Detailed Description
Exemplary embodiments of the present invention will now be described with reference to the accompanying drawings, in which various details of the embodiments of the present invention are included to facilitate understanding, and are to be considered merely exemplary. Accordingly, one of ordinary skill in the art will recognize that various changes and modifications of the embodiments described herein can be made without departing from the scope of the invention. Also, descriptions of well-known functions and constructions are omitted in the following description for clarity and conciseness.
The terminology used in the description of the various illustrated examples in this disclosure is for the purpose of describing particular examples only and is not intended to be limiting. Unless the context clearly indicates otherwise, the elements may be one or more if the number of the elements is not specifically limited. Furthermore, the term "and/or" as used in the present invention encompasses any and all possible combinations of the listed items.
Fig. 1 shows a schematic diagram of an application scenario of a method provided according to an embodiment of the present invention. The application scenario includes one or more eSIM devices 101, a cloud platform 120, and one or more communication networks 110 coupling the one or more eSIM devices 101 to the cloud platform 120. In some embodiments, the cloud platform 120 can also provide eSIM configuration services or software applications. In some embodiments, these services can be provided as web-based services or cloud services, for example, provided to users of eSIM devices 101 under a software as a service (SaaS) model.
In the configuration shown in fig. 1, cloud platform 120 may include one or more components that implement the functions performed by cloud platform 120. These components may include software components, hardware components, or a combination thereof that are executable by one or more processors. A user operating eSIM device 101 can in turn utilize one or more applications to interact with cloud platform 120 to utilize the services provided by these components. It should be appreciated that a variety of different system configurations are possible, which may differ from the application scenario. Accordingly, FIG. 1 is one example of a system for implementing the various methods described herein and is not intended to be limiting.
The eSIM device 101 can include various types of terminal devices, such as portable handheld devices, wearable devices, various messaging devices, tablet computers, and the like. Portable handheld devices may include cellular telephones, smart phones, tablet computers, personal Digital Assistants (PDAs), and the like. Wearable devices may include head mounted displays (such as smart glasses) and other devices.
Network 110 may be any type of network known to those skilled in the art that may support data communications using any of a number of available protocols, including but not limited to TCP/IP, SNA, IPX, etc. For example only, the one or more networks 110 may be a Local Area Network (LAN), an ethernet-based network, a token ring, a Wide Area Network (WAN), the internet, a virtual network, a Virtual Private Network (VPN), an intranet, an extranet, a blockchain network, a Public Switched Telephone Network (PSTN), an infrared network, a wireless network (e.g., bluetooth, WIFI), and/or any combination of these and/or other networks.
Cloud platform 120 may include one or more general-purpose computers, special-purpose server computers (e.g., PC (personal computer) servers, UNIX servers, middleend servers), blade servers, mainframe computers, server clusters, or any other suitable arrangement and/or combination. Cloud platform 120 may include one or more virtual machines running a virtual operating system, or other computing architecture that involves virtualization (e.g., one or more flexible pools of logical storage devices that may be virtualized to maintain virtual storage devices of servers). In various embodiments, cloud platform 120 may run one or more services or software applications that provide the functionality described below.
The computing units in cloud platform 120 may run one or more operating systems including any of the operating systems described above as well as any commercially available server operating systems. Cloud platform 120 may also run any of a variety of additional server applications and/or middle tier applications, including HTTP servers, FTP servers, CGI servers, JAVA servers, database servers, etc.
In some implementations, the cloud platform 120 can include one or more applications to analyze and consolidate data feeds and/or event updates received from users of eSIM devices 101. The cloud platform 120 can also include one or more applications to display data feeds and/or real-time events via one or more display devices of the eSIM device 101.
In some implementations, the cloud platform 120 may be a server of a distributed system, or a server that incorporates a blockchain. Cloud platform 120 may also be a cloud server, or an intelligent cloud computing server or intelligent cloud host with artificial intelligence technology. The cloud server is a host product in a cloud computing service system, so as to solve the defects of large management difficulty and weak service expansibility in the traditional physical host and Virtual special server (VPS PRIVATE SERVER) service.
The application scenario may also include one or more databases 130. In some embodiments, these databases may be used to store data and other information. For example, one or more of the databases 130 may be used to store historical data. Database 130 may reside in various locations. For example, a database used by cloud platform 120 may be local to cloud platform 120, or may be remote from cloud platform 120 and may communicate with cloud platform 120 via a network-based or dedicated connection. Database 130 may be of different types. In some embodiments, the database used by cloud platform 120 may be, for example, a relational database. One or more of these databases may store, update, and retrieve the databases and data from the databases in response to the commands.
Referring to fig. 2, the method for remote eSIM configuration management based on a cloud platform provided by the embodiment of the invention includes the following steps:
Step S100, acquiring a real-time configuration request of target eSIM equipment, and extracting an environment feature set and equipment identification features in the real-time configuration request, wherein the environment feature set comprises a network access mode, geographic position coordinates and signal intensity distribution parameters.
Illustratively, the real-time configuration request refers to a request instruction which is actively initiated by the target eSIM device to the cloud platform through the wireless communication network and requires to acquire specific configuration parameters, and the triggering conditions include, but are not limited to, first activation of the device, change of network environment, or manual triggering operation of a user. The environment characteristic set is a multidimensional parameter set used for describing the current operation environment of the target eSIM equipment, wherein the network access mode specifically refers to the network type and communication protocol combination of the current connection of the equipment, such as coexistence of a 4G cellular network and Wi-Fi dual mode, independent networking of 5G or single-mode connection of a narrowband Internet of things, geographic position coordinates are obtained through a global positioning system or a base station triangulation technology and used for accurately identifying longitude and latitude information of the equipment, the signal intensity distribution parameters comprise signal intensity values of each frequency band detected by the equipment in a preset time period and fluctuation characteristics thereof, such as that the signal intensity of the 2.4GHz frequency band has a mean value of-65 dBm and standard deviation of 8.2, and the signal intensity of the 5GHz frequency band shows a periodical attenuation trend. The device identification feature is a feature data combination that uniquely identifies the target eSIM device, and includes non-tamperable physical features such as an international mobile equipment identification code, an integrated circuit card identifier, and a hardware version number. By analyzing the data packet structure of the real-time configuration request, the cloud platform adopts a feature extraction algorithm to separate an environment feature set and equipment identification features from the request header field and the load content, so that the follow-up strategy generation process is ensured to have accurate input conditions.
Step 200, matching a corresponding dynamic strategy model from a pre-constructed dynamic strategy model library according to the equipment identification characteristics, wherein the dynamic strategy model is generated through historical configuration data training, and each dynamic strategy model is associated with at least one group of configuration strategy generation rules.
Illustratively, the dynamic policy model library is a set of machine learning models stored in a cloud platform distributed database, and the construction process is based on the mapping relationship between the device identification features, the environmental features and the finally effective configuration parameter set recorded in the historical configuration data, for example, a time sequence neural network algorithm can be used for training and optimizing. Each dynamic policy model corresponds to a configuration policy generation capability of a particular device type or hardware version, such as an interference-free configuration model for an industrial-level eSIM device or a low-power configuration model for a consumer-level device. The configuration policy generation rule is a set of decision logic embedded in a dynamic policy model, and comprises a network access mode priority judgment rule, a signal strength threshold triggering condition and a geographic position area division policy. When the hardware version number in the equipment identification feature is matched with a version code preset in a model library, the cloud platform calls a hash index mechanism to quickly locate an associated dynamic strategy model, and meanwhile loads a configuration strategy generation rule set bound by the model to provide a rule base for subsequent strategy sequence generation. For example, when the device identification feature indicates that the target eSIM device is a model of a vehicle terminal, the dynamic policy model will preferentially match the dedicated model supporting the multi-network fast handoff and high signal fault tolerance mechanisms.
As an embodiment, the generation process of the dynamic policy model library may include the steps of:
Step S201 is to acquire a configuration operation data set of a plurality of historical eSIM devices, where the configuration operation data set includes an environmental feature record, a configuration parameter execution record and a configuration result evaluation index of each historical eSIM device.
The configuration operation data set is a structured set of historical eSIM device full life cycle operation data collected through a cloud platform log system, and a data source of the structured set covers a device activation request, a configuration instruction issuing record and a network performance monitoring result. The environmental characteristic record refers to a time sequence data set of a network access mode, a geographic position coordinate and signal intensity distribution parameters reported by equipment in a historical time period, for example, 4G/5G dual-mode switching record, longitude and latitude tracks and mean square values of signal intensities of different frequency bands recorded by a vehicle-mounted terminal of a certain model in 30 continuous days. The configuration parameter execution record includes network access configuration, frequency band locking strategy and power consumption control parameter combination which are actually effective by the device, such as 5G NSA mode, 3.5GHz main frequency band and dynamic power adjustment threshold which are started in a certain configuration instruction. The configuration result evaluation index is an effect quantification result of each configuration operation through a preset algorithm, and comprises a configuration success rate, a signal stability gain and an energy consumption change rate, for example, the configuration operation at a certain time enables the signal switching failure rate of the equipment in a highway scene to be reduced by 12%, and the average power consumption to be increased by 8%.
