consumption of the node in an idle stage. The scheme can relieve the power consumption pressure caused by continuous standby of the node to a certain extent, and is applied to various low-power consumption communication scenes. However, most current power consumption management strategies still control based on node internal load states, residual energy, network topology information, static thresholds, or fixed scheduling periods, and lack sufficient awareness of dynamic changes in message arrival behavior in external communication links. Particularly in the schemes of low-power consumption monitoring, fixed threshold triggering awakening and coarse granularity adjustment based on average service intensity, the distribution rule of the communication message in the time dimension is difficult to be effectively reflected, and the difference of the arrival time and response requirements of the services with different priorities is less considered. In this case, when the network service presents bursty, intermittent or differential scheduling characteristics, it is often difficult to combine the energy-saving effect and the communication response performance in the prior art, on one hand, the node may not respond to the key message in time due to the relatively rigid sleep control mode, thereby causing missing report or response delay, and on the other hand, the node may generate ineffective monitoring, early awakening or excessive standby due to the relatively extensive setting of the awakening condition, thereby increasing additional energy consumption. Disclosure of Invention The application provides a remote node dormancy state machine control method, a system, a storage medium, a computer program product and electronic equipment based on message time sequence analysis, which are used for at least solving the problem that the energy-saving effect and the communication response performance are difficult to be compatible in the low-power-consumption control of a remote node in the prior art, so as to improve the power consumption management effect and the service guarantee capability of the node in a complex communication environment. In a first aspect, an embodiment of the present application provides a remote node sleep state machine control method based on message timing analysis, where the method includes obtaining arrival time stamps and message identifiers of a plurality of inbound messages received by a remote node on a communication link, classifying and storing each inbound message to a corresponding traffic class buffer queue based on the message identifiers, periodically reading arrival time stamps in each traffic class buffer queue, calculating an adjacent arrival time interval sequence of each traffic class message, extracting a statistical feature of a message interval of each traffic class based on the adjacent arrival time interval sequence, constructing a message arrival probability distribution model according to the statistical feature of the message interval of each traffic class, calculating arrival risk functions corresponding to each traffic class based on the arrival probability distribution model, where the arrival risk functions are used to characterize an instantaneous condition intensity that a message of a corresponding traffic class does not arrive at a given time under a condition of being cut to a current time, weighting the arrival risk functions corresponding to each traffic class according to a preset traffic priority, calculating an adjacent arrival risk