Abnormal device determination method, device, electronic device and storage medium

Abstract

The present disclosure provides an abnormal device determination method, apparatus, electronic device and storage medium, which relate to the field of artificial intelligence technology, and in particular to the field of collective communication technology. The specific implementation scheme of the method is: in response to the communication duration of the distributed collective communication system being greater than a predetermined threshold, the operation logs of multiple communication units used to construct the distributed collective communication system are obtained; the distributed collective communication system is used to perform a data transmission task of a deep learning model, and the operation log includes the identification of multiple communication units and the data transmission status of multiple communication units; according to the identification of multiple communication units and the data transmission status of multiple communication units, a target communication unit is determined from multiple communication units, and the data transmission status of the target communication unit is different from the data transmission status of other communication units used to perform the same data transmission task; and according to the identification of the target communication unit and the predetermined deployment quantity, the abnormal device corresponding to the target communication unit is determined.

Description

Technical Field

[0001] The present disclosure relates to the field of artificial intelligence technology, in particular to the field of collective communication technology, and specifically to an abnormal device determination method, device, electronic device, and storage medium. Background Art

[0002] With the widespread application of deep learning technology, the training data, model parameter scale and hardware resource requirements of deep learning models have also grown rapidly, and the probability of data transmission failure due to hardware failure has also increased.

[0003] Therefore, in the scenario of large-scale collective communication, there is an urgent need for a method to quickly and accurately identify abnormal devices in hardware resources to reduce the impact of hardware failures on the training and application of deep learning models. Summary of the invention

[0004] The present disclosure provides a method, an apparatus, an electronic device and a storage medium for determining an abnormal device.

[0005] According to one aspect of the present disclosure, a method for determining an abnormal device is provided, comprising: in response to a communication duration of a distributed collective communication system being greater than a predetermined threshold, obtaining operation logs of multiple communication units used to construct a distributed collective communication system; the distributed collective communication system is used to perform a data transmission task of a deep learning model, and the operation logs include identifications of the multiple communication units and data transmission states of the multiple communication units; determining a target communication unit from the multiple communication units based on the identifications of the multiple communication units and the data transmission states of the multiple communication units, the data transmission state of the target communication unit being different from the data transmission states of other communication units used to perform the same data transmission task; and determining an abnormal device corresponding to the target communication unit based on the identification of the target communication unit and a predetermined deployment number, wherein the predetermined deployment number indicates the number of communication units deployed by each device in the distributed collective communication system.

[0006] According to another aspect of the present disclosure, an abnormal device determination device is provided, including: an acquisition module, a first determination module, and a second determination module. The acquisition module is used to acquire operation logs of multiple communication units used to construct a distributed collective communication system in response to the communication duration of the distributed collective communication system being greater than a predetermined threshold; the distributed collective communication system is used to perform a data transmission task of a deep learning model, and the operation log includes the identification of multiple communication units and the data transmission status of the multiple communication units. The first determination module is used to determine a target communication unit from multiple communication units based on the identification of multiple communication units and the data transmission status of multiple communication units, and the data transmission status of the target communication unit is different from the data transmission status of other communication units used to perform the same data transmission task. The second determination module is used to determine the abnormal device corresponding to the target communication unit based on the identification of the target communication unit and a predetermined deployment number, wherein the predetermined deployment number indicates the number of communication units deployed on each device in the distributed collective communication system.

[0007] According to another aspect of the present disclosure, an electronic device is provided, comprising: at least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor so that the at least one processor can execute the method described above.

[0008] According to another aspect of the present disclosure, a non-transitory computer-readable storage medium storing computer instructions is provided, wherein the computer instructions are used to enable the computer to execute the method described above.

[0009] According to another aspect of the present disclosure, a computer program product is provided, including a computer program, and when the computer program is executed by a processor, the method described above is implemented.

[0010] It should be understood that the content described in this section is not intended to identify the key or important features of the embodiments of the present disclosure, nor is it intended to limit the scope of the present disclosure. Other features of the present disclosure will become easily understood through the following description. BRIEF DESCRIPTION OF THE DRAWINGS

[0011] The accompanying drawings are used to better understand the present solution and do not constitute a limitation of the present disclosure.

[0012] Figure 1 An exemplary system architecture to which the abnormal device determination method and apparatus according to an embodiment of the present disclosure can be applied is schematically shown;

[0013] Figure 2 A flowchart of a method for determining an abnormal device according to an embodiment of the present disclosure is schematically shown;

[0014] Figure 3A A schematic diagram schematically shows a method of determining a target communication unit from a plurality of communication units according to an embodiment of the present disclosure;

[0015] Figure 3B A schematic diagram schematically shows a method of determining a target communication unit from a plurality of communication units according to another embodiment of the present disclosure;

[0016] Figure 4A A schematic diagram of normal transmission between communication units in a symmetric transmission scenario according to an embodiment of the present disclosure is schematically shown;

[0017] Figure 4B A schematic diagram of abnormal transmission between communication units in a symmetric transmission scenario according to an embodiment of the present disclosure is schematically shown;

[0018] Figure 4C A schematic diagram of normal transmission between communication units in an asymmetric transmission scenario according to an embodiment of the present disclosure is schematically shown;

[0019] Figure 4D A schematic diagram of abnormal transmission between communication units in an asymmetric transmission scenario according to an embodiment of the present disclosure is schematically shown;

[0020] Figure 5 A block diagram schematically shows an abnormal device determination apparatus according to an embodiment of the present disclosure; and

[0021] Figure 6 A block diagram of an electronic device suitable for implementing the abnormal device determination method according to an embodiment of the present disclosure is schematically shown. DETAILED DESCRIPTION

[0022] The following is a description of exemplary embodiments of the present disclosure in conjunction with the accompanying drawings, including various details of the embodiments of the present disclosure to facilitate understanding, which should be considered as merely exemplary. Therefore, it should be recognized by those of ordinary skill in the art that various changes and modifications may be made to the embodiments described herein without departing from the scope and spirit of the present disclosure. Similarly, for the sake of clarity and conciseness, descriptions of well-known functions and structures are omitted in the following description.

[0023] In the scenario of large-scale collective communication, due to the large amount of data transmission, the data to be transmitted is usually divided or sliced ​​into blocks or slices to obtain data blocks or slices. Then the data blocks or slices are transmitted using the communication unit deployed in the device to meet the needs of large-scale data transmission in deep learning model training and application scenarios.

[0024] However, as the scale of training data and model parameters of deep learning models increases, the number of communication units used to perform data transmission tasks of deep learning models is also increasing, and the number of devices used to deploy communication units is also increasing.

