A Disaster Recovery Method and System for XR Simulation Experiment Data Based on Edge Caching
By using edge caching technology and a neighboring machine collaborative transmission mechanism, the problem of low data backhaul efficiency in weak network environments during XR virtual simulation experiments was solved, achieving efficient and reliable data backhaul in weak network environments.
Patent Information
- Application Number
- CN202611560786.1
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2026-10-10
- Publication Date
- 2026-11-17
AI Technical Summary
In XR virtual simulation experiments, existing technologies lack effective collaborative disaster recovery solutions in weak network environments, resulting in a large number of invalid response messages occupying local area network bandwidth, increasing terminal parsing pressure, and affecting data backhaul efficiency.
By using edge caching technology, simulated data packets are persistently stored for local transactions. Within the local area network, reachability detection and response eligibility determination of neighboring machines are performed, invalid response messages are filtered, and a neighboring machine collaborative transmission mechanism is established to ensure data security and efficiency.
In weak network environments, by using neighboring machines to collaboratively transmit data, the generation of invalid packets can be reduced, network bandwidth usage can be decreased, terminal parsing pressure can be alleviated, and data return efficiency and reliability can be improved.
Smart Images

Figure CN123123262A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of electronic digital data processing technology, specifically a disaster recovery and backhaul method and system for XR simulation experiment data based on edge caching. Background Technology
[0002] During the operation of XR virtual simulation experiments, weak network conditions caused by wireless network jitter and wide area link fluctuations are frequently encountered. The XR terminal equipment itself is operating normally, but the quality of the wide area service communication link between the local device and the remote virtual simulation platform is degraded, making it impossible to directly upload simulation experiment data packets to the remote virtual simulation platform. However, the terminal can rely on LAN broadcast, multicast or unicast to complete message interaction with other XR terminals in the LAN. In this scenario, it is objectively necessary to rely on neighboring devices in the LAN to complete the transmission and return of simulation experiment data. However, the existing technology lacks a complete collaborative disaster recovery solution for such weak network incomplete disconnection scenarios. In existing XR simulation experiment data disaster recovery and backhaul solutions, when the source terminal link fails and needs to find a neighboring machine to backhaul the experiment data via LAN broadcast, the output backhaul response results are not pre-filtered on the neighboring machine side. All response packets are directly sent to the source-failed XR terminal, which then handles all the parsing and filtering. This easily generates a large number of invalid response packets, consuming LAN transmission bandwidth and increasing the packet parsing and processing pressure on the source-failed XR terminal. In scenarios with a large number of terminals in the network, this further exacerbates the packet interaction overhead and affects the processing efficiency of the disaster recovery and backhaul service. Therefore, there is an urgent need for a method and system for XR simulation experiment data disaster recovery and backhaul based on edge caching. Summary of the Invention
[0003] The purpose of this invention is to provide a disaster recovery and backhaul method and system for XR simulation experimental data based on edge caching, so as to solve the problems raised in the prior art.
[0004] To achieve the above objectives, the present invention provides the following technical solution: a disaster recovery and backhaul method for XR simulation experiment data based on edge caching, the method comprising the following steps: Step S1: Local transaction persistent storage of XR simulation experiment data packets. During the process of running the virtual simulation experiment service, the source XR terminal continuously maintains the business communication link between the local machine and the remote virtual simulation platform, and performs real-time link status monitoring. The link status monitoring includes multi-dimensional link quality indicators such as transport layer connectivity detection, business application layer handshake verification, message round-trip delay and message packet loss rate, which are used to identify local link failure events. When the source XR terminal detects a local link failure and determines that the local terminal cannot directly establish an effective business communication connection with the remote virtual simulation platform and does not have the direct transmission conditions to directly deliver the simulation experiment data packets that have not been uploaded to the virtual simulation platform, it immediately performs a local transaction persistent write operation on the XR simulation experiment data packets that have not been uploaded in the current session. The XR simulation experiment data package includes simulation experiment operation trajectory data, interaction process logs, screen recording fragments of the experiment process, experimental results and scoring metadata, and business data strongly associated with this simulation session. To avoid the risk of data corruption or loss due to XR terminal power failure, system crash, or abnormal process termination during the writing process, persistent writing adopts a transactional storage mechanism. The simulation experiment data package, the data package's unique idempotent key, the source device's unique identifier, the source session's unique identifier, the data package fragment index, and timestamp metadata are included in the same transaction processing scope, and atomic transaction commits are performed. Only after the storage layer sends a transaction confirmation acknowledgment that the transaction has been completely committed is the local persistence of the XR simulation experiment data package considered complete. After the transaction is committed, the source XR terminal retains the original simulation experiment data packet and all associated metadata in the local persistent storage medium. It does not perform deletion, trimming, or overwriting operations on the original data packet. The original data packet will serve as the original business data source for subsequent neighbor machine proxy hosting, retransmission, and idempotent verification. Transaction persistence is the underlying data security foundation of the entire disaster recovery and backhaul mechanism. If a neighboring machine is subsequently entrusted with the transmission, even if the neighboring machine undertaking the transmission task experiences an abnormal situation such as crash, topology escape, or loss of transmission data packets, the source XR terminal still holds the complete original data packets locally and can re-initiate the transmission request. Only after receiving the final platform receipt from the remote virtual simulation platform, confirming that the data packets have been idempotently stored in the database, will the local original data packets be archived or deleted according to the preset archiving and cleanup strategy. The transaction persistence process will not destroy the original business data in advance. Step S2: Obtain and analyze the historical service dataset of XR terminals across the entire domain, and combine it with normalization processing and fusion mapping to obtain the equivalent relative load characteristics of each XR terminal. Collect the historical service dataset of all XR terminals in the local area network cluster registry. The historical service dataset comes from the operation log records of each historical neighbor reachability detection transaction and proxy hosting transaction. The historical service dataset includes at least the historical transaction response processing time of each XR terminal, the persistence IO time of proxy data packets, CPU sampling sequence, memory usage sampling sequence, IO queue time, transaction success sample count and transaction failure sample count. Different XR terminal hardware inherently have objective differences in performance. The CPU and memory load values obtained from the original data collection are only surface-level observation values output locally by the terminal. When the same business task is applied to XR terminals with different hardware capabilities, even if the original load observation values are similar, the actual business pressure borne by each terminal is not the same. Terminals with stronger hardware