Dynamic load test method and system for short message communication system

Abstract

The invention discloses a dynamic load test method and system for a short message communication system, and relates to a data processing technology, and the method comprises the steps: dividing a data message into a plurality of priority categories in advance; acquiring data messages in a plurality of historical emergency scenes, and classifying the data messages according to priority categories; performing clock alignment on the classified data messages of each secondary node, and selecting a data message sample of a typical time period; extracting the message characteristics of each data message in any typical time period, endowing the corresponding message characteristics with weights according to the priorities of the data messages, and constructing a dynamic load generator by using LSTM (Long Short Term Memory); extracting message features in the current emergency scene and inputting the message features into the dynamic load generator; and performing comparison based on the output of the dynamic load generator and the load condition of the current communication system so as to complete the dynamic load test. The method is used for dynamically allocating the channel resources according to the dynamic load test result and exposing the potential resource bottleneck of the system in advance.

Description

Technical Field

[0001] This application relates to the technical field of data processing, and particularly to a dynamic load testing method and system for a short message communication system. Background Art

[0002] With the wide application of the Beidou satellite navigation system, the Beidou short message communication technology has become a key communication means in fields such as water conservancy, power, and emergency disaster relief. Especially in remote areas or extreme disaster scenarios, its importance as an emergency backup communication method has become increasingly prominent. However, the existing testing methods for Beidou short message communication systems have deficiencies in aspects such as dynamic load simulation, channel resource optimization, and test scenario authenticity, mainly manifested as follows: In a hybrid networking scenario (such as the collaboration of Beidou-3 with 5G / LoRa), low-priority messages may occupy the Beidou channel bandwidth, resulting in transmission delays for high-priority emergency data (such as disaster alarms).

[0003] The setting of current test parameters (such as load intensity, perturbation type) highly depends on manual experience and lacks the ability of data-driven dynamic adjustment. For example, the manually set load curve is difficult to cover complex scenarios not seen in history (such as the superimposed disaster of "earthquake + communication interruption"), resulting in insufficient test coverage and the inability to expose potential system bottlenecks. Summary of the Invention

[0004] Embodiments of this application provide a dynamic load testing method and system for a short message communication system to solve or partially solve the above technical problems, and to achieve dynamic allocation of channel resources according to the results of dynamic load testing and to expose potential system resource bottlenecks in advance.

[0005] Embodiments of this application propose a dynamic load testing method for a short message communication system, which is applied to an emergency communication scenario based on Beidou short messages. The emergency communication scenario includes multiple secondary nodes and a primary node, and the secondary nodes send data messages to the primary node based on Beidou short messages. The dynamic load testing method includes: Dividing data messages into multiple priority categories in advance, where the higher the priority of a data message, the more suitable it is for short message communication transmission, and low-priority messages use a hybrid channel for transmission; Obtaining data messages in multiple historical emergency scenarios and classifying them according to the priority categories; Aligning the clocks of the classified data messages of each secondary node and selecting data message samples in a typical period, where the typical period is a period in which the total amount of data messages exceeds a preset threshold; Extracting the message features of each data message in any typical period, assigning weights to the corresponding message features according to the priority of each data message, and using LSTM to construct a dynamic load generator; Extract the message features in the current emergency scenario and input them into the dynamic load generator; Compare based on the output of the dynamic load generator and the load condition of the current communication system to complete the dynamic load test.

[0006] Optionally, after classifying the data messages of each secondary node, perform clock alignment, and select data message samples in typical time periods, including: Align the data messages of each secondary node classified according to the time stamps under a unified time coordinate; Intercept according to a unified time window under the time coordinate; Use the time periods in the intercepted data segment where the total number of data messages exceeds a preset threshold, and the specified number of adjacent previous time periods as data message samples in typical time periods.

[0007] Optionally, it further includes adding perturbations to the data message samples in some typical time periods by using at least one of the following methods: Obtain multiple high-priority messages in the data message samples of each typical time period, and randomly insert high-priority messages within the typical time period, where the insertion frequency follows a Poisson distribution that satisfies a preset burst intensity; For the data message sample of any typical time period as itself, and / or, add a random time offset of ±Δt to the message times of the high-priority messages in the time windows before and after it; Add Gaussian noise to the data fields in the data messages, and randomly discard some non-keyword fields.