[0025] In the training process of deep learning models, data parallel transmission strategies are generally adopted, such as shard parallel transmission, tensor parallel transmission, pipeline parallel transmission, etc. In the parallel transmission scenario, the data transmission of the communication units used to perform the same data transmission task should be synchronized. When any of the communication units fails, the data transmission of the entire collective communication will be affected. Therefore, it is very important to quickly and accurately locate the device where the faulty communication unit is located among a large number of communication units.

[0026] In view of this, the embodiments of the present disclosure analyze the common faults in the collective communication scenarios - data transmission interruption or stuck faults (Hang faults), and find that common Hang faults can be divided into the following three scenarios: Scenario 1. Part of the data to be transmitted enters the communication state of the communication group, waiting for data that has not entered the communication group; Scenario 2. All the data to be transmitted enters the communication state of the communication group, and all are stuck in the communication group; Scenario 3. All the data to be transmitted enters the communication state of the communication group, and after the communication is completed, only part of the data is output from the communication group.

[0027] It should be noted that the communication group in the embodiment of the present disclosure is a logical concept, which is composed of communication units for performing the same data transmission task. The communication units of the same communication group can be deployed in the same device or in different devices.

[0028] According to the embodiments of the present disclosure, during the collective communication process, the data to be transmitted is divided into multiple data blocks through logical segmentation. The state of scenario 1 is possible only when the communication reaches the last data block. The probability of occurrence in large-scale collective communication scenarios is low, while the probability of occurrence of scenarios 2 and 3 is high.

[0029] In view of this, the embodiments of the present disclosure mainly provide a method for determining abnormal devices for Hang failures occurring in scenarios 2 and 3. By obtaining the operation logs of multiple communication units used to construct a distributed collective communication system, a target communication unit is determined from multiple communication units according to the identifications of the multiple communication units and the data transmission status of the multiple communication units. By analyzing the actual data transmission status of each communication unit in the operation log, a communication unit with a different data transmission status from other communication units used to perform the same data transmission task is determined as the target communication unit with an abnormality, thereby achieving fine-grained troubleshooting of abnormal communication units, and being able to quickly and accurately determine abnormal communication units and devices. Moreover, the process of troubleshooting does not require information interaction between communication units, so this method has little impact on the performance of the collective communication system when executed.

[0030] Figure 1 An exemplary system architecture to which the abnormal device determination method and apparatus according to an embodiment of the present disclosure can be applied is schematically shown.

[0031] It should be noted that Figure 1 The examples shown are only examples of system architectures to which the embodiments of the present disclosure can be applied, in order to help those skilled in the art understand the technical content of the present disclosure, but do not mean that the embodiments of the present disclosure cannot be used in other devices, systems, environments or scenarios. For example, in another embodiment, the exemplary system architecture to which the abnormal device determination method and apparatus can be applied may include a terminal device, but the terminal device may implement the abnormal device determination method and apparatus provided by the embodiments of the present disclosure without interacting with the server.

[0032] like Figure 1 As shown, the system architecture 100 according to this embodiment may include terminal devices 101, 102, 103, a network 104 and a server 105. The network 104 is used to provide a medium for communication links between the terminal devices 101, 102, 103 and the server 105. The network 104 may include various connection types, such as wired and / or wireless communication links, etc.

[0033] Users can use terminal devices 101, 102, 103 to interact with server 105 through network 104 to receive or send messages, etc. Various communication client applications can be installed on terminal devices 101, 102, 103, such as knowledge reading applications, web browser applications, search applications, instant messaging tools, email clients and / or social platform software, etc. (only as examples).

[0034] The terminal devices 101 , 102 , and 103 may be various electronic devices having a display screen and supporting web browsing, including but not limited to smart phones, tablet computers, laptop computers, and desktop computers.

[0035] The server 105 may be a server that provides various services, such as a background management server (only an example) that provides support for the content browsed by the user using the terminal devices 101, 102, and 103. The background management server may analyze and process the received data such as user requests, and feed back the processing results (such as web pages, information, or data obtained or generated according to user requests) to the terminal device.

[0036] It should be noted that the abnormal device determination method provided in the embodiment of the present disclosure can generally be executed by the terminal device 101, 102, or 103. Accordingly, the abnormal device determination apparatus provided in the embodiment of the present disclosure can also be set in the terminal device 101, 102, or 103.

[0037] Alternatively, the abnormal device determination method provided in the embodiment of the present disclosure may also be generally performed by the server 105. Accordingly, the abnormal device determination apparatus provided in the embodiment of the present disclosure may generally be arranged in the server 105. The abnormal device determination method provided in the embodiment of the present disclosure may also be performed by a server or server cluster that is different from the server 105 and can communicate with the terminal devices 101, 102, 103 and / or the server 105. Accordingly, the abnormal device determination apparatus provided in the embodiment of the present disclosure may also be arranged in a server or server cluster that is different from the server 105 and can communicate with the terminal devices 101, 102, 103 and / or the server 105.

[0038] For example, when the distributed collective communication system performs the data transmission task of the deep learning model, when the terminal devices 101, 102, and 103 have a communication timeout in the distributed collective communication system, the operation logs of multiple communication units in the terminal devices 101, 102, and 103 can be obtained, and then the operation logs can be sent to the server 105, and the server 105 analyzes the operation logs, determines the target communication unit from the multiple communication units, and determines the abnormal device according to the identification and the predetermined deployment number of the target communication unit. Alternatively, a server or server cluster that can communicate with the terminal devices 101, 102, 103 and / or the server 105 analyzes the operation logs and finally determines the abnormal device.

[0039] It should be understood that Figure 1 The number of terminal devices, networks and servers in the embodiment is only for illustration. Any number of terminal devices, networks and servers may be provided according to implementation requirements.

[0040] In the technical solution of the present disclosure, the collection, storage, use, processing, transmission, provision, disclosure and application of user personal information involved comply with the provisions of relevant laws and regulations, take necessary confidentiality measures, and do not violate public order and good morals.

[0041] In the technical solution of the present disclosure, the user's authorization or consent is obtained before obtaining or collecting the user's personal information.

[0042] Figure 2 The flowchart of the abnormal device determination method according to the embodiment of the present disclosure is schematically shown.

[0043] like Figure 2 As shown, the method 200 includes operations S210 to S230.

[0044] In operation S210, in response to a communication duration of the distributed collective communication system being greater than a predetermined threshold, operation logs of a plurality of communication units used to construct the distributed collective communication system are acquired.