Step S202, extracting the equipment identification characteristics of each historical eSIM equipment from the configuration operation data set, and grouping the configuration operation data set based on the equipment identification characteristics to obtain sub-data sets corresponding to a plurality of equipment characteristic groups.
The device identification features are illustratively an immutable set of attributes extracted from hardware information and registration information of the historical eSIM device, including an international mobile equipment identification code, an embedded SIM card serial number, and a firmware version number. For example, the device identification features of a batch of industrial grade eSIM devices include a uniform hardware version number "HW-5G-IND-01" and a pre-assigned SIM card number segment "89860121". Based on the hardware version number in the device identification feature and the SIM card number segment prefix, the configuration operation data set is divided into a plurality of device feature groups, for example, all device history data with the hardware version number of "HW-5G-IND-01" and the SIM card number segment of "89860121" are categorized into the same sub-data set. In the process, a distributed hash algorithm is adopted to quickly group massive historical data, so that the sub-data set corresponding to each equipment characteristic group only contains configuration operation records of the same type of equipment, and data consistency guarantee is provided for subsequent rule extraction.
The following processes may be specifically performed for the sub-data set corresponding to each device feature group:
and step 203, constructing an initial strategy generation rule set according to the environmental characteristic records and the configuration parameter execution records in the sub-data sets, wherein the initial strategy generation rule set comprises a plurality of conditional branches, and each conditional branch corresponds to a group of environmental characteristic threshold ranges and an associated configuration parameter set.
The initial policy generation rule set is illustratively a set of decision logic extracted from the mapping relationship of historical environmental features and configuration parameters by an association rule mining algorithm, for example. For example, for a sub-dataset of a certain device feature set, analysis found that when the 3.5GHz band intensity mean of the signal intensity distribution parameters is higher than-70 dBm and the geographic location coordinates are in the urban central region, 95% of the historical configuration operations selected the 5G NSA mode and 256QAM modulation scheme. Based on the association relation, an initial rule that the conditional branch is that the IF 3.5GHz signal intensity average value is more than or equal to-70 dBm AND the geographic position is in the urban central area THEN AND the 5G NSA mode AND 256QAM modulation are started is constructed. The threshold range of the environmental characteristic of each conditional branch is determined by a clustering algorithm, for example, after K-means clustering is carried out on the signal intensity distribution parameters, the range from-70 dBm to-50 dBm is divided into high signal intensity intervals, and the range from-90 dBm to-70 dBm is divided into medium signal intensity intervals. The associated configuration parameter set selects the effective combination with the highest frequency of occurrence from the history configuration parameter execution record, for example, the 4G CA carrier aggregation and 2.4GHz Wi-Fi dual-connection configuration with the highest frequency of occurrence under the expressway scene.
And S204, optimizing the initial strategy generation rule set by using the configuration result evaluation indexes in the sub-data set to generate a dynamic strategy model corresponding to the equipment feature set, wherein the optimization process comprises merging redundant conditional branches, adjusting the environment feature threshold range and updating the priority of the configuration parameter set.
In an exemplary embodiment, the optimization process may iteratively correct the initial rule based on the configuration success rate, the signal stability gain, and the energy consumption rate of change in the configuration result evaluation index. For example, a conditional branch specifies that "IF network access mode is 4G single mode AND signal strength standard deviation is 10dB th enable CA carrier aggregation", but its configuration success rate is only 65% AND the rate of change of energy consumption rises by 15%, indicating that there is redundancy in the rule or that the threshold setting is unreasonable. By combining adjacent conditional branches covering the same environmental characteristic range and adjusting the standard deviation threshold of the signal strength to 8dB, the optimized rule configuration success rate is increased to 82%, and the energy consumption change rate is reduced to 5%. The priority update is dynamically ordered according to the historical effect of the configuration parameter sets, for example, the priority of the configuration parameter sets with the signal stability gain higher than 20% is adjusted upwards, and the configuration parameter sets with the energy consumption change rate exceeding the preset threshold are degraded to be alternative schemes. And the finally generated dynamic strategy model codes the optimized strategy generation rule set into executable decision logic and is bound with the equipment characteristic group for storage.
Step S205, storing the optimized dynamic strategy model into the dynamic strategy model library, and establishing a mapping relation between the dynamic strategy model and the equipment identification characteristics of the corresponding equipment characteristic group.
Illustratively, the dynamic policy model library may employ a graph database to store topological relationships between models, wherein each dynamic policy model is indexed by a hardware version number in the device identification feature and a SIM card number segment. For example, a dynamic policy model with a hardware version number "HW-5G-IND-01" associated with a SIM card number segment "89860121" is stored as a graph node and its applicable geographic location area and network access mode constraints are recorded by edge attributes. The mapping relation establishment process adopts a bidirectional hash table to realize quick retrieval, and when the new equipment identification characteristic is matched with the existing equipment characteristic group, the cloud platform can directly call the associated dynamic strategy model to make a real-time decision. In addition, the model version control mechanism ensures that the dynamic strategy model after each optimization generates a unique version identifier, thereby facilitating the subsequent tracing and rollback operation.
In one embodiment, the step S204 optimizes the initial policy generation rule set by using the configuration result evaluation index in the sub-data set, including:
and step S2041, extracting configuration success rate, signal stability gain and energy consumption change rate from the configuration result evaluation indexes of the sub-data sets.
For example, the configuration success rate refers to the proportion of the history configuration parameter set that is successfully validated within the corresponding environmental characteristic threshold, for example, 108 times of configuration operations triggered by a certain rule successfully complete APN authentication and frequency band locking, and the configuration success rate is 90%. The signal stability gain is calculated by comparing the standard deviation of the signal intensity before and after configuration, for example, the standard deviation of the signal intensity of the device is reduced from 15dB to 8dB in 24 continuous hours after configuration, and the gain is 46.7%. The energy consumption change rate is quantified by adopting the average power consumption ratio before and after configuration, for example, the standby current of the equipment is increased from 12mA to 13mA after configuration, and the change rate is 8.3%.
For example, for each conditional branch in the initial policy generation rule set, the following is performed:
step S2042, calculating the configuration success rate of the history configuration parameter set corresponding to the conditional branch and the weighted score of the signal stability gain.
Illustratively, the weighted score is determined in such a way that the score=0.6×configuration success rate+0.4×signal stability gain, wherein the weight coefficients (0.6 and 0.4) are adaptively adjusted according to the device type, e.g., the industrial level device weights 0.5 and 0.5 for signal stability. If the allocation success rate of a conditional branch is 85% and the signal stability gain is 30%, the weighting score is 0.6x85+0.4x30=63 minutes.
And step S2043, if the weighted score is lower than a preset score threshold, deleting the conditional branch, and merging the environmental characteristic threshold range covered by the conditional branch into the adjacent conditional branch.
Illustratively, the preset scoring threshold is set based on a performance baseline of the device feature set, e.g., a consumer grade device threshold of 60 points and an industrial grade device threshold of 70 points. When a conditional branch score is 55, the signal strength range covered by the conditional branch score is [ -80dBm, -70dBm ] to be combined into adjacent [ -70dBm, -60dBm ] intervals, and the combined rule score is recalculated.
And step S2044, if the weighted score is higher than or equal to a preset score threshold value but the energy consumption change rate exceeds the preset energy consumption threshold value, adjusting the power consumption control parameters in the configuration parameter set associated with the conditional branch, and recalculating the configuration result evaluation index.
The preset energy consumption threshold is set according to the power supply capacity of the device, for example, the vehicle-mounted terminal is 10%, and the wearable device is 5%. If the score of a certain rule is 75 minutes and the energy consumption change rate is 12 percent, the transmitting power in the configuration parameter set is reduced from 23dBm to 20dBm, and the updated energy consumption change rate is 7 percent by recapturing the historical data.
Step S2045, generating an updated strategy generation rule set according to the adjusted conditional branches, and inputting the updated strategy generation rule set into the strategy decision tree model for rule conflict detection.
Illustratively, rule conflict detection identifies whether there are multiple paths pointing to the same environmental feature range but associating different sets of configuration parameters by traversing all paths of the policy decision tree model. For example, when two regulations simultaneously specify that the IF network access mode is 5G AND the geographic position is the city center THEN enabled frequency band a "AND the IF signal intensity is more than or equal to-65 dBm AND the geographic position is the city center THEN enabled frequency band B", a conflict alarm is triggered in the overlapping area of the coverage of the frequency band a AND the frequency band B.