processing capabilities still have more business capacity margin under the original load conditions. Terminals with weaker hardware processing capabilities have their business capacity margin nearly exhausted under the same original load. The original load observation results cannot be directly used for horizontal comparison of business pressure between different terminals. The upper limit of the performance that a single XR terminal's hardware can achieve is selected as a reference benchmark. Each historical service record corresponds to a complete neighbor detection or data transmission long-term transaction. The average overall load observation result of the transaction within the limited load statistics window is extracted. The single historical service record is treated as a complete whole and normalized within the device. The average overall load observation result of the transaction is transformed to obtain the relative load characteristics of the whole machine relative to the upper limit of the machine's hardware performance. This unifies the feature expression dimension of heterogeneous XR terminals across the entire domain and eliminates the problem that the observation indicators cannot be compared horizontally due to differences in hardware baselines. For all valid historical service records of a single XR terminal after normalization, the arithmetic mean level and the mode level of the sample are calculated respectively. The arithmetic mean level can reflect the overall service performance of the sample set, and the mode level reflects the normal service conditions corresponding to most records. It can weaken the interference of a small number of occasional high or low extreme transaction samples on the evaluation of equipment capabilities. For all valid historical service records after normalization, calculate the statistical measure of the degree of discrete fluctuation of the sample; pre-set two sets of configuration boundaries, corresponding to the acceptable lower limit boundary and the acceptable upper limit boundary of the discrete fluctuation of the sample, respectively. Execution boundary comparison and contribution percentage calculation: When the statistical measurement result reaches or exceeds the acceptable upper limit of discrete fluctuation, all contributions are directly allocated to the sample mode level, and the arithmetic mean level is no longer allocated a contribution percentage. When the statistical measurement result reaches or falls below the acceptable lower limit of discrete fluctuation, all contributions are directly allocated to the arithmetic mean level, and the contribution ratio of the sample mode level is no longer allocated. When the statistical measure result lies between the two sets of configuration boundaries, linear interpolation is performed based on the position of the statistical measure result relative to the two sets of configuration boundaries to solve for the contribution ratio corresponding to the arithmetic mean level and the sample mode level respectively; the contribution ratios of the two types are added together to keep the total contribution constant. After solving for the contribution ratio, the arithmetic average level and the sample mode level are multiplied by their respective calculated contribution ratios and then combined to obtain the comprehensive inherent processing background index that characterizes the inherent service carrying capacity of a single XR terminal. The higher the comprehensive inherent processing background index, the stronger the inherent service carrying capacity of the XR terminal at the hardware level. Based on the comprehensive inherent processing background index obtained from the XR terminals across the entire domain, a cluster business reference benchmark is constructed. For the business task to be executed, the average overall load observation results of the whole machine obtained from the single business transaction reported by the XR terminal within the limited load statistics window are combined with the comprehensive inherent processing background index corresponding to the terminal for mapping and conversion processing to obtain the equivalent relative load characteristics of the business task applied to the corresponding XR terminal. The equivalent relative load characteristic enables a horizontal comparison of the service load of the same business task across XR terminals with different hardware specifications. For the same business task, the stronger the hardware carrying capacity of the terminal, the lower the equivalent relative load characteristic is calculated; the weaker the hardware carrying capacity of the terminal, the higher the equivalent relative load characteristic is calculated. The mapping and conversion process retains the relative magnitude relationship between the indicators of different terminals, and the conversion output result is consistent with the objective state of the actual resource occupation of the terminal. The mapping and conversion parameters corresponding to the integrated inherent processing background index and the equivalent relative load characteristics are persistently saved to the local area network cluster registry. The integrated inherent processing background index serves as the basic reference for determining the qualification of neighboring machines to respond, and the equivalent relative load characteristics are used to complete the cross-terminal service load comparison. This solution addresses the practical problem of inconsistent hardware baselines in heterogeneous XR terminals, which prevents direct horizontal comparison of overall load observation results. Based on the historical sample discrete statistical measurement results formed by the average load of a single business transaction window, and combined with preset configuration boundaries, it quantitatively solves the contribution ratio of two levels through boundary judgment and linear interpolation, quantitatively suppressing the evaluation interference caused by extreme abnormal service transaction samples. By mapping and converting to obtain equivalent relative load characteristics, it unifies the overall load observation results of different hardware terminals based on transaction window statistics to a comparable dimension, providing statistical support for the objective determination of neighboring machine response qualifications. Step S3: Construct a neighbor reachability detection request message based on the unique identifier of the source device of the XR terminal with the source fault. The XR terminal that receives the message combines the sliding time window load analysis and the conversion of the inherent processing background index to obtain the equivalent relative load characteristics, and performs a joint determination of the response qualification. After the XR terminal with the source fault completes the persistence of the local simulation experiment data packet transaction, it reads the unique identifier of the source device of the local machine, and binds the unique identifier of the source session corresponding to this XR simulation experiment, the resource requirement metadata of the simulation data packet to be uploaded, the source end return address information protected by signature and the unique request identifier, assembles and generates a neighbor reachability detection request message, and broadcasts the neighbor reachability detection request message to the logical neighbor in the local area network. Among them, the unique identifier of the source device is carried in the message header, and the unique identifier of the source session, the resource requirement metadata of the simulation data packet to be uploaded, the signature-protected source end return address information, and the unique request identifier are carried in the message payload. Any XR terminal within the local area network that receives a neighbor reachability probe request message first performs a basic validity check on the message. If the message fails the check, it is discarded without further processing. If the message passes the check, a sliding time analysis window is opened with itself as the analysis object. Within the sliding time analysis window period, relevant sampling data on the overall load of the local machine are continuously collected. Based on the sampling sequence within the window, the average overall load and load fluctuation amplitude are calculated. The average overall load represents the overall busy level of the terminal at the current stage, and the load fluctuation amplitude represents the intensity of short-term load oscillations, which is used to identify short-term business peak disturbances. The system reads the pre-stored comprehensive inherent processing background index in the local area network cluster registry, combines it with the currently calculated window average overall system load, and maps and converts it to obtain the equivalent relative load characteristics corresponding to this service. At