[0008] Optionally, it further includes adding spatio-temporal perturbations to the data message samples by using the following method: Obtain the geographical distribution data of historical disaster events in the communication coverage area; Determine the intersection area with the target monitoring range in the geographical distribution data; Extract the geographical feature information of the intersection area, and determine the similarity between the extracted geographical feature information and the local geographical features of the target monitoring range; Based on the local area corresponding to the local geographical features with a similarity higher than a preset threshold, associate and insert logically related data messages.

[0009] Optionally, when the target monitoring range includes a river channel, the extracted geographical feature information and local geographical features include: terrain features, hydrological features, and meteorological features, where: The terrain features are obtained based on DEM data; The hydrological features include river channel hydrological data, historical flood frequency, and soil permeability coefficient; The meteorological features include extreme rainfall and extreme rainfall event frequency; Adopt a random forest model to dynamically allocate the weights of features based on the relevance between historical emergency events and geographical features.

[0010] Optionally, determining the similarity between the extracted geographical feature information and the local geographical features of the target monitoring range also includes introducing a time dimension and superimposing a time decay factor when calculating the geographical feature similarity, satisfying: Among them, represents the weight of the i th type of geographical feature, represents the similarity of the i th type of geographical feature normalized by cosine similarity or Euclidean distance, represents the timestamp of the current data packet, represents the time of the historical emergency event, is the time decay coefficient, used to make the impact of recent emergency events greater.

[0011] Optionally, using LSTM to construct a dynamic load generator includes: Obtain the historical load data corresponding to any typical period and extract the load characteristics of the historical load data. The load data includes channel packet loss rate, end-to-end delay, CPU, and memory occupancy rate; Take the packet characteristics with weights assigned to any typical period and the corresponding load characteristics as a group and input them into the LSTM to use the LSTM to generate a predicted load curve to construct a dynamic load generator.

[0012] Optionally, it further includes configuring the comprehensive optimization objective function for load testing: Among them, , , are adjustable hyperparameters, represents the load coverage rate of the typical period t, the average success rate of high-priority packets in the typical period t, represents the channel overload risk of the typical period t; Solve the dynamic optimization parameter combination in the comprehensive optimization objective function to balance the test efficiency and system stability in real time.

[0013] The embodiment of the present application also proposes a dynamic load testing system for a short message communication system, including a processor and a memory. A computer program is stored on the memory, and when the computer program is executed by the processor, the steps of the dynamic load testing method for the short message communication system as described above are implemented.

[0014] The method of the embodiment of the present application can realize dynamic allocation of channel resources according to the results of dynamic load testing and expose potential resource bottlenecks in the system in advance.

[0015] The above description is only an overview of the technical solution of the present application. In order to be able to understand the technical means of the present application more clearly, it can be implemented according to the content of the specification. And in order to make the above and other purposes, features and advantages of the present application more obvious and understandable, the following specifically illustrates the embodiments of the present invention. Description of the Drawings

[0016] By reading the following detailed description of the preferred embodiments, various other advantages and benefits will become clear to those of ordinary skill in the art. The drawings are only for the purpose of showing the preferred embodiments and are not considered to be a limitation of the present invention. Moreover, throughout the drawings, the same reference numerals are used to represent the same components. In the drawings: Figure 1 It is a schematic diagram of the basic process of the dynamic load testing method for the short message communication system of this embodiment. Detailed Embodiments

[0017] The exemplary embodiments of the present disclosure will be described in more detail below with reference to the drawings. Although the exemplary embodiments of the present disclosure are shown in the drawings, it should be understood that the present disclosure can be implemented in various forms and should not be limited by the embodiments set forth herein. On the contrary, these embodiments are provided so that the present disclosure can be more thoroughly understood and the scope of the present disclosure can be fully conveyed to those skilled in the art.

[0018] The embodiment of the present application proposes a dynamic load testing method for a short message communication system, which is applied to an emergency communication scenario based on Beidou short messages. In some specific examples, the emergency communication scenario can be, for example, a cascade power station scenario, a wide-area power monitoring scenario, etc. The emergency communication scenario includes a plurality of secondary nodes and a primary node. In a specific example, the primary node can correspond to a centralized control center in the emergency communication scenario, and the secondary nodes can correspond to each monitoring site. The secondary nodes send data messages to the primary node based on Beidou short messages. As Figure 1 shown, the dynamic load testing method includes: In step S101, multiple priority categories are pre-divided for the data messages. Among them, the data messages with higher priority are adapted to short message communication transmission, and the messages with lower priority are transmitted using a hybrid channel. For example, in a high-load scenario, the Beidou-3 channel resources are preferentially allocated to the high-priority messages, and the low-priority data is transmitted using a hybrid channel (such as 5G / LoRa).