And step S2046, if the rule conflict is detected, reordering the priority of the conflicting conditional branches until no conflict path exists in the strategy decision tree model.
For example, the prioritization may be based on a weighted score of the set of configuration parameters and device type preferences, e.g., a set of configuration parameters with high signal stability gain is preferentially selected in an industrial-scale device, and a set of configuration parameters with low rate of energy consumption change is preferentially selected in a consumer-scale device. After reordering, the strategy decision tree model only reserves the highest priority path in the same environment characteristic range, and ensures the uniqueness of decision logic.
As one embodiment, the construction process of the policy decision tree model includes the following steps:
Step S20461 is to extract all conditional branch environmental feature threshold ranges and associated configuration parameter sets from the updated policy generation rule set.
Illustratively, the environmental characteristic threshold range is stored in interval coded form, e.g., the network access mode is {4G,5G }, the geographic location coordinates are longitude [116.3 °,116.5 ° ] and latitude [39.9 °,40.1 ° ], the signal strength distribution parameters are mean [ -70dBm, -60dBm ] and the standard deviation is less than or equal to 8dB. The associated set of configuration parameters is then converted to a binary instruction template, e.g. "0x1A3B" indicating that 5G NSA mode, 3.5GHz band and dynamic power control are enabled.
And step S20462, constructing a multi-layer decision node by taking the network access mode as a root node according to the sequence of the geographic position coordinates and the signal intensity distribution parameters.
Illustratively, the root node first divides the network access mode in which the device is located, e.g., using 4G, 5G, and hybrid modes as three main branches. Each main branch is followed by dividing the child nodes by latitude and longitude ranges of the geographic location coordinates, e.g., creating "city center", "suburban", and "remote area" child nodes under the 5G branch. The final node is subdivided according to the mean value and standard deviation of signal intensity distribution parameters, for example, paths of signal intensity mean value of more than or equal to-65 dBm and signal intensity mean value of less than or equal to-65 dBm are created under the subnode of the city center.
And step 20463, setting a corresponding environmental characteristic threshold segmentation point at each decision node, and directing the path meeting the segmentation point condition to the decision node or leaf node of the next layer.
Illustratively, the threshold split point may be determined using a dichotomy or an equal frequency bin method, for example, dividing the signal strength mean into intervals of [ -90dBm, -85 dBm), [ -85dBm, -80 dBm), etc., at 5dB intervals. When the environmental characteristics of the equipment match a certain segmentation point condition, the decision path is transmitted downwards along the corresponding branch until the leaf node is reached.
Step S20464, associating corresponding configuration parameter sets at the leaf nodes, and setting the execution sequence of the leaf nodes based on the priorities of the configuration parameter sets.
For example, two parameter sets of "5G NSA high power configuration" and "5G SA power configuration" are associated at leaf nodes with signal strength mean ∈65dBm, and the former is set as the default execution option according to its weighted score. If the default configuration fails to execute, the subsequent parameter sets are tried in order of priority.
And step 20465, pruning the constructed initial strategy decision tree, deleting redundant paths which do not cover any history environmental characteristic records, and taking the pruned strategy decision tree as a core decision structure of the dynamic strategy model.
Illustratively, the pruning algorithm may traverse all decision paths, removing branches not covered by the historical data instance, e.g., deleting branches with "signal strength mean ≡-50dBm under 4G networks," as this condition never occurs in the historical data. The strategy decision tree after pruning obviously improves the execution efficiency and the decision accuracy of the model by removing the path with low generalization capability.
As an implementation manner, the step S200, matching, according to the device identification feature, a corresponding dynamic policy model from a pre-constructed dynamic policy model library, includes:
and S210, decomposing the equipment identification feature into a plurality of sub-feature sets, wherein the sub-feature sets comprise hardware model codes, firmware version sequences and operator code combinations.
Illustratively, the device identification feature is a set of unique identification data extracted from a non-volatile memory of the target eSIM device, the decomposition of which can be accomplished by a feature parsing algorithm. The hardware model code represents a fifth generation industrial grade 5G communication module for a hardware version identifier defined by a device manufacturer, for example, "HW-5G-IND-01", where "IND" identifies an industrial application scenario. The firmware version sequence consists of a compile time stamp and version number in the device firmware upgrade log, e.g. "FW-2.1.3_20231001" represents version 2.1.3 firmware released on day 10 of 2023. The operator codes are combined into an authorized code set of the network access authority of the operator to which the equipment belongs, for example, "CN-MOB-001" represents China mobile eSIM subscription codes, and "CN-TEL-005" represents China telecommunication Internet of things special access codes. Through regular expression matching and field separator recognition, the cloud platform disassembles the original equipment identification characteristic character string into three independent sub-characteristic sets, and provides structural input for subsequent multi-stage matching.
Step S220, traversing first-level matching conditions in a level index table of a dynamic strategy model library according to the hardware model code, screening out all candidate dynamic strategy models containing the hardware model code, and forming a first candidate model list.
Illustratively, the hierarchical index table is a multi-level index table constructed based on a B+ tree structure, wherein a first level node of the multi-level index table is encoded by a hardware model as key values, and each key value is associated with a dynamic strategy model linked list. For example, when the hardware model code is "HW-5G-IND-01," the index table traversal algorithm starts matching from the root node, locates to the leaf node that contains the code, and retrieves the linked list head address that it points to. And storing metadata of all dynamic strategy models supporting the hardware model in a linked list, wherein the metadata comprises a model identifier, an applicable firmware version range and an operator white list. Prefix matching mechanisms can be adopted in the traversal process, so that partial code matching models can be screened, for example, HW-5G-IND-01 and HW-5G-IND-01A share the same prefix branch. Finally, the first candidate MODEL list contains 12 hardware-compatible dynamic policy MODELs, such as the MODELs identified as "MODEL_5G_IND_V3" and "MODEL_5G_IND_LEGACY".
And step S230, carrying out second-stage matching condition filtering on the first candidate model list based on the firmware version sequence, extracting a model of which the firmware adaptation range covers the firmware version sequence from the candidate dynamic strategy model, and generating a second candidate model list.
Illustratively, the firmware adaptation range is a firmware version compatibility interval defined in the dynamic policy model metadata, e.g. "FW-2.0.0 to FW-2.2.0" means that all firmware versions between 2.0.0 to 2.2.0 are supported. In the matching process, the firmware version sequence FW-2.1.3_20231001 of the target device can be converted into a semantic version number of 2.1.3, and interval inclusion detection is carried out with the adaptation range of the candidate model. For example, if the fit range of a model is ". Gtoreq. 2.1.0 and <2.2.0", then the target firmware version meets the condition and the model is retained to the second candidate model list. For an adaptation range defined with a timestamp (e.g., "20230901 to 20231130"), then "20231001" in the firmware version sequence is extracted for time window matching. After the second filtering, 12 MODELs in the first candidate MODEL list are reduced to 5, for example, "MODEL_5G_IND_V3" is reserved for the adaptation range of "FW-2.1.0 to FW-2.3.0", and "MODEL_5G_IND_LEGACY" is eliminated for supporting only "FW-1.4.0 to FW-2.0.0".
And step S240, performing third-level matching condition mapping on the second candidate model list according to the operator code combination, and matching the target dynamic strategy model which completely contains the operator code combination in the operator white list associated with the candidate dynamic strategy model.
Illustratively, the operator whitelist is a predefined set of authorized operator codes in dynamic policy model metadata, e.g. "{ CN-MOB-001, CN-TEL-005}" indicates that the model is applicable only to chinese mobile and chinese telecom dual operator bound devices. The matching algorithm verifies whether the operator code combination of the target device is a subset of the whitelist, e.g., when the device code combination is "{ CN-MOB-001, CN-TEL-005}" and the model whitelist is "{ CN-MOB-001, CN-TEL-005, CN-UNI-007}", it is determined to be fully inclusive. If the device code combination contains a whitelist of unauthorized codes (e.g. "CN-GLB-010"), the model is excluded. In this process, the 5 MODELs in the second candidate MODEL list are further reduced to 2, for example, "MODEL_5G_IND_V3" is kept for the white list containing "CN-MOB-001" and "CN-TEL-005", and "MODEL_5G_IND_HYBRID" is culled for the white list containing only "CN-UNI-007".
Step S250, if a plurality of target dynamic strategy models exist after the third-level matching condition mapping, priority ranking is carried out according to the difference value between the activation time stamp in the equipment identification feature and the update time stamp of the target dynamic strategy model, and the target dynamic strategy model with the smallest difference value is selected as a matching result.