the same time, it parses and extracts the resource requirement metadata of the simulated data packet to be uploaded from the payload of the neighbor reachability probe request message that has passed the legality verification. This resource requirement metadata represents the overall system resource overhead corresponding to this service. Based on the equivalent relative load characteristics, load fluctuation amplitude, and the resource requirement metadata corresponding to this service, a joint determination of response eligibility is made. The judgment criteria include: First, the equivalent relative load characteristics obtained by mapping and conversion do not exceed the allowable load limit corresponding to the business; Second, the load fluctuation amplitude calculated by the sliding time window does not exceed the preset fluctuation threshold; Third, the remaining resources of the local machine are evaluated by combining the equivalent relative load characteristics. The remaining resources of the local machine can cover the resource overhead of the current upload service represented by the resource requirement metadata. If all three conditions are met simultaneously, the XR terminal is deemed to be qualified to respond; if any one condition is not met, the terminal is deemed not qualified to respond. The upper-layer business logic of this XR terminal will not process the broadcast request message or generate a response message. Other XR terminals in the local area network can still receive the reachability probe request message through the local area network broadcast mechanism and execute their respective judgment processes. Each XR terminal within the local area network independently completes the response qualification determination; the XR terminal that is determined to be qualified to respond generates a response message and, based on the source return address information carried in the message, unicasts the response message back to the source-failed XR terminal; the XR terminals are unaware of each other's determination results; after sending a neighbor reachability probe request message, the source-failed XR terminal starts a preset waiting time. If it receives a response message within the preset waiting time, the probe is considered successful; if it does not receive a response message after the preset waiting time expires, the neighbor reachability probe is considered to have failed. The preset waiting time is configured based on the round-trip latency of messages, the number of online XR terminals in the local area network, and the size of the data packets to be transmitted, as statistically analyzed by the cluster edge gateway node. The cluster edge gateway node continuously collects historical message interaction latency and node load assessment time in the local area network, outputs a duration parameter adapted to the current cluster status, and sends it to the XR terminal with the source fault. The system reuses the inherent processing base color index to generate equivalent relative load characteristics, eliminating load assessment bias caused by differences in hardware baselines of different XR terminals. It obtains overall load statistics through a sliding time analysis window, combines them with the resource overhead of the service to be uploaded obtained from message parsing, and makes a joint judgment based on three dimensions: equivalent relative load characteristics, load fluctuation amplitude, and the matching relationship between the remaining overall resources and service resource overhead. This avoids misjudgments caused by instantaneous service spikes and prevents situations where load indicators are qualified but remaining resources are insufficient to handle the proxy transmission service. Nodes that do not meet the qualifications only abandon the response business logic and do not interfere with broadcast message distribution. Qualified neighboring nodes independently generate response messages and unicast them back to the source faulty XR terminal. Neighboring nodes are unaware of each other's judgment results. The source faulty XR terminal judges the detection result based on the preset waiting time dynamically configured by the cluster edge gateway node. This filters neighboring nodes with insufficient hardware capabilities, severe service fluctuations, or resources that cannot match service overhead, reducing the generation of invalid response messages, lowering LAN message interaction overhead, and providing a reliable admission screening mechanism for the generation and output of neighboring response messages. Step S4: XR terminals that are qualified to process responses, as determined by the joint assessment of response eligibility, generate response messages and perform cross-comparison of historical responses to filter response messages; The joint assessment of response eligibility determines which XR terminals are qualified to respond. Based on the assessment result, the XR terminal performs reachability detection on the remote virtual simulation platform, calculates the remaining available resources of the terminal that can be used for response and proxy services, and generates a neighboring terminal reachability detection response message. The neighboring terminal reachability detection response message includes a unique request identifier, a unique source device identifier, the platform reachability detection result, the terminal's comprehensive inherent processing background index, and the amount of remaining available resources of the terminal. Before sending a neighbor reachability probe response message to the XR terminal with the source failure, the local response cache is queried to retrieve historical response records corresponding to the same source device unique identifier and the same request unique identifier. The remaining available resources carried in the response message to be sent are cross-compared with the remaining available resources carried in the response messages that have been returned by the local machine in response to this probe request in the past. If the remaining available resources in the response message to be sent are worse than the remaining available resources in the historical records, the response is determined to be a quality degradation response message, and the neighbor reachability probe response message to be sent is not sent to the XR terminal with the source failure. If the cross-comparison is successful, the neighbor reachability probe response message is sent to the XR terminal with the source failure. Based on the joint determination of response eligibility, neighbor reachability probe response messages are only generated for XR terminals that are qualified to respond according to the joint determination of response eligibility. Nodes that fail the joint determination of response eligibility do not enter the neighbor reachability probe response message generation process. Before the response message is sent to the source fault XR terminal, a cross-comparison mechanism of local cached historical response records is introduced to filter degraded neighbor reachability probe response messages corresponding to the deterioration of the remaining available resources under the same probe request, so as to avoid the source fault XR terminal receiving response information with degraded quality. This reduces the transmission of invalid neighbor reachability probe response messages, reduces the overhead of local area network message interaction, and ensures that the source fault XR terminal obtains stable and reliable neighbor proxy candidate information. Step S5: The XR terminal with the source fault performs a completeness check on the response message and cross-confirms the fault classification with the neighboring machine, and performs disaster recovery backhaul based on the fault classification result. The source failure XR terminal experienced a failure in the service communication link between itself and the remote virtual simulation platform, but its ability to access the local area network remained normal. The reachability probe request messages from neighboring machines and the simulation experiment data packets to be transmitted were transmitted between the source failure XR terminal and the neighboring machines via the local area network, without passing through the remote virtual simulation platform, and were not affected by the aforementioned service communication link failure. At the same time, the neighboring machines qualified to respond had completed the reachability probe of the remote virtual simulation platform, and the service communication link between the neighboring machines and the remote virtual simulation platform remained normal, enabling them to undertake the data upload task to the remote virtual simulation platform. After sending a neighbor reachability probe request message, the source fault XR terminal starts a preset waiting time. During the preset waiting time, it continuously listens for and receives neighbor reachability