[0019] In step S102, data packets in multiple historical emergency scenarios are obtained and classified according to priority categories. That is, the data packets in the historical emergency scenarios are classified according to the multiple priority categories divided above.

[0020] In step S103, the data packets classified by each secondary node are clock-aligned, and data packet samples in a typical period are selected, where the typical period is a period in which the total amount of data packets therein exceeds a preset threshold. The dynamic load testing method of the embodiment of the present application is used to test the resource congestion situation of the channel in extreme cases. By clock-aligning the data packets in the historical emergency scenarios, the peak data packet volume that may exist at the same moment can be determined.

[0021] In step S104, the packet features of each data packet in any typical period are extracted, weights are assigned to the corresponding packet features according to the priorities of the data packets, and a dynamic load generator is constructed using LSTM. In a specific example, a dynamic load generator can be constructed in combination with parameters such as channel packet loss rate and end-to-end delay.

[0022] In step S105, the packet features in the current emergency scenario are extracted and input into the dynamic load generator.

[0023] In step S106, a comparison is made based on the output of the dynamic load generator and the load situation of the current communication system to complete the dynamic load test. After the dynamic load test is completed, channel resources can be dynamically allocated according to the test results, and the method based on the embodiment of the present application can expose potential resource bottlenecks of the system in advance based on the dynamic load generator, forming a "test - analysis - optimization" closed loop to improve the self - adaptability of the test system.

[0024] In some embodiments, clock-aligning the data packets classified by each secondary node and selecting data packet samples in a typical period includes: The data packets classified by each secondary node are aligned under a unified time coordinate according to the timestamps. In a specific example, alignment can be performed according to the sending timestamp of the data packet and the normal delay.

[0025] Intercepting is performed under the time coordinate according to a unified time window. For example, a time window with a specified duration can be set, and the classified data packets are intercepted under the aligned unified time coordinate. The intercepted data includes multiple segments of data packets in this time window.

[0026] The time periods in the intercepted data segment where the total number of data packets exceeds a preset threshold, as well as the specified number of adjacent previous time periods, are used as data packet samples for typical time periods. Based on the long short-term memory effect of LSTM, further using the specified number of adjacent previous time periods as packets for typical time periods can improve the learning and prediction effect of LSTM.

[0027] In some embodiments, it further includes adding perturbations to the data packet samples of some typical time periods by at least one of the following methods: Obtain multiple high-priority packets in the data packet samples of each typical time period, and randomly insert high-priority packets within the typical time period, where the insertion frequency follows a Poisson distribution that satisfies the preset burst intensity of the Poisson distribution.

[0028] For the data packet sample of any typical time period, add a random time offset of ±Δt to the packet time of itself, and / or the high-priority packets in the time windows before and after it; Add Gaussian noise to the data fields in the data packets, and randomly discard some non-keyword fields. Under normal circumstances, since the overall sample size of emergency data packets is small, through the above methods, on the one hand, the emergency data packet samples can be expanded, and on the other hand, the randomness and suddenness of the emergency data packet samples can be improved, so as to ensure that the system has a certain amount of redundancy according to the results of dynamic load testing.

[0029] Based on the foregoing embodiments, since the overall number of emergency data packet samples is small, and in actual use, the emergency communication scenario may cover a large area. In some embodiments, it further includes adding spatio-temporal perturbations to the data packet samples in the following way to further supplement the emergency data packet samples: Obtain the geographical distribution data of historical disaster events in the communication coverage area.

[0030] Determine the intersection area with the target monitoring range in the geographical distribution data, that is, determine the historical disaster events that occurred in the target monitoring range, and the corresponding historical disaster events may generate emergency data packets.

[0031] Extract the geographical feature information of the intersection area, and determine the similarity between the extracted geographical feature information and the local geographical features of the target monitoring range.

[0032] Based on the local area corresponding to the local geographical features with a similarity higher than the preset threshold, logically related data packets are associated and inserted, that is, according to historical disaster events, for areas with a higher similarity, similar disaster events may occur and generate data packets. This possibility is represented by associating and inserting logically related data packets. In a specific example, a certain event occurrence probability can be set, and data packets are inserted according to the probability.