[0045] In operation S220, a target communication unit is determined from among the plurality of communication units according to the identifications of the plurality of communication units and the data transmission states of the plurality of communication units.

[0046] In operation S230, an abnormal device corresponding to the target communication unit is determined according to the identification of the target communication unit and the predetermined deployment quantity.

[0047] According to an embodiment of the present disclosure, a distributed collective communication system is used to perform data transmission tasks of a deep learning model. The distributed collective communication system can be constructed by multiple communication units, and the multiple communication units can be deployed on the same device or on different devices. The communication unit can be a GPU (Graphics Processing Unit).

[0048] According to an embodiment of the present disclosure, when executing a data transmission task of a deep learning model, the same data transmission task can be performed by a group of communication units, and the data transmission of the group of communication units should be synchronized. When a Hang failure occurs in any of the communication units, the data transmission status of the communication unit with the Hang failure will differ from the data transmission status of other communication units in the group.

[0049] According to an embodiment of the present disclosure, the operation log may include identifications of the plurality of communication units and data transmission states of the plurality of communication units.

[0050] For example: For a communication group used to perform data transmission task A, the communication group may include 4 communication units. The data transmission tasks of the training and application scenarios of deep learning models are generally iterative data processing processes. When each communication unit performs data transmission, each completed data transmission can be recorded once in the operation log. Therefore, the target communication unit can be determined by comparing whether the number of completed data transmissions of the 4 communication units is consistent. For example: the first communication unit completed 5 transmissions, the second communication unit completed 5 transmissions, the third communication unit completed 5 transmissions, and the fourth communication unit completed 4 transmissions. The target communication unit can be determined to be the fourth communication unit.

[0051] According to an embodiment of the present disclosure, the predetermined deployment quantity indicates the number of communication units deployed by each device in the distributed collective communication system. According to the identification of the target communication unit and the predetermined deployment quantity, an abnormal device corresponding to the target communication unit is determined.

[0052] For example, if 8 communication units are deployed on each device, the identification of the communication unit can be a global identification, and the distributed collective communication system can include 10 devices, then the identification of the communication unit can be from 0 to 79. The identification of the target communication unit can be 18, and the identification of the device 2 can be obtained by rounding the quotient of the identification of the target communication unit and the predetermined number of deployments, so that the abnormal device can be determined as the device with identification 2.

[0053] According to an embodiment of the present disclosure, by obtaining the operation logs of multiple communication units used to construct a distributed collective communication system, a target communication unit is determined from the multiple communication units according to the identifications of the multiple communication units and the data transmission status of the multiple communication units, and by analyzing the actual data transmission status of each communication unit in the operation log, a communication unit with a different data transmission status from other communication units used to perform the same data transmission task is determined as a target communication unit with an abnormality, thereby achieving fine-grained troubleshooting of abnormal communication units, and being able to quickly and accurately determine the communication units and devices with abnormalities. Moreover, the process of troubleshooting does not require information interaction between communication units, so the method has little impact on the performance of the collective communication system when executed.

[0054] According to an embodiment of the present disclosure, determining a target communication unit from multiple communication units based on the identifications of the multiple communication units and the data transmission status within the multiple communication units may include the following operations: dividing the multiple communication units according to the identifications of the multiple communication units to obtain multiple communication groups; for the multiple communication units within each communication group, determining the target communication unit from the multiple communication units according to the data transmission status of the multiple communication units.

[0055] According to an embodiment of the present disclosure, the identification of a communication unit may include a type identification for identifying a data transmission task. Typically, data transmission tasks of the same type are assigned to communication units within the same communication group. Within the communication group, the communication units are used to execute the same data transmission task according to a parallel strategy.

[0056] Figure 3A The diagram schematically shows a schematic diagram of determining a target communication unit from multiple communication units according to an embodiment of the present disclosure.

[0057] like Figure 3A As shown, in embodiment 300A, the communication group 321 includes communication unit Cu1 321_1 , communication unit Cu2 321_2 , communication unit Cu3 321_3 , and communication unit Cu4 321_4 .

[0058] According to an embodiment of the present disclosure, the data transmission state includes a data transmission task identifier. The data transmission task identifier can be used to indicate that the currently completed data transmission task is a task in which iteration round of the deep learning model.

[0059] According to an embodiment of the present disclosure, the data transmission task identifiers of the multiple communication units in each communication group may be analyzed to obtain a first analysis result.

[0060] For example: In the data transmission state 322A, the data transmission state of the communication unit Cu1321_1 is task T 101 Completed 322_1 indicates that the communication unit Cu1 321_1 has completed the data transmission task in the 101st iteration of the deep learning model. Similarly, the communication unit Cu2 321_2 and the communication unit Cu3 321_3 have also completed the data transmission task in the 101st iteration of the deep learning model. However, the data transmission status of the communication unit Cu4 321_4 is task T 100 Completed 322_4, indicating that communication unit Cu4321_4 has completed the data transmission task in the 100th iteration round of the deep learning model. By comparing the data transmission task identifiers of communication unit Cu1321_1, communication unit Cu2321_2, communication unit Cu3321_3 and communication unit Cu4321_4, the first analysis result that can be obtained is "the data transmission task identifier of communication unit Cu4321_4 is less than the data transmission task identifier of communication unit Cu3321_3", that is, the data transmission status of communication unit Cu4321_4 is different from that of other communication units 323.

[0061] Therefore, according to the first analysis result, the communication unit Cu4321_4 can be determined as the target communication unit 324.

[0062] According to an embodiment of the present disclosure, by comparing the identifiers of the data transmission tasks of the communication units in each communication group, the abnormality troubleshooting process can be reduced to one communication group, thereby narrowing the troubleshooting scope in the target communication unit determination process and improving processing efficiency.

[0063] Since in the collective communication process, for the data transmission task, the components include the transmission process and the transmission preparation process, therefore, in order to more accurately determine the transmission status of the communication units in the communication group, the comparison of the transmission tasks can be added on the basis of comparing the transmission task identifiers.

[0064] Figure 3B The diagram schematically shows a schematic diagram of determining a target communication unit from multiple communication units according to an embodiment of the present disclosure.

[0065] like Figure 3B As shown, in embodiment 300B, the communication group 321 includes communication unit Cu1 321_1 , communication unit Cu2 321_2 , communication unit Cu3 321_3 and communication unit Cu4 321_4 .

[0066] According to an embodiment of the present disclosure, the data transmission status also includes: a transmission progress corresponding to the data transmission task. For multiple communication units in each communication group, in response to the data transmission task identifiers of the multiple communication units being the same, the transmission progress corresponding to the data transmission task of the multiple communication units can be analyzed to obtain a second analysis result.