probe response messages from each neighbor. When the preset waiting time expires, it stops receiving neighbor reachability probe response messages. At this time, the source fault XR terminal may not receive, receive one or more neighbor reachability probe response messages. Based on all received valid neighbor reachability probe response messages, cross-confirmation of neighboring devices is performed to complete fault classification. The fault classification results include four states: local link available, local link abnormal, platform or public link abnormal, and undetermined. When the reachability detection result of the local machine to the remote virtual simulation platform indicates that the local link is reachable, it is determined that the local link is available; If the local machine is unreachable but the platform reachability detection result of at least one valid neighboring machine indicates that it is reachable, it is determined that the local machine link is abnormal; When the local machine is unreachable and all valid neighboring machines that have reached the preset confirmation threshold are unreachable, it is determined that the platform or public link is abnormal. If the received valid neighbor reachability probe response messages are insufficient to support the above determination, the state is determined to be uncertain. Execute the corresponding disaster recovery backhaul logic based on the fault classification results obtained from cross-confirmation with neighboring machines: When the local link is determined to be abnormal, if the XR terminal with the source fault receives multiple neighbor reachability probe response messages returned by neighboring machines, it selects the best neighboring machine as the receiving neighboring machine based on the platform reachability probe results carried in each neighboring machine reachability probe response message, the local integrated inherent processing background index, and the current remaining available resources of the local machine. The best receiving neighboring machine is the candidate XR terminal with the highest current remaining available resources of the local machine. If only one neighbor reachability probe response message is received, that neighbor is directly designated as the receiving neighbor. The simulation experiment data packet is sent to the receiving neighbor via the local area network. The receiving neighbor persists the simulation experiment data packet and its relay status transactionally, generates a takeover receipt, and returns it to the source fault XR terminal. The source fault XR terminal verifies the takeover receipt and retains the original simulation experiment data packet. The receiving neighbor uploads the simulation experiment data packet to the remote virtual simulation platform through a normal business communication link maintained between itself and the remote virtual simulation platform. The remote virtual simulation platform verifies the source, integrity, resource version, fragment sequence number, and idempotency flag of the simulation experiment data packet. If the verification passes and the idempotency flag has not yet been submitted, the data packet is transactionally entered into the database, and a platform final receipt is generated. This receipt is then returned to the source fault XR terminal via the receiving neighbor or the recovered direct link. After verifying the platform final receipt, the source fault XR terminal marks the simulation experiment data packet as having been entered into the platform database and performs archiving or deletion processing according to the preset archiving and cleanup strategy. If the platform final receipt times out, the source fault XR terminal returns to the pending transmission state and reselects the return path using the original idempotency flag. When a platform or public link is determined to be abnormal, a local delayed backhaul strategy is implemented. The simulation experiment data packets are kept in the local persistent storage medium. The retesting cycle is set according to the fault type. A longer backoff time is used for platform or public link abnormalities, and a shorter retesting time is used when the status cannot be determined. The local link status detection is re-initiated. When the local link is determined to be available again, the XR terminal that caused the source fault directly transmits the simulation experiment data packets to the virtual simulation platform, and uploads them according to the fragment sequence number. The entire process combines idempotent verification logic to avoid duplicate data packet submission, and completes the disaster recovery backhaul processing of XR simulation experiment data. The system monitors and receives neighbor reachability probe response messages based on a preset waiting time. Message reception stops after the preset waiting time expires, making it compatible with various real-world networking scenarios, including no received messages, receiving a single message, and receiving multiple neighbor reachability probe response messages. Based on the acquired valid neighbor reachability probe response messages, it performs cross-confirmation with neighboring machines, resulting in four fault classifications: local link available, local link abnormal, platform or public link abnormal, and indeterminate. The fault classification distinguishes between local link faults and platform or public link faults. When the local link is abnormal, the simulation experiment data packets are delivered to the receiving neighboring machine via the local area network for relay transmission. A double-receipt verification using both the takeover receipt and the platform's final receipt clarifies the data responsibility between the source and the relaying end. When the platform or public link is abnormal, local delayed transmission is performed; when the local link is available, direct transmission is performed. Idempotent verification prevents duplicate data packet submissions, improving the reliability of disaster recovery transmission and the targeted fault handling of simulation experiment data in heterogeneous XR terminal networking scenarios. Furthermore, an XR simulation experiment data disaster recovery and backhaul system based on edge caching is provided. This XR simulation experiment data disaster recovery and backhaul system includes a local transaction persistence module, a load feature parsing module, a neighbor machine response qualification determination module, and a fault classification disaster recovery and backhaul module. The local transaction persistence module is used to complete the local transaction persistence storage of XR simulation experiment data packets based on local link events; the load feature parsing module is used to obtain the historical service dataset of XR terminals across the entire domain and perform mapping processing to obtain the equivalent relative load features of each XR terminal; the neighbor response qualification determination module is used for the source fault XR terminal to construct a neighbor reachability detection request message to detect and compare and filter response messages; the fault classification disaster recovery backhaul module is used to perform disaster recovery backhaul based on the fault classification results. The local transaction persistence module includes a link failure monitoring unit and an atomic transaction writing unit; the link failure monitoring unit is used to monitor and identify local link failure events of the source XR terminal; the atomic transaction writing unit is used to perform atomic transaction persistence writing. The load feature analysis module includes a historical service data acquisition unit and an equivalent load conversion unit; the historical service data acquisition unit is used to collect the historical service dataset of XR terminals in the local area network cluster registry; the equivalent load conversion unit is used to solve the comprehensive inherent processing background index and equivalent relative load features; The neighboring machine response qualification determination module includes a probe message sending and receiving unit and a historical response cross-comparison unit; the probe message sending and receiving unit is used to generate and broadcast neighboring machine reachability probe request messages and determine response qualification; the historical response cross-comparison unit is used to control the sending of neighboring machine reachability probe response messages; The fault classification and disaster recovery backhaul module includes a response message monitoring unit and a disaster recovery backhaul execution unit; the response message monitoring unit is used for cross-confirmation with neighboring machines to obtain fault classification results; the disaster recovery backhaul execution unit is used to select the corresponding processing method to execute backhaul based on the fault classification results.