[0033] In some embodiments, when the target monitoring range includes a river channel, the extracted geographical feature information and local geographical features include: topographical features, hydrological features, and meteorological features, where: The topographical features are obtained based on DEM data and include, for example, topographical slope, elevation standard deviation, surface roughness, etc.

[0034] The hydrological features include river channel hydrological data, historical flood frequency, and soil permeability coefficient.

[0035] The meteorological features include extreme rainfall amount and extreme rainfall event frequency.

[0036] The random forest model is used to dynamically assign weights to features based on the correlation between historical emergency events and geographical features. For example: in areas with a high incidence of debris flows, the weight of the topographical slope is increased to 60%; in areas with frequent floods, the weight of hydrological features accounts for more than 70%.

[0037] In some embodiments, determining the similarity between the extracted geographical feature information and the local geographical features of the target monitoring range further includes introducing a time dimension and superimposing a time decay factor when calculating the geographical feature similarity, satisfying: where, represents the weight of the i th type of geographical feature, represents the similarity of the i th type of geographical feature normalized by cosine similarity or Euclidean distance, represents the timestamp of the current data packet, represents the time of the historical emergency event, is the time decay coefficient, which is used to make the impact of recent emergency events greater.

[0038] Through such a design, the accuracy of similarity calculation can be effectively improved, avoiding coarse-grained matching that only relies on spatial overlap, and the dynamic weight assignment enables the model to adapt to different disaster types.

[0039] In some embodiments, using LSTM to construct a dynamic load generator includes: Obtaining historical load data corresponding to any typical period and extracting the load characteristics of the historical load data. The load data includes channel packet loss rate, end-to-end delay, CPU, and memory occupancy rate; Taking the message features with weights assigned to any typical period and the corresponding load features as a group and inputting them into the LSTM to generate a predicted load curve using the LSTM to construct a dynamic load generator. That is, in some examples, a certain time interval can be set for the message features and load features as a group, and then the load curve after the corresponding time interval is predicted through the LSTM.

[0040] In some embodiments, it further includes configuring a comprehensive optimization objective function for load testing: Among them, 、 、 are adjustable hyperparameters, represents the load coverage rate of the typical period t, the average success rate of high-priority messages in the typical period t, represents the channel overload risk of the typical period t; Solving the dynamic optimization parameter combination in the comprehensive optimization objective function to balance the test efficiency and system stability in real time. In this way, an integrated process of "generation - test - optimization" closed-loop is realized, reducing manual participation, and multi-objective optimization ensures that both the test coverage rate and system stability meet the standards. The method of the embodiment of the present application realizes the suddenness of the test scenario and system stability, improves the test efficiency, and provides a comprehensive solution for the reliability evaluation of the Beidou short message communication system through means such as feature extraction and adaptive weight assignment, multi-objective optimization, and geographical feature coupling perturbation injection.

[0041] The embodiment of the present application also proposes a dynamic load testing system for a short message communication system, including a processor and a memory. A computer program is stored on the memory, and when the computer program is executed by the processor, the steps of the dynamic load testing method for the short message communication system as described above are implemented.

[0042] In addition, although exemplary embodiments have been described herein, their scope includes any and all embodiments based on the present disclosure having equivalent elements, modifications, omissions, combinations (e.g., solutions that cross various embodiments), adaptations, or changes. It is not limited to the examples described in this specification or during the implementation of the present application, and the examples will be interpreted as non-exclusive.

[0043] The above description is intended to be illustrative rather than restrictive. For example, the above examples (or one or more of their solutions) can be used in combination with each other. For example, those of ordinary skill in the art can use other embodiments when reading the above description.

[0044] The above embodiments are only exemplary embodiments of the present disclosure. Those skilled in the art can make various modifications or equivalent replacements to the present invention within the essence and protection scope of the present disclosure, and such modifications or equivalent replacements should also be regarded as falling within the protection scope of the present invention.