[0067] For example, in the data transmission state 322B, the data transmission state corresponding to the communication unit Cu1321_1 is task T 101 Completed 322_1, the data transmission status corresponding to the communication unit Cu2321_2 is task T 101 Completed 322_2, the data transmission status corresponding to the communication unit Cu3321_3 is task T 101 Completed 322_3, the data transmission status corresponding to the communication unit Cu4321_4 is task T 101 By comparing the data transmission progress of communication unit Cu1 321_1, communication unit Cu2 321_2, communication unit Cu3 321_3 and communication unit Cu4 321_4, the second analysis result can be obtained as "the data transmission task progress of communication unit Cu4 321_4 is less than the data transmission task progress of communication unit Cu3 321_3", that is, the data transmission status of communication unit Cu4 321_4 is different from that of other communication units 326.

[0068] Therefore, according to the second analysis result, the communication unit Cu4321_4 can be determined as the target communication unit 327.

[0069] According to an embodiment of the present disclosure, on the basis of comparing transmission task identifiers, comparison of transmission tasks is added to more accurately determine the transmission status of communication units in a communication group, thereby improving the accuracy of abnormal device determination.

[0070] According to an embodiment of the present disclosure, the identification of the target device can be obtained according to the identification of the target communication unit and the predetermined deployment quantity; and the target device can be determined according to the identification of the target device.

[0071] For example, if 8 communication units are deployed on each device, the identification of the communication unit can be a global identification, and the distributed collective communication system can include 10 devices, then the identification of the communication unit can be from 0 to 79. The identification of the target communication unit can be 18, and the identification of the device 2 can be obtained by rounding the quotient of the identification of the target communication unit and the predetermined number of deployments, so that the abnormal device can be determined as the device with identification 2.

[0072] In the scenario of large-scale distributed collective communication, in most cases, the communication units used to perform the same task are deployed in the same device, for example, in the tensor parallel collective communication scenario of the deep learning model, so the method described above can quickly and accurately identify abnormal devices, so as to remove abnormal devices from the distributed collective communication system and reduce the impact of abnormal devices on the deep learning model training process.

[0073] However, the communication units in the same communication group may also be deployed in multiple devices. In this case, due to the large number of communication units, it is not conducive to quickly identifying the abnormal device that has failed.

[0074] In view of this, the abnormal device determination method provided in the embodiment of the present disclosure can also determine the communication unit with Hang failure according to the transmission characteristics of the distributed collective communication. For example: according to the data transmission type of the distributed collective communication system, the data transmission characteristics are obtained; and according to the data transmission status and data transmission characteristics of the multiple communication units, the target communication unit is determined from the multiple communication units.

[0075] According to an embodiment of the present disclosure, in a distributed collective communication system, the characteristics of data transmission correspond to the collective communication algorithm. The characteristics of data transmission can be divided into symmetric transmission and asymmetric transmission. The data transmission state can represent the number of data blocks that the communication unit has completed transmission.

[0076] Figure 4A A schematic diagram of normal transmission between communication units in a symmetric transmission scenario according to an embodiment of the present disclosure is schematically shown.

[0077] like Figure 4AAs shown, in the symmetric transmission scenario shown in embodiment 400A, the number of data blocks transmitted between each communication unit should be the same. Figure 4A The black squares in the figure represent data blocks.

[0078] For example, the number of data blocks transmitted by communication unit Cua421 to communication unit Cub422, the number of data blocks transmitted by communication unit Cub422 to communication unit Cuc423, the number of data blocks transmitted by communication unit Cuc423 to communication unit Cud424, and the number of data blocks transmitted by communication unit Cud424 to communication unit Cue425 are all 3.

[0079] Figure 4B A schematic diagram of abnormal transmission between communication units in a symmetric transmission scenario according to an embodiment of the present disclosure is schematically shown.

[0080] like Figure 4B As shown, in the symmetric transmission scenario shown in embodiment 400B, a difference occurs in the number of data blocks transmitted between each communication unit, indicating a Hang failure occurs in the communication group.

[0081] According to an embodiment of the present disclosure, the data transmission status of the sth communication unit can be compared with the data transmission status of the previous s-1 communication units, where s is an integer greater than or equal to 1 and less than or equal to S; and in response to the data transmission characteristic being symmetric transmission and the data transmission status of the sth communication unit being different from the data transmission status of the previous s-1 communication units, a target communication unit is determined from multiple communication units.

[0082] For example: the number of data blocks transmitted by communication unit Cua421 to communication unit Cub422, the number of data blocks transmitted by communication unit Cub422 to communication unit Cuc423, and the number of data blocks transmitted by communication unit Cuc423 to communication unit Cud424 are all still 3, but the number of data blocks transmitted by communication unit Cud424 to communication unit Cue425 is 2.

[0083] Therefore, it can be determined that the data transmission state of the communication unit Cue425 is different from the data transmission states of other communication units, and it can be determined that the communication unit Cue425 is the target communication unit, that is, the communication unit with the Hang failure.

[0084] Figure 4C A schematic diagram of normal transmission between communication units in an asymmetric transmission scenario according to an embodiment of the present disclosure is schematically shown.

[0085] like Figure 4CAs shown, in the asymmetric transmission scenario shown in Example 400C, the number of data blocks transmitted between each communication unit should decrease step by step, and the difference in the number of data blocks transmitted between two adjacent communication units is the same as the difference in the number of data blocks transmitted between other two adjacent communication units. Figure 4C The black squares in the figure represent data blocks.

[0086] For example, the number of data blocks transmitted from communication unit Cua421 to communication unit Cub422 is 4, the number of data blocks transmitted from communication unit Cub422 to communication unit Cuc423 is 3, the number of data blocks transmitted from communication unit Cuc423 to communication unit Cud424 is 2, and the number of data blocks transmitted from communication unit Cud424 to communication unit Cue425 is 1. The difference in the number of data blocks transmitted by adjacent communication units is 1.

[0087] Figure 4D A schematic diagram of abnormal transmission between communication units in an asymmetric transmission scenario according to an embodiment of the present disclosure is schematically shown.

[0088] like Figure 4D As shown, in the asymmetric transmission scenario shown in embodiment 400D, the number of data blocks transmitted between each communication unit no longer decreases step by step, indicating that a Hang failure occurs in the communication group.