[0005] Compared with the prior art, the beneficial effects of the present invention are: 1. This invention addresses the scenario of incomplete network disconnection in weak networks. Under the condition that the wide area link of the XR terminal is degraded but the local area network communication is normal, message interaction between neighboring machines is achieved through local area network broadcast, multicast, and unicast. A neighboring machine collaborative data transmission mechanism adapted to this scenario is established. When the terminal cannot directly connect to the remote virtual simulation platform, simulation experiment data can be reliably transmitted and returned by relying on neighboring machines in the local area network, ensuring that simulation experiment service data is not lost in the weak network environment.
[0006] 2. This invention completes the joint determination of response eligibility on the neighboring machine side, and directly blocks the generation of response messages for XR terminals that do not meet the load and resource conditions. It filters the output messages of nodes that do not have the ability to relay messages from the source, eliminating the need for the source fault XR terminal to process a large number of invalid messages, reducing the number of message transmissions in the local area network, reducing network bandwidth occupation, and alleviating the message parsing and processing pressure of the source fault XR terminal.
[0007] 3. This invention adds a historical response cross-comparison step on the neighboring machine side to intercept and filter response messages with degraded quality, further eliminating response outputs with deteriorated performance under the same probe request, reducing the interactive transmission of degraded response messages in the local area network, reducing the overall message interaction overhead in the network, and improving the overall processing efficiency of disaster recovery backhaul services under heterogeneous XR terminal networking. Attached Figure Description
[0008] Figure 1 This is a flowchart illustrating a disaster recovery and backhaul method for XR simulation experimental data based on edge caching, according to the present invention. Detailed Implementation
[0009] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.
[0010] Example 1: As Figure 1 As shown, this invention provides a technical solution: a disaster recovery and backhaul method for XR simulation experiment data based on edge caching. The XR simulation experiment data disaster recovery and backhaul method includes the following steps: Step S1: Persistently store the XR simulation experiment data packet in local transactions; During the operation of virtual simulation experiment services, the source XR terminal continuously maintains the business communication link between the local machine and the remote virtual simulation platform, and performs real-time link status monitoring. The link status monitoring includes multi-dimensional link quality indicators such as transport layer connectivity detection, business application layer handshake verification, message round-trip delay, and message packet loss rate, which are used to identify local link failure events. When the source XR terminal detects a local link failure and determines that the local terminal cannot directly establish an effective business communication connection with the remote virtual simulation platform and does not have the direct transmission conditions to directly deliver the simulation experiment data packets that have not been uploaded to the virtual simulation platform, it immediately performs a local transaction persistent write operation on the XR simulation experiment data packets that have not been uploaded in the current session. The XR simulation experiment data package includes simulation experiment operation trajectory data, interaction process logs, screen recording fragments of the experiment process, experimental results and scoring metadata, and business data strongly associated with this simulation session. Persistent writing is performed using a transactional storage mechanism, incorporating the simulation experiment data package, the data package's unique idempotent key, the source device's unique identifier, the source session's unique identifier, the data package fragment index, and timestamp metadata into the same transaction processing scope, executing atomic transaction commits. Only after the storage layer sends a transaction confirmation receipt indicating successful complete transaction commit is the local persistence of the XR simulation experiment data package considered complete. After the transaction is committed, the source XR terminal retains the original simulation experiment data packet and all associated metadata in the local persistent storage medium. It does not perform deletion, trimming, or overwriting operations on the original data packet. The original data packet serves as the original business data source for subsequent neighbor machine proxy hosting, retransmission, and idempotent verification. In practice, during a virtual simulation experiment, if the wireless network at the location of the XR terminal experiences fluctuations and becomes a weak network environment, the upload channel between the local machine and the remote virtual simulation platform cannot remain stable. At this time, the simulation experiment operation trajectory data, interaction process logs, and screen recording fragments generated by the current session have not yet been uploaded. The aforementioned data packets are then sent to the local transaction storage for writing. The writing action includes the data packet body, the data packet's unique idempotent key, the source device's unique identifier, the source session's unique identifier, the data packet fragment index, and the timestamp metadata into the same transaction scope. Local persistence is only complete after the storage layer sends a transaction confirmation receipt confirming the complete commit of the transaction. After this, the original data packets are always completely retained in the local persistent storage medium without being deleted, trimmed, or overwritten. Only after the remote virtual simulation platform returns the final platform receipt and confirms that the data packets are idempotently stored is the archiving or deletion process performed according to the preset archiving and cleanup strategy.