Illustratively, the activation timestamp is a time identification of the first successful access of the device to the cloud platform, e.g., "2023-10-01 08:00:00". The update time stamp of the target dynamic policy MODEL records the time of its last optimization iteration, e.g. the update time stamp of "MODEL_5G_IND_V3" is "2023-09-15:30:00", and the update time stamp of "MODEL_5G_IND_V3B" is "2023-10-05:00:00". The absolute time difference between the activation time stamp and the update time stamp is calculated, for example, the time difference of "MODEL_5G_IND_V3" is 15 days 6 hours, and the time difference of "MODEL_5G_IND_V3B" is 3 days 23 hours. By ascending order, "MODEL_5G_IND_V3B" with the smallest time difference is selected as the final matching MODEL, so that the device is ensured to use the latest optimized policy rules.
Step S260, if the target dynamic strategy model does not exist after the third-level matching condition mapping, the hardware model code and the firmware version sequence are input to a similarity evaluation engine of a dynamic strategy model library, compatibility scores of the hardware model code and the firmware version sequence are calculated, a dynamic strategy model with the compatibility scores exceeding a preset threshold is selected as a substitute matching result, and unmatched operator codes are recorded and combined to a difference feature log for subsequent model incremental training.
Illustratively, the similarity assessment engine may employ a multidimensional feature weighting algorithm, wherein the hardware model code compatibility weights 60% and the firmware version sequence compatibility weights 40%. For example, when the degree of similarity of the target hardware MODEL code "HW-5G-IND-01" and the hardware code "HW-5G-IND-02" of the candidate MODEL "modem_5g_ind_v2" is 85%, and the degree of compatibility of the firmware version sequence "FW-2.1.3" and the MODEL adaptation range "FW-2.0.0 to FW-2.2.0" is 90%, the overall compatibility score is 0.6x85+0.4x90=87 minutes. If the preset threshold is 80 minutes, the model is selected as the alternative matching result. The unmatched operator code combination "CN-GLB-010" is recorded to a difference feature log, which contains the device identification features, the set of environmental features, and the configuration policy sequence for triggering subsequent model incremental training tasks, such as taking the operator code into a whitelist extension during the next model training period.
And step 300, generating a configuration strategy sequence of the target eSIM equipment based on the environment feature set and the matched dynamic strategy model, wherein the configuration strategy sequence comprises a plurality of candidate configuration parameter sets with ordered priorities.
Illustratively, the configuration policy sequence refers to a parameter combination queue dynamically generated according to real-time environmental characteristics and policy generation rules and arranged in descending order of execution priority, and each candidate configuration parameter set contains a complete network access configuration, a frequency band selection policy and a signal strength tolerance threshold. The priority ranking algorithm comprehensively considers network connection stability, data transmission rate and equipment energy consumption indexes, for example, the 5G independent networking configuration is arranged on the 4G/Wi-Fi dual-mode configuration, or the priority of the low-frequency band configuration is automatically improved when the standard deviation of the signal strength exceeds a preset threshold. The dynamic strategy model analyzes the time-frequency characteristics of the signal intensity distribution parameters through convolution operation, and finally generates a strategy sequence containing 3 candidate configuration parameter groups by combining geographical position coordinates with a preset regional network coverage thermodynamic diagram. For example, the first priority parameter set is configured to enable a 5G SA mode and lock a 3.5GHz band with the strongest signal strength, the second priority parameter set is configured to enable 4G CA carrier aggregation and 2.4GHz Wi-Fi dual connectivity, and the third priority parameter set is connected with NB-IoT narrowband internet of things and enables a signal strength adaptive adjustment module. The generation of each parameter set strictly follows the rule constraint preset in the dynamic policy model, for example, the heartbeat packet interval optimization policy is forcefully enabled when the network access mode is cellular single connection.
As an implementation manner, the step S300, based on the environmental feature set and the matched dynamic policy model, generates a configuration policy sequence of the target eSIM device, and may specifically include the following steps:
And step S310, performing conditional branch matching on the network access mode in the environment characteristic set and the strategy generation rule in the dynamic strategy model, and screening out all effective rules of the network access mode threshold range covering the current network access type.
Illustratively, the network access mode threshold range is a predefined set of network types in the dynamic policy model rule base, e.g. "5G NSA (non-independent networking)", "4G CA (carrier aggregation)", and "3G/Wi-Fi dual mode" are taken as independent threshold intervals. When the current network access type of the target eSIM device is '5G NSA', the conditional branch matching algorithm traverses the network access mode field in the strategy generation rule, and screens out the effective rule containing '5G NSA' or '5G NSA and 4G CA mixed mode'. For example, a rule defines a network access mode threshold of "5G NSA or 4G CA" and the signal strength average value is ≡75dBm, then the rule is marked as a valid rule. In this process, invalid rules, such as rules that support only "3G single mode" or "Wi-Fi first" are automatically excluded, ensuring that subsequent processing is focused only on policies compatible with the current network environment.
Step 320, extracting signal strength adapting conditions associated with the geographic position coordinates in the dynamic strategy model based on the screened effective rules, and calculating geographic position coverage overlap ratio of each effective rule according to the signal strength distribution parameters in the environment feature set.
Illustratively, the signal strength adaptation condition associated with the geographic position coordinates refers to a longitude and latitude boundary range and a corresponding signal strength constraint set in a rule, for example, a rule specifies that a signal strength mean value is more than or equal to-70 dBm and a standard deviation is less than or equal to 10dB in a longitude [116.30 degrees, 116.50 degrees ] and latitude [39.90 degrees, 40.10 degrees ] area. The geographic location coverage overlap is determined by calculating the ratio of the current coordinates of the target device to the area of overlap of the geofence in the rule, e.g., 92% of its total coverage area when the device coordinates are located at [116.35 °,39.95 ° ]. Meanwhile, the mean value and the standard deviation in the signal intensity distribution parameters are substituted into the signal intensity constraint condition in the rule for verification, and if the current signal intensity mean value of the equipment is-68 dBm and the standard deviation is 8dB, the rule adaptation condition is judged to be met.
And S330, pre-sequencing the priorities of the effective rules of which the geographic position coverage overlap ratio reaches a preset overlap threshold and the signal strength distribution parameters meet the adaptation conditions, and generating an initial candidate rule queue.
Illustratively, the preset coincidence threshold is dynamically adjusted according to the moving speed of the device, for example, the threshold of the fixed device is 80%, and the threshold of the vehicle-mounted terminal is reduced to 60% to adapt to the moving scene at high speed. When the geographic location coverage of a rule is 85% and the signal strength parameters are fully matched, the rule is added to the initial candidate rule queue. The pre-ordering algorithm performs a descending order based on the signal strength average requirement of the rules, e.g., the rule with higher signal strength threshold is preferentially selected to maximize connection stability. For example, rule A requires a signal strength of-65 dBm or more and rule B requires a signal strength of-70 dBm or more, then rule A orders higher in the queue than rule B, although its geographic location coverage overlap is slightly lower.
And step 340, carrying out weight optimization adjustment on the effective rules in the initial candidate rule queue according to the rule history execution success rate and the energy consumption efficiency index recorded in the dynamic policy model, and redefining the priority order.
Illustratively, rule history execution success rate refers to the number of times the rule successfully completes a configuration operation in the past 100 triggers, e.g., rule a has a success rate of 95% and rule B has a success rate of 88%. The energy consumption efficiency index is calculated by the ratio of the average power consumption to the reference power consumption in the history execution of the configuration parameter set associated with the rule, for example, the power consumption ratio of the rule A is 110%, and the power consumption ratio of the rule B is 98%. The weight optimization adopts a multi-objective weighting algorithm, gives the execution success rate of 60% weight and the energy consumption efficiency of 40% weight, and reorders after comprehensive scoring is calculated. For example, rule a overall score is 0.6x95+0.4× (100-110 x 0.5) =77 points, rule B is 0.6x88+0.4× (100-98 x 0.5) =82.4 points, and therefore rule B priority is adjusted up to the head of the queue.
And step 350, mapping the adjusted effective rule to a corresponding candidate configuration parameter set according to a priority order, and eliminating redundant parameter items which conflict with the device identification characteristics of the target eSIM device in the candidate configuration parameter set.
Illustratively, the candidate configuration parameter set is bound to the rules one by one, e.g., rule B maps to parameter set "5G NSA primary band 3.5ghz+4g CA secondary band 1.8GHz". The device identification feature collision detection may be implemented by comparing the hardware compatibility list in the parameter set with the hardware model code of the target device, for example, if the parameter set requires support of "HW-5G-IND-02" and the target device is "HW-5G-IND-01", the configuration item of "4G CA auxiliary band 1.8GHz" in the parameter set is removed because it depends on the specific hardware filter module. After the redundant parameter items are removed, the remaining parameter sets are packaged into a conflict-free standardized instruction template.