[0089] According to an embodiment of the present disclosure, a first difference between the data transmission state of the tth communication unit and the data transmission state of the t-1th communication unit is compared with a second difference between the data transmission state of the tth communication unit and the data transmission state of the t+1th communication unit, where t is an integer greater than or equal to 1 and less than or equal to T; and in response to the data transmission characteristic being asymmetric transmission and the first difference and the second difference satisfying a preset condition, a target communication unit is determined from multiple communication units.

[0090] According to an embodiment of the present disclosure, the preset condition includes: the first difference is not equal to the second difference.

[0091] For example: the number of data blocks transmitted by communication unit Cua421 to communication unit Cub422 is 4, the number of data blocks transmitted by communication unit Cub422 to communication unit Cuc423 is 2, the number of data blocks transmitted by communication unit Cuc423 to communication unit Cud424 is 2, and the number of data blocks transmitted by communication unit Cud424 to communication unit Cue425 is 1.

[0092] Therefore, the difference between the number of data blocks that communication unit Cua421 has completed transmitting and the number of data blocks that communication unit Cub422 has completed transmitting is 2; the difference between the number of data blocks that communication unit Cub422 has completed transmitting and the number of data blocks that communication unit Cuc423 has completed transmitting is 0, and communication unit Cuc423 can be determined as the target communication unit.

[0093] According to the embodiments of the present disclosure, by analyzing the data transmission status of the communication unit based on the data transmission characteristics, the target communication unit deployed on different devices can be accurately determined, thereby accurately determining the abnormal device.

[0094] Figure 5 The block diagram of the abnormal device determining apparatus according to an embodiment of the present disclosure is schematically shown.

[0095] like Figure 5 As shown, the device 500 may include: an acquisition module 510 , a first determination module 520 , and a second determination module 530 .

[0096] The acquisition module 510 is used to obtain the operation logs of multiple communication units used to construct the distributed collective communication system in response to the communication duration of the distributed collective communication system being greater than a predetermined threshold; the distributed collective communication system is used to execute the data transmission task of the deep learning model, and the operation log includes the identification of the multiple communication units and the data transmission status of the multiple communication units.

[0097] The first determination module 520 is used to determine a target communication unit from multiple communication units based on the identifiers of the multiple communication units and the data transmission status of the multiple communication units, wherein the data transmission status of the target communication unit is different from the data transmission status of other communication units used to perform the same data transmission task.

[0098] The second determination module 530 is used to determine the abnormal device corresponding to the target communication unit according to the identifier of the target communication unit and the predetermined deployment quantity, wherein the predetermined deployment quantity indicates the number of communication units deployed in each device in the distributed collective communication system.

[0099] According to an embodiment of the present disclosure, the first determination module may include: a division submodule and a first determination submodule. The division submodule is used to divide the multiple communication units according to the type of data transmission task to obtain multiple communication groups. The first determination submodule is used to determine the target communication unit from the multiple communication units according to the data transmission status of the multiple communication units in each communication group.

[0100] According to an embodiment of the present disclosure, the data transmission status includes a data transmission task identifier. The first determination submodule may include: a first molecular unit, configured to analyze the data transmission task identifiers of multiple communication units in each communication group to obtain a first analysis result. A first determination unit, configured to determine a target communication unit from the multiple communication units based on the first analysis result.

[0101] According to an embodiment of the present disclosure, the communication group includes I communication units, where I is an integer greater than 1, and the first analysis unit may include: a first comparison subunit and a first generation subunit. The first comparison subunit is used to compare the data transmission task identifier of the i-th communication unit with the data transmission task identifier of the i+1-th communication unit, where i is an integer greater than or equal to 1 and less than or equal to I. The first generation subunit is used to generate a first analysis result in response to the data transmission task identifier of the i-th communication unit being greater than the data transmission task identifier of the i+1-th communication unit.

[0102] According to an embodiment of the present disclosure, the first determining unit includes: a first determining subunit, configured to determine, according to the first analysis result, the (i+1)th communication unit as a target communication unit.

[0103] According to an embodiment of the present disclosure, the data transmission status also includes: a transmission progress corresponding to the data transmission task, and the first determination submodule includes: a second analysis unit and a second determination unit. The second analysis unit is used to analyze the transmission progress corresponding to the data transmission task of the multiple communication units in each communication group in response to the same data transmission task identifier of the multiple communication units, and obtain a second analysis result. The second determination unit is used to determine the target communication unit from the multiple communication units according to the second analysis result.

[0104] According to an embodiment of the present disclosure, the communication group includes I communication units, where I is an integer greater than 1, and the second analysis unit includes: a second comparison subunit and a second generation subunit. The second comparison subunit is used to compare the transmission progress of the i-th communication unit with the transmission progress of the i+1-th communication unit, where i is an integer greater than or equal to 1 and less than or equal to I. The second generation subunit is used to generate a second analysis result in response to the transmission progress of the i-th communication unit being greater than the transmission progress of the i+1-th communication unit.

[0105] According to an embodiment of the present disclosure, the second determining unit includes: a second determining subunit, configured to determine, according to the second analysis result, the (i+1)th communication unit as the target communication unit.

[0106] According to an embodiment of the present disclosure, the above-mentioned device further includes: an acquisition module and a third determination module. The acquisition module is used to obtain data transmission characteristics according to the data transmission type of the distributed collective communication system. The third determination module is used to determine the target communication unit from the multiple communication units according to the data transmission status and data transmission characteristics of the multiple communication units.

[0107] According to an embodiment of the present disclosure, the plurality of communication units includes S, where S is an integer greater than 1, and the third determination module includes: a first comparison submodule and a second determination submodule. The first comparison submodule is used to compare the data transmission state of the sth communication unit with the data transmission state of the first s-1 communication units, where s is an integer greater than or equal to 1 and less than or equal to S. The second determination submodule is used to determine the target communication unit from the plurality of communication units in response to the data transmission feature being symmetrical transmission and the data transmission state of the sth communication unit being different from the data transmission state of the first s-1 communication units.

[0108] According to an embodiment of the present disclosure, the second determination submodule includes: a third determination unit, which is used to determine the sth communication unit as the target communication unit from multiple communication units in response to the data transmission characteristic being symmetric transmission and the data transmission state of the sth communication unit being different from the data transmission state of the previous s-1 communication units.

[0109] According to an embodiment of the present disclosure, the plurality of communication units includes T, where T is an integer greater than 1, and the third determination module includes: a second comparison submodule and a third determination submodule. The second comparison submodule is used to compare a first difference between a data transmission state of the tth communication unit and a data transmission state of the t-1th communication unit and a second difference between a data transmission state of the tth communication unit and a data transmission state of the t+1th communication unit, where t is an integer greater than or equal to 1 and less than or equal to T. The third determination submodule is used to determine a target communication unit from the plurality of communication units in response to the data transmission feature being asymmetric transmission and the first difference and the second difference satisfying a preset condition.