Claims

1. A dynamic load testing method for a short message communication system, characterized in that, Applied to an emergency communication scenario based on Beidou short message communication. The emergency communication scenario includes multiple secondary nodes and a primary node. The secondary nodes send data messages to the primary node based on Beidou short messages. The dynamic load testing method includes: Pre-divide data messages into multiple priority categories. Among them, the data messages with higher priority are adapted to short message communication transmission, and the messages with lower priority use hybrid channel transmission; Obtain data messages in multiple historical emergency scenarios and classify them according to priority categories; Align the data messages classified by each secondary node in terms of clock, and select data message samples in a typical period, where the typical period is a period in which the total amount of data messages within it exceeds a preset threshold; Extract the message features of each data message in any typical period, assign weights to the corresponding message features according to the priority of each data message, and use LSTM to construct a dynamic load generator; Extract the message features in the current emergency scenario and input them into the dynamic load generator; Compare based on the output of the dynamic load generator and the load situation of the current communication system to complete the dynamic load test.

2. The dynamic load testing method for the short message communication system according to claim 1, characterized in that Align the data messages classified by each secondary node in terms of clock, and selecting data message samples in a typical period includes: Align the data messages classified by each secondary node according to the timestamps under a unified time coordinate; Intercept according to a unified time window under the time coordinate; The period in which the total amount of data messages in the intercepted data segment exceeds the preset threshold, and the specified number of adjacent previous periods are used as data message samples in the typical period.

3. The dynamic load testing method for the short message communication system according to claim 1, characterized in that It also includes adding perturbations to the data message samples in some typical periods in at least one of the following ways: Obtain multiple high-priority messages in the data message samples of each typical period, and randomly insert high-priority messages within the typical period, where the insertion frequency follows a Poisson distribution that satisfies the preset burst intensity; For the data message sample of any typical period as itself, and / or, add a random time offset of ±Δt to the message time of the high-priority messages in the time windows before and after it; Add Gaussian noise to the data fields in the data messages, and randomly discard some non-keyword fields.

4. The dynamic load testing method for the short message communication system according to claim 3, wherein It also includes adding spatio-temporal perturbations to the data message samples in the following way: Obtain the geographical distribution data of historical disaster events in the communication coverage area; Determine the intersection area with the target monitoring range in the geographical distribution data; Extract the geographical feature information of the intersection area and determine the similarity between the extracted geographical feature information and the local geographical features of the target monitoring range; Based on the local area corresponding to the local geographical features with a similarity higher than the preset threshold, logically related data messages are associated and inserted.

5. The dynamic load testing method for the short message communication system according to claim 4, characterized in that When the target monitoring range includes a river channel, the extracted geographical feature information and local geographical features include: terrain features, hydrological features, and meteorological features, where: The terrain features are obtained based on DEM data; The hydrological features include river channel hydrological data, historical flood frequency, and soil permeability coefficient; The meteorological features include extreme rainfall amount and extreme rainfall event frequency; Adopt a random forest model to dynamically allocate the weights of features based on the relevance between historical emergency events and geographical features.

6. The dynamic load testing method for the short message communication system according to claim 4, characterized in that Determining the similarity between the extracted geographical feature information and the local geographical features of the target monitoring range also includes introducing a time dimension and superimposing a time decay factor when calculating the geographical feature similarity, satisfying: Among them, represents the weight of the i th type of geographical feature, represents the similarity of the i th type of geographical feature after cosine similarity or Euclidean distance normalization, represents the timestamp of the current data packet, represents the time of the historical emergency event, is the time decay coefficient, which is used to make the impact of recent emergency events greater.

7. The dynamic load testing method for the short message communication system according to claim 1, characterized in that, Using LSTM to construct a dynamic load generator includes: Obtain the historical load data corresponding to any typical period and extract the load features of the historical load data, where the load data includes channel packet loss rate, end-to-end delay, CPU, and memory occupancy rate; Take the packet features with weights assigned to any typical period and the corresponding load features as a group and input them into the LSTM to generate a predicted load curve using the LSTM to construct a dynamic load generator.

8. The dynamic load testing method for the short message communication system according to claim 7, characterized in that It also includes configuring the comprehensive optimization objective function for load testing: Among them, , , are adjustable hyperparameters, represents the load coverage rate at the typical time period t, the average success rate of high-priority packets at the typical time period t, represents the channel overload risk at the typical time period t; Solve the dynamic optimization parameter combination in the comprehensive optimization objective function to balance the test efficiency and system stability in real time.

9. A dynamic load testing system for a short message communication system, characterized in that, It includes a processor and a memory, and a computer program is stored on the memory. When the computer program is executed by the processor, it realizes the steps of the dynamic load testing method of the short message communication system as described in any one of claims 1 to 7.

Images