And step S360, dynamically calibrating the connection stability parameters in the candidate configuration parameter set according to the real-time fluctuation trend of the signal intensity distribution parameters to form a configuration strategy sequence comprising calibrated parameters and priority labels.
Illustratively, the real-time fluctuation trend of the signal strength may calculate the variance change rate of the signal strength in the last 5 minutes by a sliding window algorithm, for example, the variance rising from 8dB2 to 15dB2 indicates that the signal stability is degraded. The dynamic calibration module adjusts the heartbeat packet interval (from 30 seconds to 20 seconds) and the handover decision threshold (from-75 dBm to-70 dBm) in the parameter set accordingly. The calibrated parameter set is accompanied by a priority label, e.g. "high stability priority" or "low energy priority", the label assignment being based on the scene identifier in the device identification feature, the industrial device being labeled "high stability" and the consumer electronics being labeled "low energy".
And step S370, when the network access mode or the geographic position coordinate in the environmental feature set is detected to be changed, triggering the local reordering of the configuration strategy sequence, and performing secondary verification on the signal strength adaptation condition of the candidate configuration parameter set based on the changed environmental feature set.
For example, when the device switches from 5G NSA to 4G CA network access mode, the parameter set dependent on the 5G frequency band in the configuration policy sequence is downgraded, and simultaneously the parameter set dedicated to 4G CA is newly added to the head of the queue. The change in the geographic location coordinates triggers a geofence overlap recalculation, and if the device is driving into a regular uncovered area, the set of associated parameters is temporarily disabled. The suitability of the parameter set is re-evaluated by the signal strength data acquired in real time by a second verification, for example, after the signal mean falls to-72 dBm, only the parameter set with the threshold value less than or equal to-72 dBm is reserved.
In one embodiment, in the step S300, generating a configuration policy sequence of the target eSIM device further includes:
step 301, monitoring the network connection state of the target eSIM device in real time, and triggering dynamic adjustment of a configuration policy sequence when detecting that the network delay exceeds a preset delay threshold.
Illustratively, the network delay threshold is set according to the device type, e.g., 50ms for industrial control devices and 200ms for video monitoring devices. The monitoring module generates a delay alert event by continuously transmitting probe packets and calculating a Round Trip Time (RTT) when a mean of 3 consecutive RTTs exceeds a threshold. For example, the RTT of the vehicle terminal increases from 40ms to 80ms, triggering the dynamic adjustment procedure.
Step S302, selecting a secondary priority candidate configuration parameter set from the configuration strategy sequence, and adjusting a connection retry interval and a data compression mode in the secondary priority candidate configuration parameter set based on a current network delay parameter.
Illustratively, the secondary priority candidate configuration parameter set is the parameter set that ranks second in the sequence, e.g., the parameter set that is "high stability" in primary priority is replaced with the "low latency" parameter set. The connection retry interval is adjusted from a default 5 seconds to 2 seconds to speed up connection recovery, and the data compression mode is switched from lossless compression to lossy compression, reducing the amount of transmitted data by 30%.
Step S303, inserting the adjusted secondary priority candidate configuration parameter set into the first position of the configuration strategy sequence, and generating a new remote configuration instruction.
Illustratively, the new remote configuration instruction overrides the connection parameters in the original instruction, e.g., changes the APN (access point name) from "reduce. APN" to "lowlatency. APN", and adds an urgent reconfiguration identifier at the instruction header. The cloud platform interface preferentially sends the instruction to the target equipment to ensure that the target equipment preempts the current transmission channel.
Step S304, after the target eSIM equipment successfully executes the new remote configuration instruction, the adjusted secondary priority candidate configuration parameter set and the corresponding network delay parameter are recorded into the strategy generation rule of the dynamic strategy model.
Illustratively, after successful execution, the configuration response data uploaded by the device side includes a new network delay indicator (e.g., RTT is reduced to 45 ms) and compression efficiency data. The dynamic policy model updates the rule base through an online learning mechanism, for example, under the condition of 'network delay >50 ms', a low-delay APN and a lossy compression branch are newly started, and the historical execution success rate is initialized to 100% for subsequent preferential calling.
And step 400, generating a remote configuration instruction according to the highest priority candidate configuration parameter set in the configuration strategy sequence, and transmitting the remote configuration instruction to the target eSIM equipment through a cloud platform interface.
Illustratively, the remote configuration instruction is, for example, a binary data packet conforming to the GSMA remote SIM configuration specification, and the encapsulation process includes, for example, converting network access parameters in the candidate configuration parameter set to an APN access point name, authentication key, and QoS quality of service class identifier, while adding a specific instruction header check code according to the hardware version number of the target eSIM device. The cloud platform interface establishes a bidirectional data channel with the target eSIM equipment by adopting an asynchronous communication protocol, and sends the configuration instruction to the equipment end in a slicing way through an HTTPS encryption transmission layer. In the process of issuing the instruction, the cloud platform monitors network transmission delay and data packet retransmission rate in real time, and automatically triggers a re-sequencing mechanism of a configuration strategy sequence when detecting that the signal intensity distribution parameters change obviously. For example, if the standard deviation of the signal intensity of the target device suddenly increases beyond the threshold value in the instruction transmission stage, the cloud platform immediately stops issuing the current instruction, and re-executes step S300 to generate a new configuration policy sequence, so as to ensure the real-time suitability of the configuration instruction.
As an embodiment, the step S400, generating a remote configuration instruction according to the highest priority candidate configuration parameter group in the configuration strategy sequence, wherein the method specifically comprises the following steps of:
Step S410 is to extract the network connection parameters, the security authentication parameters and the service subscription parameters in the highest priority candidate configuration parameter set, and to verify the compatibility of the network connection parameters with the hardware model codes in the device identification feature of the target eSIM device.
Illustratively, the network connection parameters are a set of network access configurations defined in the configuration policy, including an APN (access point name), qoS (quality of service) class, and a band locking policy, for example, an APN of "reduced. Iot", a QoS class of "priority 5", and locking to a 3.5GHz band. The security authentication parameters include a pre-shared key identifier, a certificate fingerprint, and a type of mutual authentication protocol, such as a key identifier "KID_2023_V2" that associates the AES-256 encryption algorithm with the SHA-256 signature mechanism. The service subscription parameters define cloud service endpoints and API rights lists that the device can access, e.g., allow access to "https:// api.iotaoplatform.com/v 1/data" but prohibit access to "https:// api.otaupdate.com". The compatibility verification is realized by comparing hardware dependent items in network connection parameters with hardware model codes of target equipment, for example, a frequency band locking strategy requires hardware to support a 3.5GHz frequency band filtering module, and the module is explicitly contained in a hardware specification document of the equipment model code HW-5G-IND-01', so that the verification is judged to pass.
Step S420, if verification is passed, combining the security authentication parameter and the service subscription parameter according to a preset instruction template structure to generate an initial configuration instruction draft, wherein the instruction template structure comprises a parameter type identifier, a parameter value placeholder and a check code generation bit.
Illustratively, the instruction template structure is an XML architecture conforming to the GSMA specification, wherein parameter type identifiers are used to distinguish network connection, security authentication, and service subscription parameter categories, such as "< NetworkConfig >" "< SecurityAuth >" "< ServiceSub >" tags. The parameter value placeholder marks the actual parameter value to be filled in a "${ }" format, e.g. "APN _value" in "< APN > $ { APN _value }" corresponds to "invalid. Iot". Check code generation bits are reserved at the tail of the instruction and are used for storing CRC32 check codes calculated subsequently. In the combination process, a key identifier ' KID_2023_V2 ' in the security authentication parameter is filled in with a ' KeyID > $ { key_id }/KeyID > -placeholder, and an API endpoint list in the service subscription parameter is serialized into a JSON array format of ' AllowURL > $ { url_list }/AllowURL > ', so as to form an unencrypted initial configuration instruction draft.
And step S430, carrying out protocol adaptation conversion on the initial configuration instruction draft according to the type of the transmission protocol of the current connection cloud platform interface of the target eSIM equipment, filling a protocol header field and replacing a parameter value placeholder as an actual parameter value.
Illustratively, the transport protocol types include HTTPS, coAP, and MQTT, for example, when the device is currently using HTTPS long connection, the protocol adaptation module adds the HTTP header fields "Host: io t. Config. Com" and "Content-Type: application/xml" before the instruction draft. Parameter value placeholders are replaced by a key-value mapping mechanism, e.g., replacing "$ { apn _value }" with "reduced. Iot", and "$ { url_list }" with "[ 'https:// api.iota iotaotplane.com/v 1/data' ]. For binary protocols such as CoAP, the instruction draft needs to be converted into CBOR (concise binary object representation) format, and MESSAGE ID and Token fields in the CoAP message header are added to ensure protocol compatibility.