[0110] According to an embodiment of the present disclosure, the third determination submodule includes: a fourth determination unit, used to determine the t+1th communication unit as the target communication unit from multiple communication units in response to the data transmission characteristic being asymmetric transmission and the first difference and the second difference satisfying a preset condition.

[0111] According to an embodiment of the present disclosure, the preset condition includes: the first difference is not equal to the second difference.

[0112] According to an embodiment of the present disclosure, the second determination module includes: an acquisition submodule and a fourth determination submodule. The acquisition submodule is used to obtain the identification of the target device according to the identification of the target communication unit and the predetermined deployment quantity. The fourth determination submodule is used to determine the target device according to the identification of the target device.

[0113] According to an embodiment of the present disclosure, the present disclosure also provides an electronic device, a readable storage medium and a computer program product.

[0114] According to an embodiment of the present disclosure, an electronic device includes: at least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor so that the at least one processor can execute the method as described above.

[0115] According to an embodiment of the present disclosure, a non-transitory computer-readable storage medium storing computer instructions is provided, wherein the computer instructions are used to cause a computer to execute the method as described above.

[0116] According to an embodiment of the present disclosure, a computer program product includes a computer program, and when the computer program is executed by a processor, the computer program implements the method as described above.

[0117] Figure 6 A schematic block diagram of an example electronic device 600 that can be used to implement an embodiment of the present disclosure is shown. The electronic device is intended to represent various forms of digital computers, such as laptop computers, desktop computers, workstations, personal digital assistants, servers, blade servers, mainframe computers, and other suitable computers. The electronic device can also represent various forms of mobile devices, such as personal digital processing, cellular phones, smart phones, wearable devices, and other similar computing devices. The components shown herein, their connections and relationships, and their functions are merely examples and are not intended to limit the implementation of the present disclosure described and / or required herein.

[0118] like Figure 6 As shown, the device 600 includes a computing unit 601, which can perform various appropriate actions and processes according to a computer program stored in a read-only memory (ROM) 602 or a computer program loaded from a storage unit 608 into a random access memory (RAM) 603. In the RAM 603, various programs and data required for the operation of the device 600 can also be stored. The computing unit 601, the ROM 602, and the RAM 603 are connected to each other via a bus 604. An input / output (I / O) interface 605 is also connected to the bus 604.

[0119] A number of components in the device 600 are connected to the I / O interface 605, including: an input unit 606, such as a keyboard, a mouse, etc.; an output unit 607, such as various types of displays, speakers, etc.; a storage unit 608, such as a disk, an optical disk, etc.; and a communication unit 609, such as a network card, a modem, a wireless communication transceiver, etc. The communication unit 609 allows the device 600 to exchange information / data with other devices through a computer network such as the Internet and / or various telecommunication networks.

[0120] The computing unit 601 may be a variety of general and / or special processing components with processing and computing capabilities. Some examples of the computing unit 601 include, but are not limited to, a central processing unit (CPU), a graphics processing unit (GPU), various dedicated artificial intelligence (AI) computing chips, various computing units running machine learning model algorithms, digital signal processors (DSPs), and any appropriate processors, controllers, microcontrollers, etc. The computing unit 601 performs the various methods and processes described above, such as the abnormal device determination method. For example, in some embodiments, the abnormal device determination method may be implemented as a computer software program, which is tangibly contained in a machine-readable medium, such as a storage unit 608. In some embodiments, part or all of the computer program may be loaded and / or installed on the device 600 via the ROM 602 and / or the communication unit 609. When the computer program is loaded into the RAM 603 and executed by the computing unit 601, one or more steps of the abnormal device determination method described above may be performed. Alternatively, in other embodiments, the computing unit 601 may be configured to perform the abnormal device determination method in any other appropriate manner (e.g., by means of firmware).

[0121] Various implementations of the systems and techniques described above herein can be implemented in digital electronic circuit systems, integrated circuit systems, field programmable gate arrays (FPGAs), application specific integrated circuits (ASICs), application specific standard products (ASSPs), systems on chips (SOCs), complex programmable logic devices (CPLDs), computer hardware, firmware, software, and / or combinations thereof. These various implementations can include: being implemented in one or more computer programs that can be executed and / or interpreted on a programmable system including at least one programmable processor, which can be a special purpose or general purpose programmable processor that can receive data and instructions from a storage system, at least one input device, and at least one output device, and transmit data and instructions to the storage system, the at least one input device, and the at least one output device.

[0122] The program code for implementing the method of the present disclosure may be written in any combination of one or more programming languages. These program codes may be provided to a processor or controller of a general-purpose computer, a special-purpose computer, or other programmable data processing device, so that the program code, when executed by the processor or controller, enables the functions / operations specified in the flow chart and / or block diagram to be implemented. The program code may be executed entirely on the machine, partially on the machine, partially on the machine and partially on a remote machine as a stand-alone software package, or entirely on a remote machine or server.

[0123] In the context of the present disclosure, a machine-readable medium may be a tangible medium that may contain or store a program for use by or in conjunction with an instruction execution system, device, or equipment. A machine-readable medium may be a machine-readable signal medium or a machine-readable storage medium. A machine-readable medium may include, but is not limited to, an electronic, magnetic, optical, electromagnetic, infrared, or semiconductor system, device, or equipment, or any suitable combination of the foregoing. A more specific example of a machine-readable storage medium may include an electrical connection based on one or more lines, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing.

[0124] To provide interaction with a user, the systems and techniques described herein can be implemented on a computer having: a display device (e.g., a CRT (cathode ray tube) or LCD (liquid crystal display) monitor) for displaying information to the user; and a keyboard and pointing device (e.g., a mouse or trackball) through which the user can provide input to the computer. Other types of devices can also be used to provide interaction with the user; for example, the feedback provided to the user can be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user can be received in any form (including acoustic input, voice input, or tactile input).

[0125] The systems and techniques described herein may be implemented in a computing system that includes back-end components (e.g., as a data server), or a computing system that includes middleware components (e.g., an application server), or a computing system that includes front-end components (e.g., a user computer with a graphical user interface or a web browser through which a user can interact with implementations of the systems and techniques described herein), or a computing system that includes any combination of such back-end components, middleware components, or front-end components. The components of the system may be interconnected by any form or medium of digital data communication (e.g., a communication network). Examples of communication networks include: a local area network (LAN), a wide area network (WAN), and the Internet.