[0011] Step S2: Obtain and analyze the historical service dataset of XR terminals across the entire domain, and combine it with normalization processing and fusion mapping to obtain the equivalent relative load characteristics of each XR terminal. Collect the historical service dataset of all XR terminals in the local area network cluster registry. The historical service dataset comes from the operation log records of each historical neighbor reachability detection transaction and proxy hosting transaction. The historical service dataset includes at least the historical transaction response processing time of each XR terminal, the persistence IO time of proxy data packets, CPU sampling sequence, memory usage sampling sequence, IO queue time, transaction success sample count and transaction failure sample count. The upper limit of the performance that a single XR terminal’s hardware can achieve is selected as a reference benchmark. Each historical service record corresponds to a complete neighbor detection or data transmission long-term transaction. The average overall load observation result of the transaction within the limited load statistics window is extracted. The single historical service record is treated as a complete whole and normalized within the device is processed. The average overall load observation result of the transaction is transformed to obtain the overall relative load characteristics of the record relative to the upper limit of the hardware performance of the machine. For all valid historical service records of a single XR terminal after normalization, the arithmetic mean and mode of the samples are statistically analyzed. For all valid historical service records after normalization, statistical measures characterizing the degree of sample discrete fluctuation are calculated. Two sets of configuration boundaries are pre-set, corresponding to the acceptable lower boundary and the acceptable upper boundary of sample discrete fluctuation, respectively. Boundary comparison and contribution ratio calculation are then performed. When the statistical measurement result reaches or exceeds the acceptable upper limit of discrete fluctuation, all contributions are directly allocated to the sample mode level, and the arithmetic mean level is no longer allocated a contribution percentage. When the statistical measurement result reaches or falls below the acceptable lower limit of discrete fluctuation, all contributions are directly allocated to the arithmetic mean level, and the contribution ratio of the sample mode level is no longer allocated. When the statistical measure result lies between the two sets of configuration boundaries, linear interpolation is performed based on the position of the statistical measure result relative to the two sets of configuration boundaries to solve for the contribution ratio corresponding to the arithmetic mean level and the sample mode level respectively; the contribution ratios of the two types are added together to keep the total contribution constant. After solving for the contribution ratio, the arithmetic average level and the sample mode level are multiplied by their respective calculated contribution ratios and then combined to obtain a comprehensive inherent processing background index that characterizes the inherent service carrying capacity of a single XR terminal. Based on the comprehensive inherent processing background index obtained from the XR terminals across the entire domain, a cluster business reference benchmark is constructed. For the business task to be executed, the average overall load observation results of the whole machine obtained from the single business transaction reported by the XR terminal within the limited load statistics window are combined with the comprehensive inherent processing background index corresponding to the terminal for mapping and conversion processing to obtain the equivalent relative load characteristics of the business task applied to the corresponding XR terminal. The mapping and conversion parameters corresponding to the integrated inherent processing background index and the equivalent relative load characteristics are persistently saved to the local area network cluster registry. The integrated inherent processing background index serves as the basic reference for determining the qualification of neighboring machines to respond, and the equivalent relative load characteristics are used to complete the cross-terminal service load comparison. In practical implementation, the hardware specifications of the XR terminals within the local area network cluster are different. The historical service dataset is taken from the operation logs of historical neighbor reachability detection transactions and proxy hosting transactions. Each historical service record corresponds to a complete long-transmission transaction. The average overall load of the transaction within a limited load statistics window is used as the load observation result for that record. Normalization is performed within the device based on the upper limit of the hardware performance of a single XR terminal. After normalization, the arithmetic mean and sample mode are calculated for all valid historical service records. Simultaneously, statistical measures characterizing the degree of discrete fluctuation of the samples are calculated and compared with pre-configured lower and upper limits of discrete fluctuation. When the measurement result reaches or exceeds the upper limit of acceptable discrete fluctuation, the entire contribution is assigned to the mode level of the sample. When it reaches or falls below the lower limit of acceptable discrete fluctuation, the entire contribution is assigned to the arithmetic average level. When it is between the two sets of configuration boundaries, linear interpolation is performed to solve the contribution ratio of the two sets, and the total contribution remains constant. The arithmetic average level, the mode level of the sample, and their respective contribution ratios are multiplied and combined to obtain the comprehensive inherent processing background index. Then, this index is used to perform mapping and conversion on the average overall load observation result of a single business transaction within the limited load statistics window to obtain the equivalent relative load characteristic. The mapping and conversion parameters corresponding to the comprehensive inherent processing background index and the equivalent relative load characteristic are persistently saved to the local area network cluster registry.