Step S440, based on the key identifier in the security authentication parameter, a corresponding encryption algorithm is called from a security repository of the cloud platform to carry out end-to-end encryption on the initial configuration instruction draft after protocol adaptation, and an encryption configuration instruction data packet is generated.
Illustratively, the secure store employs a Hardware Security Module (HSM) to manage the keying material, which returns a corresponding AES-256-GCM encryption key and initialization vector when a key identifier "KID_2023_V2" is entered. The encryption process divides the plaintext instruction draft into 128-byte data blocks, sequentially performs encryption operation and attaches an authentication tag to generate an encryption configuration instruction data packet in a binary format. The data packet structure complies with the ISO/IEC 7816-4 specification, and includes an encryption algorithm identifier "0x01" representing AES-256-GCM, a ciphertext data segment, and a 16-byte integrity check tag.
And S450, determining a fragmentation strategy and a transmission priority mark of the encryption configuration instruction data packet according to the bandwidth limit and the signal intensity distribution parameter in the network connection parameter, wherein the fragmentation strategy comprises a fragmentation size upper limit and retransmission redundancy fragmentation number.
The bandwidth limitation is determined by the network access mode of the target device, for example, when the bandwidth of the 4G network is 50Mbps, the upper limit of the slice size is set to 1024 bytes to avoid the timeout of single slice transmission, and if the signal strength distribution parameter shows a standard deviation of 12dB (high volatility), the number of retransmission redundancy slices is increased from 1 to 3 by default to improve the transmission reliability. The transmission priority mark is set according to the emergency degree of the configuration strategy, for example, the mark is 'priority high' when the equipment is in a roaming state, and a priority queue scheduling mechanism of the cloud platform interface is triggered.
Step S460, injecting the slicing strategy and the transmission priority mark into the metadata segment of the encryption configuration instruction data packet to generate a complete remote configuration instruction.
Illustratively, the metadata fragment is appended to the encrypted packet header in a TLV (type-length-value) format, e.g., the fragmentation policy is encoded as a type code "0x1F", a length "0x04", a value "0400" (representing a fragment size of 1024 bytes) and "0301" (representing a redundant fragment number of 3). The transmission priority flag is encoded as a type code "0x2A", a length "0x01", a value "0x01" (high priority). After the injection is completed, the length of the complete instruction packet is expanded to be the sum of the encrypted data segment and the metadata segment, and the structural integrity is verified through a state checking module of the cloud platform interface.
And step S470, if the verification is not passed, triggering a replacement mechanism of the secondary priority candidate configuration parameter set in the configuration strategy sequence, and recording invalid network connection parameters in the highest priority candidate configuration parameter set to a strategy generation rule exception list of the dynamic strategy model.
For example, when the band locking policy in the highest priority parameter set requires a 4.9GHz band and the hardware of the device model code "HW-5G-IND-01" is not supported, the system automatically selects the sub-priority parameter set (e.g., 3.5GHz band policy) to re-perform steps S410 to S460. The invalid parameter "4.9GHz band lock" is recorded to an exception list, the list entry contains a device identification feature, an environmental feature set, and a failure cause code "err_hw_ INCOMPATIBLE", and the conflicting relation between the parameter and the device model code is added to the exclusion condition of the policy generation rule in the subsequent model increment training period.
And step S500, receiving configuration response data returned by the target eSIM equipment, and updating a policy generation rule of a dynamic policy model associated with the equipment identification feature in the dynamic policy model library according to the configuration response data.
The configuration response data illustratively includes, for example, an execution result code of the remote configuration instructions by the target eSIM device, an identifier of the set of configuration parameters that are actually in effect, and network performance metrics measured at the device side, such as data transmission rate, signal strength steady state values, and power consumption level. The cloud platform analyzes the deviation degree between the preset configuration parameter set and the actual effective parameter set through a difference comparison algorithm, and automatically triggers a weight adjustment mechanism of the strategy generation rule when detecting that a plurality of devices with similar geographic coordinates have failure execution of the same configuration parameter set. For example, a specific updating process may include strengthening decision weights of standard deviation factors in signal strength distribution parameters, adding a network switching success rate penalty term to a loss function of a dynamic policy model, and performing online iterative optimization on configuration policy generation rules based on a reinforcement learning algorithm. For example, if a 5G configuration connection timeout frequently occurs in a highway scene for a certain type of vehicle-mounted terminal, the dynamic policy model will automatically increase the speed threshold condition in the policy generation rule, and forcedly demote to the 4G CA configuration parameter set when the device moving speed exceeds 120km/h, and synchronize the rule to the dynamic policy models of all the associated device types in the model library.
In one embodiment, in the step S500, the updating the dynamic policy model library according to the configuration response data may specifically include the following steps:
step S510, analyzing the actual configuration parameter execution result and the equipment state index in the configuration response data.
For example, the actual configuration parameter execution result refers to an operation status code and an effective parameter list returned after the target eSIM device executes the remote configuration instruction, for example, the status code "0x00" indicates that the APN (access point name) and the band locking policy have been successfully applied, and the status code "0xE1" indicates that the security authentication parameter verification fails. The device status indicator contains real-time network performance data after the configuration is validated, e.g., network delay decreases from 120ms to 45ms, signal strength average stabilizes at-68 dBm and standard deviation of 7dB. The parsing process separates data fields through regular expression matching and JSON deserialization technology, for example, extracts key value pairs such as "apn=reduce. Iot" and "qos=5" from response data, and converts a network delay index "latency=45" into a floating point type numerical value to store the floating point type numerical value in an analysis queue.
And step S520, performing difference comparison between the actual configuration parameter execution result and the highest priority candidate configuration parameter set in the configuration strategy sequence, and determining the parameter deviation type and the deviation amplitude.
Illustratively, the discrepancy comparison algorithm employs a field-by-field comparison mechanism, e.g., an APN defined in the highest priority parameter set is "invalid. Iot" and the actual effective APN is "backup. Iot", then it is determined as "APN parameter bias", and if the signal strength average is expected to be-65 dBm and the actual value is-68 dBm, then the calculated bias amplitude is 3dB. The parameter deviation types are classified into compatible deviation and incompatible deviation, wherein the parameter value difference does not affect the core function (such as QoS grade is adjusted from 5 to 4), and the parameter deviation type refers to abnormal function caused by parameter deficiency or boundary crossing (such as frequency band locking strategy is not effective).
And step S530, if the parameter deviation type is compatible deviation and the deviation amplitude is within a preset tolerance range, keeping the strategy generation rule of the dynamic strategy model unchanged, and recording the execution result of the actual configuration parameters as the supplementary training data.
Illustratively, the preset tolerance range is dynamically set according to the parameter type, such as signal strength mean tolerance ±5dB, network delay tolerance ±20ms. When the APN deviation is a compatible standby access point of the same operator and the signal strength deviation is 2dB, the dynamic policy model rule base maintains the original conditional branch, and meanwhile takes an actual parameter value of 'APN=backup. Iot' and 'signal_mean= -68 dB' as a new added sample to be added into the training data set. The supplemental training data is stored in an unstructured database of the cloud platform and its associated device identification features and environmental feature sets are marked by data versioning.
Step S540, extracting the environmental feature set and the equipment identification feature and starting the incremental training of the dynamic strategy model if the parameter deviation type is incompatible deviation or the deviation amplitude exceeds a preset tolerance range, wherein the incremental training comprises the steps of recalculating conditional branch weights in a strategy generation rule by taking the environmental feature set, the equipment identification feature and an actual configuration parameter execution result as newly added sample data and updating a segmentation point threshold of the strategy decision tree model.
For example, when the actual effective frequency band and the expected frequency band deviate by more than 10MHz due to incompatibility of hardware, the incremental training module loads the device identification feature HW-5G-IND-01 and the environmental feature set of '5G NSA mode+urban center area', and re-evaluates the condition weight of the frequency band selection rule in the strategy decision tree. The segmentation point threshold adjustment employs a gradient descent algorithm, such as optimizing the signal strength mean threshold from-70 dBm to-68 dBm to match the actual validation conditions.
As an embodiment, the incremental training may further include the steps of:
Step S541 is to extract the changing trend of the network access mode and the moving track of the geographic position coordinates from the newly added sample data.
Illustratively, the network access mode trend is at an average interval of 15 minutes for switching from 5G NSA to 4G CA by the sliding window statistics device at the last 24 hours network type switching frequency. The geographical position coordinate movement track adopts a Kalman filtering algorithm to smooth the original coordinate sequence, so that longitude and latitude change rate and direction angle are generated, for example, the equipment moves along the direction of longitude 116.30 DEG to 116.45 DEG at the speed of 60 km/h. The trajectory data is encoded in GeoJSON format and associated with a sequence of time stamps for spatio-temporal feature analysis.