[0126] A computer system may include a client and a server. The client and the server are generally remote from each other and typically interact through a communication network. The relationship of client and server arises through computer programs running on respective computers and having a client-server relationship to each other. The server may be a cloud server, a server in a distributed system, or a server combined with a blockchain.

[0127] It should be understood that the various forms of processes shown above can be used to reorder, add or delete steps. For example, the steps recorded in this disclosure can be executed in parallel, sequentially or in different orders, as long as the desired results of the technical solutions disclosed in this disclosure can be achieved, and this document does not limit this.

[0128] The above specific implementations do not constitute a limitation on the protection scope of the present disclosure. It should be understood by those skilled in the art that various modifications, combinations, sub-combinations and substitutions can be made according to design requirements and other factors. Any modification, equivalent substitution and improvement made within the spirit and principle of the present disclosure shall be included in the protection scope of the present disclosure.

Claims

1. A method for determining an abnormal device, comprising: In response to a communication duration of the distributed collective communication system being greater than a predetermined threshold, obtaining operation logs of a plurality of communication units used to construct the distributed collective communication system; The distributed collective communication system is used to perform a data transmission task of a deep learning model, and the operation log includes the identifiers of the multiple communication units and the data transmission status of the multiple communication units; The identifier is used to identify the type of the data transmission task; Determine a target communication unit from the plurality of communication units according to the identifiers of the plurality of communication units and the data transmission states of the plurality of communication units, wherein the data transmission state of the target communication unit is different from the data transmission states of other communication units for performing the same data transmission task; as well as According to the identification and the predetermined deployment quantity of the target communication unit, the abnormal device corresponding to the target communication unit is determined, wherein the predetermined deployment quantity indicates the number of communication units deployed by each device in the distributed collective communication system.

2. The method according to claim 1, wherein: The step of determining a target communication unit from the plurality of communication units according to the identifiers of the plurality of communication units and the data transmission status in the plurality of communication units comprises: Dividing the multiple communication units according to the identifiers of the multiple communication units to obtain multiple communication groups; For a plurality of communication units in each communication group, the target communication unit is determined from the plurality of communication units according to data transmission states of the plurality of communication units.

3. The method according to claim 2, wherein: The data transmission status includes a data transmission task identifier; For each of the plurality of communication units in the communication group, determining the target communication unit from the plurality of communication units according to data transmission states of the plurality of communication units includes: Analyzing the data transmission task identifiers of the multiple communication units in each communication group to obtain a first analysis result; The target communication unit is determined from the plurality of communication units according to the first analysis result.

4. The method according to claim 3, wherein: The communication group includes I communication units, where I is an integer greater than 1, and the data transmission task identifiers of the multiple communication units in each communication group are analyzed to obtain a first analysis result, including: Compare the data transmission task identifier of the i-th communication unit with the data transmission task identifier of the i+1-th communication unit, where i is an integer greater than or equal to 1 and less than or equal to 1; and In response to the data transmission task identifier of the i-th communication unit being greater than the data transmission task identifier of the (i+1)-th communication unit, the first analysis result is generated.

5. The method according to claim 4, wherein: The step of determining the target communication unit according to the first analysis result includes: According to the first analysis result, the (i+1)th communication unit is determined to be the target communication unit.

6. The method according to claim 3, wherein: The data transmission status also includes: a transmission progress corresponding to the data transmission task, and also includes: For the multiple communication units in each communication group, in response to the data transmission task identifiers of the multiple communication units being the same, analyzing the transmission progress corresponding to the data transmission task of the multiple communication units to obtain a second analysis result; and The target communication unit is determined from the plurality of communication units according to the second analysis result.

7. The method according to claim 6, wherein: The communication group includes I communication units, where I is an integer greater than 1, and the transmission progress corresponding to the data transmission task of the multiple communication units is analyzed to obtain a second analysis result, including: Compare the transmission progress of the i-th communication unit with the transmission progress of the i+1-th communication unit, where i is an integer greater than or equal to 1 and less than or equal to 1; and In response to the transmission progress of the i-th communication unit being greater than the transmission progress of the (i+1)-th communication unit, a second analysis result is generated.

8. The method according to claim 7, wherein: The step of determining the target communication unit according to the second analysis result includes: According to the second analysis result, the (i+1)th communication unit is determined to be the target communication unit.

9. The method according to claim 1, further comprising: According to the data transmission type of the distributed collective communication system, a data transmission characteristic is obtained; as well as A target communication unit is determined from among the plurality of communication units according to the data transmission states of the plurality of communication units and the data transmission characteristics.

10. The method according to claim 9, wherein: The plurality of communication units include S, where S is an integer greater than 1, and determining a target communication unit from the plurality of communication units according to the data transmission states of the plurality of communication units and the data transmission characteristics includes: Compare the data transmission status of the s-th communication unit with the data transmission status of the previous s-1 communication units, where s is an integer greater than or equal to 1 and less than or equal to S; In response to the data transmission characteristic being symmetric transmission, and the data transmission state of the sth communication unit being different from the data transmission state of the first s-1 communication units, a target communication unit is determined from the multiple communication units.

11. The method according to claim 10, wherein: In response to the data transmission characteristic being symmetric transmission, and the data transmission state of the sth communication unit being different from the data transmission state of the first s-1 communication units, determining a target communication unit from the plurality of communication units comprises: In response to the data transmission feature being symmetric transmission and the data transmission state of the sth communication unit being different from the data transmission state of the first s-1 communication units, the sth communication unit is determined as the target communication unit from among the multiple communication units.

12. The method according to claim 9, wherein: The plurality of communication units include T, where T is an integer greater than 1, and determining a target communication unit from the plurality of communication units according to the data transmission states of the plurality of communication units and the data transmission characteristics includes: comparing a first difference between a data transmission state of the t-th communication unit and a data transmission state of the t-1-th communication unit with a second difference between a data transmission state of the t-th communication unit and a data transmission state of the t+1-th communication unit, where t is an integer greater than or equal to 1 and less than or equal to T; and In response to the data transmission characteristic being asymmetric transmission and the first difference and the second difference satisfying a preset condition, a target communication unit is determined from the plurality of communication units.

13. The method according to claim 12, wherein: In response to the data transmission characteristic being asymmetric transmission, and the first difference and the second difference satisfying a preset condition, determining a target communication unit from the plurality of communication units comprises: In response to the data transmission feature being asymmetric transmission, and the first difference and the second difference satisfying a preset condition, determining the t+1th communication unit from the plurality of communication units as the target communication unit.