[0012] Step S3: Construct a neighbor reachability detection request message based on the unique identifier of the source device of the XR terminal with the source fault. The XR terminal that receives the message combines the sliding time window load analysis and the conversion of the inherent processing background index to obtain the equivalent relative load characteristics, and performs a joint determination of the response qualification. After the XR terminal with the source fault completes the persistence of the local simulation experiment data packet transaction, it reads the unique identifier of the source device on the local machine, and binds the unique identifier of the source session corresponding to this XR simulation experiment, the resource requirement metadata of the simulation data packet to be uploaded, the signature-protected source end return address information, and the unique request identifier. It then assembles and generates a neighbor reachability detection request message and broadcasts the neighbor reachability detection request message to logical neighboring machines in the local area network. The unique identifier of the source device is carried in the message header, and the unique identifier of the source session, the resource requirement metadata of the simulation data packet to be uploaded, the signature-protected source end return address information, and the unique request identifier are carried in the message payload. Any XR terminal within the local area network that receives a neighbor reachability probe request message first performs a basic validity check on the message. If the message fails the check, it is discarded and no further processing is performed. If the message passes the check, a sliding time analysis window is opened with itself as the analysis object. Within the sliding time analysis window period, the local machine load-related sampling data is continuously collected. Based on the sampling sequence within the window, the window average overall machine load and load fluctuation amplitude are calculated respectively. The system reads the pre-stored comprehensive inherent processing background index in the local area network cluster registry, combines it with the currently calculated window average overall system load, and maps and converts it to obtain the equivalent relative load characteristics corresponding to this service. At the same time, it parses and extracts the resource requirement metadata of the simulated data packet to be uploaded from the payload of the neighbor reachability probe request message that has passed the legality verification. This resource requirement metadata represents the overall system resource overhead corresponding to this service. Based on the equivalent relative load characteristics, load fluctuation amplitude, and the resource requirement metadata corresponding to this service, a joint determination of response eligibility is made. The judgment criteria include: First, the equivalent relative load characteristics obtained by mapping and conversion do not exceed the allowable load limit corresponding to the business; Second, the load fluctuation amplitude calculated by the sliding time window does not exceed the preset fluctuation threshold; Third, the remaining resources of the local machine are evaluated by combining the equivalent relative load characteristics. The remaining resources of the local machine can cover the resource overhead of the current upload service represented by the resource requirement metadata. If all three conditions are met simultaneously, the XR terminal is deemed to be qualified to respond; if any one condition is not met, the terminal is deemed not qualified to respond. The upper-layer business logic of this XR terminal will not process the broadcast request message or generate a response message. Other XR terminals in the local area network can still receive the reachability probe request message through the local area network broadcast mechanism and execute their respective judgment processes. Each XR terminal within the local area network (LAN) independently completes the response qualification determination; XR terminals qualified to respond generate response messages and, based on the source return address information carried in the message, unicast response messages back to the source-failed XR terminal; after sending a neighbor reachability probe request message, the source-failed XR terminal starts a preset waiting time. If a response message is received within the preset waiting time, the probe is considered successful; if no response message is received after the preset waiting time expires, the neighbor reachability probe is considered to have failed; wherein, the preset waiting time is configured based on the message round-trip latency, the number of online XR terminals in the LAN, and the size of the data packets to be transmitted, as statistically analyzed by the cluster edge gateway node; the cluster edge gateway node continuously collects historical message interaction latency and node load assessment time within the LAN, outputs duration parameters adapted to the current cluster status, and sends them to the source-failed XR terminal; In practical implementation, after the source XR terminal completes the local transaction persistence of the simulation experiment data packet, it reads the unique identifier of its own source device, binds the unique identifier of the source session, the resource requirement metadata of the simulation data packet to be uploaded, the source end return address information protected by signature, and the unique request identifier, and assembles it into a neighbor reachability probe request message, which is then broadcast to logical neighboring machines in the local area network. Any XR terminal in the local area network that receives this message first performs a basic validity check on the message. If the check fails, the message is discarded directly. If the check passes, a sliding time analysis window is opened, and the local machine load sampling data is continuously collected within the window period. The system calculates the average overall load and load fluctuation of the entire machine within a window, then reads the pre-stored comprehensive inherent processing background index in the local area network cluster registry, and combines it with the average overall load of the entire machine within a window to obtain the equivalent relative load characteristics. At the same time, it extracts the resource requirement metadata of the simulated data packet to be uploaded from the packet payload parsing. Only when the three judgment conditions are met simultaneously—that the equivalent relative load characteristics do not exceed the allowed load limit corresponding to the service, that the load fluctuation amplitude does not exceed the preset fluctuation threshold, and that the remaining resources of the entire machine can cover the resource overhead of the service to be uploaded represented by the resource requirement metadata—can the XR terminal be qualified to respond and process.
[0013] Step S4: XR terminals that are qualified to process responses, as determined by the joint assessment of response eligibility, generate response messages and perform cross-comparison of historical responses to filter response messages; The joint assessment of response eligibility determines which XR terminals are qualified to respond. Based on the assessment result, the XR terminal performs reachability detection on the remote virtual simulation platform, calculates the remaining available resources of the terminal that can be used for response and proxy services, and generates a neighboring terminal reachability detection response message. The neighboring terminal reachability detection response message includes a unique request identifier, a unique source device identifier, the platform reachability detection result, the terminal's comprehensive inherent processing background index, and the amount of remaining available resources of the terminal. Before sending a neighbor reachability probe response message to the XR terminal with the source failure, the local response cache is queried to retrieve historical response records corresponding to the same source device unique identifier and the same request unique identifier. The remaining available resources carried in the response message to be sent are cross-compared with the remaining available resources carried in the response messages that have been returned by the local machine in response to this probe request in the past. If the remaining available resources in the response message to be sent are worse than the remaining available resources in the historical records, the response is determined to be a quality degradation response message, and the neighbor reachability probe response message to be sent is not sent to the XR terminal with the source failure. If the cross-comparison is successful, the neighbor reachability probe response message is sent to the XR terminal with the source failure. In practice, an XR terminal that has been jointly determined to be qualified to process responses will first perform a reachability probe on a remote virtual simulation platform before generating a reachability probe response message. This will allow the terminal to count the remaining available resources that can be used for response and proxy transmission services. During the transmission process, the terminal will query its local response cache and retrieve historical response records that share the same source device unique identifier and the same request unique identifier as the current probe request. The remaining available resources carried in the current response message to be sent will be cross-compared with the remaining available resources carried in the historical returned response messages. If the current remaining available resources are inferior to those of the historical returned records, the response message will be deemed a quality degradation response message and will not be sent to the faulty XR terminal. Only if the cross-comparison passes will the response message be sent. The response message will carry the request unique identifier, the source device unique identifier, the platform reachability probe result, the terminal's inherent comprehensive processing background index, and the terminal's current remaining available resources.