And step S542, predicting expected environmental characteristics of the target eSIM equipment in a future time window according to the change trend and the movement track.
Illustratively, the future time window length is adaptively adjusted according to the movement rate, e.g., the high speed mobile device is set to 5 minutes and the stationary device is set to 1 hour. The expected environmental feature prediction uses an ARIMA (autoregressive integrated moving average) model, for example, the prediction device switches the incoming signal strength mean-72 dBm, network access mode to the region of 4G CA after 5 minutes. The expected coordinates of the geographic location are calculated by linear extrapolation, e.g., the current longitude 116.40 ° increases at a rate of 0.002 °/min, then after 5 minutes the expected longitude is 116.40 ° + (0.002×5) = 116.41 °.
Step S543, adding a temporary decision path in the strategy decision tree model, wherein the temporary decision path maps the expected environmental characteristics to a preset preparation configuration parameter set.
Illustratively, the temporary decision path is identified with an "expected_" prefix, e.g., the newly added path "expected_4g_ca_116.41 °" associates the set of provisioning configuration parameters "enables 4G CA carrier aggregation with 2.4GHz Wi-Fi dual connectivity. The path condition is set to "IF network access mode=4gcaand longitude is equal to or greater than 116.40 ° AND longitude is equal to or less than 116.42 ° THEN execution of the preparation parameter set". The validity period of the temporary path is set to be the length of a future time window, and the temporary path is automatically marked as a state to be recovered after overtime.
Step S544, when it is detected that the matching degree between the environmental features of the target eSIM device and the expected environmental features exceeds the preset matching threshold, preferentially executing the remote configuration instruction corresponding to the preparation configuration parameter set.
Illustratively, the matching degree calculation adopts a cosine similarity algorithm, for example, the similarity between the actual environment feature vector and the expected vector is 0.92 (threshold value is 0.85), and the preparation parameter set is triggered to be issued. The set of provisioning parameters is preemptively issued through the cloud platform interface, for example, when the device enters 300 meters before the 116.41 ° longitude zone, the 4G CA configuration is activated in advance to avoid signal switching delays.
Step S545, after the future time window is finished, if the preparation configuration parameter set is not triggered to be executed, the temporary decision path is removed from the strategy decision tree model.
For example, the removal mechanism may be implemented by a timer and a status flag bit, e.g., after the end of the future time window time stamp is reached, the system scans all "expected_" prefix paths, and if its "triggered" flag is false, the paths are deleted and associated memory resources are released. The removal operation simultaneously cleans the instruction cache of the preparation parameter set, and ensures that the policy decision tree model maintains optimal decision efficiency.
As one embodiment, the method further includes a process of exception configuration processing, specifically including:
and S600, after the remote configuration instruction is issued, starting configuration execution countdown.
Illustratively, the configuration execution countdown is a timeout monitoring window dynamically set based on the network connection quality of the target eSIM device, e.g., setting the countdown duration to 30 seconds when the network delay is 50ms, and extending to 90 seconds when the delay increases to 150 ms. The countdown trigger mechanism is realized through a task scheduler of the cloud platform, the internal clock precision is in the millisecond level, and the follow-up abnormal processing flow is ensured to be accurately triggered when the configuration response data is not received in a preset time window. For example, when the remote configuration instruction includes a frequency band locking policy and a security authentication parameter, the countdown window is adjusted to 45 seconds according to the instruction complexity, and a sufficient time is reserved for completing radio frequency calibration and key negotiation at the device side.
And step S700, if the configuration response data is not received before the countdown is finished, sending a configuration state query request to the target eSIM equipment.
Illustratively, the configuration status query request employs a dual composite transmission mechanism, the primary request is sent over a currently active cloud platform interface (e.g., HTTPS long connection), and the backup request is switched to a low-power wide area network protocol (e.g., NB-IoT) for redundant transmission. The request message structure contains the instruction sequence number, the device identification feature hash value, and a query type identifier, e.g., query type identifier "0x03" indicates the execution status code that requires the device to return the configuration instruction and error details of the outstanding operations. After receiving the inquiry request, the equipment end interrupts the ongoing data transmission task, responds to the state inquiry preferentially and returns a response data packet containing the snapshot of the current configuration process.
Step S800, judging the abnormal type according to the feedback result of the configuration state query request. The anomaly type judging module is mapped to a predefined anomaly classification tree by analyzing the state code and the error log in the response data packet.
For example, the status code "0xE2" corresponds to "security certificate chain verification failure" and is classified as security authentication anomaly, and the status code "0xD4" corresponds to "frequency band resource unavailable" and is classified as hardware compatibility anomaly. The stack trace information in the error log further assists in refining the exception subclasses, for example, in the case of unavailable frequency band resources, distinguishing between the two sub-types of "hardware filter failure" and "spectrum grant expiration" to provide accurate input for subsequent processing.
And step 900, if the feedback result is that the configuration instruction does not arrive, switching the data transmission channel of the cloud platform interface and re-issuing the remote configuration instruction.
Illustratively, the data transmission channel switching policy is dynamically selected according to the network access mode and the signal strength distribution parameter of the target device, for example, switching from a default HTTPS channel to MQTT over TCP protocol to bypass firewall restrictions, or enabling a fast retransmission mode of UDP protocol to promote the instruction arrival rate. The re-issued remote configuration instruction is added with a transmission priority mark and a fragmentation redundancy check code, for example, a transmission strategy of 512 bytes in the fragmentation size and 2 redundant fragments is adopted in an NB-IoT channel, so that the instruction integrity in a high packet loss rate environment is ensured.
If the feedback result is that the configuration instruction is received but the execution fails, selecting the candidate configuration parameter set of the next priority from the configuration policy sequence, and adding the debug instruction to regenerate the remote configuration instruction.
Illustratively, the next priority candidate configuration parameter set is the next to highest priority alternative in the current sequence, e.g., when the highest priority parameter set fails due to a band collision, the next priority parameter set enables the alternate band 3.5GHz and simplified version security authentication procedure. The debug instructions include enabling the device-side log tracking function, boosting the diagnostic information reporting frequency to once per second, and setting the debug level to "VERBOSE" to capture the underlying drive errors. The regenerated instruction is adapted to the current transmission channel through a protocol conversion module, for example, a Base64 coded debug parameter is embedded in the CoAP protocol to a message option field.
And step S1100, recording the exception type and the processing measure to an exception processing knowledge base of the dynamic policy model, and preferentially avoiding configuration parameter combinations causing exceptions in the subsequent policy generation process.
Illustratively, the exception handling knowledge base employs a graph database to store associations of exception events and handling measures, e.g., node "security certificate chain verification failed" association edge "switches to secondary CA certificate" and "offline signature verification enabled". After the knowledge base is updated, the dynamic strategy model calls a graph traversal algorithm when a configuration strategy sequence is generated, and configuration parameter combinations which historically cause similar anomalies are actively removed, for example, when the fact that the equipment identification features contain HW-5G-IND-01 "is detected, parameter groups which depend on a 4.9GHz frequency band are automatically skipped.
As an embodiment, the updating process of the exception handling knowledge base may include the steps of:
Step S1110, counting the frequency of triggering the same abnormal type for a plurality of times under the same equipment identification characteristic.
For example, the frequency statistics may employ a sliding time window counting method, such as counting the number of times the device identification feature "HW-5G-IND-01_89860121" triggers a "band resource unavailable" anomaly within the past 24 hours. The counting result is stored in a time sequence database, and the abnormal triggering rate is calculated through an exponential weighted moving average algorithm, for example, if the number of the abnormal times exceeds 3 in each hour, the high-frequency abnormal source is judged. In the statistical process, the environment feature set and the configuration parameter group identifier are associated, and a multidimensional abnormal feature image is established.
And S1120, marking the associated configuration parameter groups in the dynamic strategy model if the frequency exceeds a preset frequency threshold value, and reducing the priority of the marked configuration parameter groups when a configuration strategy sequence is generated.
Illustratively, the preset frequency threshold may be set hierarchically according to device type, e.g., 2 times per hour for industrial-grade devices and 5 times per hour for consumer-grade devices. The tagging operation adds a "risk tag" identifier in the metadata area of the dynamic policy model and multiplies the initial priority weight of the associated parameter set by the degradation coefficient of 0.7. For example, the parameter set with the original priority of 90 points is marked and then falls to 63 points, so that the sorting position of the parameter set at the time of policy generation is shifted backwards.
Step S1130, when the priority of the marked configuration parameter set is lower than a preset priority threshold, the marked configuration parameter set is moved out of the configuration strategy sequence, and emergency retraining of the dynamic strategy model is triggered, wherein the emergency retraining comprises the steps of screening the configuration parameter set irrelevant to the current abnormal type from historical configuration data, and reconstructing a strategy based on the screening result to generate a rule set.