14. The method according to claim 12, wherein: The preset condition includes: the first difference is not equal to the second difference.

15. The method according to claim 1, wherein: The determining, according to the identifier of the target communication unit and the predetermined deployment quantity, the abnormal device corresponding to the target communication unit comprises: Obtaining an identification of the abnormal device according to the identification of the target communication unit and the predetermined deployment quantity; and The abnormal device is determined according to the identifier of the abnormal device.

16. An abnormal device determination device, comprising: an acquisition module, configured to acquire operation logs of a plurality of communication units used to construct the distributed collective communication system in response to a communication duration of the distributed collective communication system being greater than a predetermined threshold; The distributed collective communication system is used to perform a data transmission task of a deep learning model, and the operation log includes the identifiers of the multiple communication units and the data transmission status of the multiple communication units; The identifier is used to identify the type of the data transmission task; A first determining module, configured to determine a target communication unit from the plurality of communication units according to the identifiers of the plurality of communication units and the data transmission states of the plurality of communication units, wherein the data transmission state of the target communication unit is different from the data transmission states of other communication units used to perform the same data transmission task; as well as The second determination module is used to determine the abnormal device corresponding to the target communication unit according to the identification and the predetermined deployment quantity of the target communication unit, wherein the predetermined deployment quantity indicates the number of communication units deployed by each device in the distributed collective communication system.

17. The device according to claim 16, wherein: The first determining module comprises: A division submodule, used for dividing the multiple communication units according to the identifiers of the multiple communication units to obtain multiple communication groups; The first determination submodule is used to determine the target communication unit from the multiple communication units in each communication group according to the data transmission status of the multiple communication units.

18. The device according to claim 17, wherein: The data transmission status includes a data transmission task identifier; The first determining submodule includes: A first analysis unit, configured to analyze the data transmission task identifiers of the plurality of communication units in each communication group to obtain a first analysis result; A first determining unit is used to determine the target communication unit from the multiple communication units according to the first analysis result.

19. The device according to claim 18, wherein: The communication group includes I communication units, where I is an integer greater than 1, and the first analysis unit includes: A first comparison subunit is used to compare the data transmission task identifier of the i-th communication unit with the data transmission task identifier of the i+1-th communication unit, where i is an integer greater than or equal to 1 and less than or equal to 1; and The first generating subunit is configured to generate the first analysis result in response to the data transmission task identifier of the i-th communication unit being greater than the data transmission task identifier of the (i+1)-th communication unit.

20. The device according to claim 19, wherein The first determining unit includes: The first determining subunit is used to determine the (i+1)th communication unit as the target communication unit according to the first analysis result.

21. The device according to claim 18, wherein The data transmission status also includes: a transmission progress corresponding to the data transmission task, and the first determination submodule includes: a second analyzing unit, configured to analyze, for each of the plurality of communication units in the communication group, transmission progress corresponding to the data transmission task of the plurality of communication units in response to the plurality of communication units having the same data transmission task identifier, to obtain a second analysis result; and A second determining unit is used to determine the target communication unit from the multiple communication units according to the second analysis result.

22. The device according to claim 21, wherein The communication group includes I communication units, where I is an integer greater than 1, and the second analysis unit includes: a second comparing subunit, configured to compare the transmission progress of the i-th communication unit with the transmission progress of the i+1-th communication unit, where i is an integer greater than or equal to 1 and less than or equal to 1; and The second generating subunit is used to generate a second analysis result in response to the transmission progress of the i-th communication unit being greater than the transmission progress of the (i+1)-th communication unit.

23. The device according to claim 22, wherein: The second determining unit includes: The second determining subunit is used to determine the (i+1)th communication unit as the target communication unit according to the second analysis result.

24. The apparatus of claim 16, further comprising: An acquisition module, used to obtain data transmission characteristics according to the data transmission type of the distributed collective communication system; as well as A third determining module is configured to determine a target communication unit from among the plurality of communication units according to the data transmission states of the plurality of communication units and the data transmission characteristics.

25. The device according to claim 24, wherein: The plurality of communication units include S, where S is an integer greater than 1, and the third determining module includes: A first comparison submodule is used to compare the data transmission status of the sth communication unit with the data transmission status of the previous s-1 communication units, where s is an integer greater than or equal to 1 and less than or equal to S; The second determination submodule is used to determine a target communication unit from the multiple communication units in response to the data transmission feature being symmetric transmission and the data transmission state of the sth communication unit being different from the data transmission state of the first s-1 communication units.

26. The device according to claim 25, wherein The second determining submodule includes: The third determination unit is used to determine the sth communication unit as the target communication unit from the multiple communication units in response to the data transmission characteristic being symmetric transmission and the data transmission state of the sth communication unit being different from the data transmission state of the first s-1 communication units.

27. The device according to claim 24, wherein: The plurality of communication units include T, where T is an integer greater than 1, and the third determining module includes: a second comparison submodule, configured to compare a first difference between a data transmission state of a t-th communication unit and a data transmission state of a t-1-th communication unit and a second difference between a data transmission state of the t-th communication unit and a data transmission state of the t+1-th communication unit, where t is an integer greater than or equal to 1 and less than or equal to T; and The third determination submodule is configured to determine a target communication unit from the plurality of communication units in response to the data transmission characteristic being asymmetric transmission and the first difference and the second difference satisfying a preset condition.

28. The device according to claim 27, wherein The third determining submodule comprises: A fourth determining unit is configured to determine, in response to the data transmission characteristic being asymmetric transmission and the first difference and the second difference satisfying a preset condition, the t+1th communication unit from the multiple communication units as the target communication unit.

29. The device according to claim 27, wherein: The preset condition includes: the first difference is not equal to the second difference.

30. The device according to claim 16, wherein: The second determining module comprises: an obtaining submodule, configured to obtain an identification of the abnormal device according to an identification of the target communication unit and the predetermined deployment quantity; and The fourth determining submodule is used to determine the abnormal device according to the identifier of the abnormal device.

31. An electronic device comprising: at least one processor; as well as a memory communicatively connected to the at least one processor; wherein, The memory stores instructions that can be executed by the at least one processor, and the instructions are executed by the at least one processor to enable the at least one processor to perform the method according to any one of claims 1 to 15.

32. A non-transitory computer-readable storage medium storing computer instructions, wherein: The computer instructions are used to cause the computer to execute the method according to any one of claims 1-15.

33. A computer program product comprising a computer program, which, when executed by a processor, implements the method according to any one of claims 1 to 15.

Images