Illustratively, the preset priority threshold is set to 50 points, and when the parameter set priority drops to 49 points, the system automatically moves it out of the configuration policy sequence for the current device. The emergency retraining module loads historical configuration data which does not trigger the similar abnormality in the last 30 days, recalculates conditional branch weights of the strategy generation rules by adopting a random forest algorithm, removes a decision path which has strong association with the abnormal parameter group through pruning operation, AND for example, deletes the rule of 'IF network access mode = 5G AND frequency band = 4.9GHz THEN enables high power mode'.
As an embodiment, the method further includes a process of cross-model migration learning, and specifically may include the following steps:
And step 1200, when the fact that the equipment identification characteristics of the newly added eSIM equipment are not matched with any dynamic strategy model is detected, selecting a reference dynamic strategy model with highest equipment identification characteristic similarity from a dynamic strategy model library.
Illustratively, the similarity calculation adopts a mixed feature weighting algorithm, the hardware model code similarity weight is 60%, the firmware version sequence similarity is 30%, and the operator code combination similarity is 10%. For example, the hardware coding edit distance of the newly added device identification feature "HW-5G-IND-02_89860122" and the reference MODEL "modem_5g_ind_v3" is 1 (similarity 95%), the firmware version difference is a minor version number (similarity 85%), the operator code overlap ratio is 80%, and the comprehensive similarity score is 0.6x95+0.3x85+0.1x80=90.5 points, so that the newly added device identification feature becomes the best reference MODEL.
And step S1300, extracting a policy generation rule set of the reference dynamic policy model, and removing rule conditions strongly associated with the equipment identification features.
Illustratively, a strong association rule condition refers to a decision branch directly referencing a particular hardware model or operator code, e.g. "IF hardware model code = HW-5G-IND-01 THEN enable proprietary radio frequency calibration" is identified as a device specific rule. The removing operation analyzes the rule expression through a grammar analyzer, deletes the logic judgment containing the equipment identification characteristic field, and reserves the general environment characteristic conditions such as 'signal intensity mean value is more than or equal to-70 dBm' and 'network access mode=5GNSA'.
And step 1400, taking the removed strategy generation rule set as an initial rule set, loading real-time configuration request data of the new eSIM equipment to carry out rule filling, wherein the rule filling comprises dynamically adding a network access mode and a self-adaptive threshold condition of signal intensity according to the environment feature set of the new eSIM equipment, and associating a general configuration parameter set.
Illustratively, after the feature extraction of the environmental feature set in the real-time configuration request data, the driving rule generating engine creates a new conditional branch, for example, IF the signal intensity standard deviation of the newly added device in the 4G CA mode is detected to be stabilized within 5dB, the rule "IF network access mode=4g CA AND signal intensity standard deviation is less than or equal to 6dB THEN enables CA carrier aggregation optimization configuration". The set of generic configuration parameters is selected from a common pool of model libraries, such as "low power base configuration" and "high throughput default configuration".
And S1500, packaging the filled policy generation rule set into a newly added dynamic policy model, and adding the newly added dynamic policy model into a dynamic policy model library.
Illustratively, the encapsulation process includes assigning a unique MODEL identifier "MODEL_5G_IND_V4" to the rule set, generating version metadata (e.g., creating a timestamp "20231001T143000" and applicable device feature range "HW-5G-IND-02 series"), and compiling the rule set into executable code for the policy decision tree. After the newly added model passes the consistency verification test, the newly added model is registered in an index catalog of a dynamic strategy model library, and a subsequent configuration request can directly call the model through equipment identification feature mapping.
Referring to fig. 3, which is a block diagram of the cloud platform 120 of the present invention, the cloud platform 120 includes a computing unit 1001 that can perform various appropriate actions and processes according to a computer program stored in a Read Only Memory (ROM) 1002 or a computer program loaded from a storage unit 1008 into a Random Access Memory (RAM) 1003. In the RAM 1003, various programs and data required for the operation of the cloud platform 120 may also be stored. The computing unit 1001, the ROM 1002, and the RAM 1003 are connected to each other by a bus 1004. An input/output (I/O) interface 1005 is also connected to bus 1004.
The various components in cloud platform 120 are connected to I/O interface 1005, including input unit 1006, output unit 1007, storage unit 1008, and communication unit 1009. The input unit 1006 may be any type of device capable of inputting information to the cloud platform 120, the input unit 1006 may receive input numeric or character information and generate key signal inputs related to user settings and/or function control of the server, and may include, but is not limited to, a mouse, a keyboard, a touch screen, a trackpad, a trackball, a joystick, a microphone, and/or a remote control. The output unit 1007 may be any type of device capable of presenting information and may include, but is not limited to, a display, speakers, video/audio output terminals, vibrators, and/or printers. Storage unit 1008 may include, but is not limited to, magnetic disks, optical disks. Communication unit 1009 allows cloud platform 120 to exchange information/data with other devices via a computer network, such as the internet, and/or various telecommunications networks, and may include, but is not limited to, modems, network cards, infrared communication devices, wireless communication transceivers and/or chipsets, such as bluetooth (TM) devices, 802.11 devices, wiFi devices, wiMax devices, cellular communication devices, and/or the like.
The computing unit 1001 may be a variety of general and/or special purpose processing components having processing and computing capabilities. Some examples of computing unit 1001 include, but are not limited to, a Central Processing Unit (CPU), a Graphics Processing Unit (GPU), various specialized Artificial Intelligence (AI) computing chips, various computing units running machine learning model algorithms, a Digital Signal Processor (DSP), and any suitable processor, controller, microcontroller, etc. The computing unit 1001 performs the various methods and processes described above, such as the cloud platform-based eSIM remote configuration management method. For example, in some embodiments, the cloud platform-based eSIM remote configuration management method can be implemented as a computer software program tangibly embodied on a machine-readable medium, such as the storage unit 1008. In some embodiments, part or all of the computer program may be loaded and/or installed onto cloud platform 120 via ROM 1002 and/or communication unit 1009. When the computer program is loaded into RAM 1003 and executed by the computing unit 1001, one or more steps of the cloud platform-based eSIM remote configuration management method described above can be performed. Alternatively, in other embodiments, the computing unit 1001 may be configured to perform the cloud platform-based eSIM remote configuration management method in any other suitable manner (e.g., by means of firmware).
Various implementations of the systems and techniques described here above may be implemented in digital electronic circuitry, integrated circuit systems, field Programmable Gate Arrays (FPGAs), application Specific Integrated Circuits (ASICs), application Specific Standard Products (ASSPs), systems On Chip (SOCs), complex Programmable Logic Devices (CPLDs), computer hardware, firmware, software, and/or combinations thereof. These various embodiments may include being implemented in one or more computer programs that are executable and/or interpretable on a programmable system including at least one programmable processor, which may be a special or general purpose programmable processor, operable to receive data and instructions from, and to transmit data and instructions to, a storage system, at least one input device, and at least one output device.
Program code for carrying out methods of the present invention may be written in any combination of one or more programming languages. These program code may be provided to a processor or controller of a general purpose computer, special purpose computer, or other programmable data processing apparatus such that the program code, when executed by the processor or controller, causes the functions/operations specified in the flowchart and/or block diagram to be implemented. The program code may execute entirely on the machine, partly on the machine, as a stand-alone software package, partly on the machine and partly on a remote machine or entirely on the remote machine or server.
In the context of the present invention, a machine-readable medium may be a tangible medium that can contain, or store a program for use by or in connection with an instruction execution system, apparatus, or device. The machine-readable medium may be a machine-readable signal medium or a machine-readable storage medium. The machine-readable medium may include, but is not limited to, an electronic, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any suitable combination of the foregoing. More specific examples of a machine-readable storage medium would include an electrical connection based on one or more wires, a portable computer diskette, 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 disc read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing.
In summary, the present invention provides a cloud platform, which includes at least one processor, and a memory communicatively connected to the at least one processor, where 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 perform the method provided by the foregoing embodiments of the present invention.
It should be appreciated that various forms of the flows shown above may be used to reorder, add, or delete steps. For example, the steps recited in the present invention may be performed in parallel, sequentially or in a different order, provided that the desired results of the disclosed embodiments are achieved, and are not limited herein.
Although embodiments or examples of the present invention have been described with reference to the accompanying drawings, it is to be understood that the above-described methods, systems, and apparatus are merely illustrative embodiments or examples, and that the scope of the present invention is not limited by these embodiments or examples but only by the claims following the grant and their equivalents. Various elements of the embodiments or examples may be omitted or replaced with equivalent elements thereof. Furthermore, the steps may be performed in a different order than described in the present invention. Further, various elements of the embodiments or examples may be combined in various ways. It is important that as technology evolves, many of the elements described herein may be replaced by equivalent elements that appear after the invention.