[0014] Step S5: The XR terminal with the source fault performs a completeness check on the response message and cross-confirms the fault classification with the neighboring machine, and performs disaster recovery backhaul based on the fault classification result. The source failure XR terminal experienced a failure in the service communication link between itself and the remote virtual simulation platform, but its ability to access the local area network remained normal. The reachability probe request messages from neighboring machines and the simulation experiment data packets to be transmitted were transmitted between the source failure XR terminal and the neighboring machines via the local area network, without passing through the remote virtual simulation platform, and were not affected by the aforementioned service communication link failure. The neighboring machines qualified to respond had completed the reachability probe of the remote virtual simulation platform, and the service communication link between the neighboring machines and the remote virtual simulation platform remained normal, enabling them to undertake the data upload task to the remote virtual simulation platform. After sending a neighbor reachability probe request message, the source fault XR terminal starts a preset waiting time. During the preset waiting time, it continuously listens for and receives neighbor reachability probe response messages from each neighbor. When the preset waiting time expires, it stops receiving neighbor reachability probe response messages. At this time, the source fault XR terminal may not receive, receive one or more neighbor reachability probe response messages. Based on all received valid neighbor reachability probe response messages, cross-confirmation of neighboring devices is performed to complete fault classification. The fault classification results include four states: local link available, local link abnormal, platform or public link abnormal, and undetermined. When the reachability detection result of the local machine to the remote virtual simulation platform indicates that the local link is reachable, it is determined that the local link is available; If the local machine is unreachable but the platform reachability detection result of at least one valid neighboring machine indicates that it is reachable, it is determined that the local machine link is abnormal; When the local machine is unreachable and all valid neighboring machines that have reached the preset confirmation threshold are unreachable, it is determined that the platform or public link is abnormal. If the received valid neighbor reachability probe response messages are insufficient to support the above determination, the state is determined to be uncertain. Execute the corresponding disaster recovery backhaul logic based on the fault classification results obtained from cross-confirmation with neighboring machines: When a local link anomaly is detected, if the source fault XR terminal receives multiple reachability probe response messages from neighboring machines, it selects the optimal neighbor as the accepting neighbor based on the platform reachability probe results carried in each neighboring machine reachability probe response message, the local integrated inherent processing background index, and the current remaining available resources of the local machine. The optimal accepting neighbor is the candidate XR terminal with the highest current remaining available resources of the local machine. If only one neighboring machine reachability probe response message is received, that neighbor is directly selected as the accepting neighbor. The simulation experiment data packet is sent to the accepting neighbor via the local area network. The accepting neighbor persists the simulation experiment data packet and its relay status transactionally, generates a takeover receipt, and returns it to the source fault XR terminal. The source fault XR terminal verifies the acceptance. After receiving the receipt, the original simulation experiment data packet is retained. The receiving neighboring machine, maintaining a normal business communication link with the remote virtual simulation platform, uploads the simulation experiment data packet to the remote virtual simulation platform. The remote virtual simulation platform verifies the source, integrity, resource version, fragment sequence number, and idempotency flag of the simulation experiment data packet. If the verification passes and the idempotency flag has not yet been committed, the data packet is transacted into the database and a final platform receipt is generated. This receipt is then returned to the source fault XR terminal via the receiving neighboring machine or the recovered direct link. After verifying the final platform receipt, the source fault XR terminal marks the simulation experiment data packet as already in the database and performs archiving or deletion processing according to the preset archiving and cleanup strategy. If the final platform receipt times out, the source fault XR terminal returns to the pending transmission state and reselects the transmission path using the original idempotency flag. When a platform or public link is determined to be abnormal, a local delayed backhaul strategy is implemented. The simulation experiment data packets are kept in the local persistent storage medium. The retesting cycle is set according to the fault type. A longer backoff time is used for platform or public link abnormalities, and a shorter retesting time is used when the status cannot be determined. The local link status detection is re-initiated. When the local link is determined to be available again, the XR terminal that caused the source fault directly transmits the simulation experiment data packets to the virtual simulation platform, and uploads them according to the fragment sequence number. The entire process combines idempotent verification logic to avoid duplicate data packet submission, and completes the disaster recovery backhaul processing of XR simulation experiment data. In practical implementation, after the XR terminal with the source fault sends a neighbor reachability probe request message, a preset waiting time is initiated. Upon expiration of the preset waiting time, reception of neighbor reachability probe response messages ceases. In reality, no response message may be received, one response message may be received, or multiple response messages may be received. Based on all received valid neighbor reachability probe response messages, neighbor cross-confirmation is performed, resulting in four fault classifications: local link available, local link abnormal, platform or public link abnormal, and undetermined. If the local link is abnormal and multiple response messages are received, the candidate XR terminal with the highest remaining available resources is selected as the receiving neighbor. If only one response message is received, that neighbor is directly selected as the receiving neighbor. Simulation experiments are conducted. The data packet is sent to the receiving neighboring machine via the local area network. After the receiving neighboring machine performs transactional persistence on the simulation experiment data packet and its relay status, it returns a takeover receipt. Subsequently, it is uploaded to the remote virtual simulation platform. The platform completes the verification of source, integrity, resource version, fragment sequence number, and idempotency flag, and returns a final receipt after transactional storage. If the platform's final receipt times out, the XR terminal with the source failure will reselect the backhaul path using the original idempotency flag. If the platform or public link is determined to be abnormal, the simulation experiment data packet is kept in the local persistent storage medium and a local delayed backhaul is performed. When the local link is restored and available, the XR terminal with the source failure directly transmits the simulation experiment data packet to the virtual simulation platform. Idempotency verification is performed throughout the process to avoid duplicate submission of data packets.
[0015] It will be apparent to those skilled in the art that the present invention is not limited to the details of the exemplary embodiments described above, and that the invention can be implemented in other specific forms without departing from the spirit or essential characteristics of the invention. Therefore, the embodiments should be considered in all respects as exemplary and non-limiting, and the scope of the invention is defined by the appended claims rather than the foregoing description. Thus, it is intended that all variations falling within the meaning and scope of equivalents of the claims be included within the present invention.
Claims
1. A disaster recovery and backhaul method for XR simulation experimental data based on edge caching, characterized in that: The method for disaster recovery and backhaul of XR simulation experimental data includes: Based on local link events, persist XR simulation experiment data packets for local transactions; The historical service dataset of XR terminals across the entire domain is obtained and analyzed. The equivalent relative load characteristics of each XR terminal are obtained by combining normalization processing and fusion mapping. The XR terminal with the source fault constructs a neighbor reachability probe request message and performs probe, and jointly determines the eligibility to respond based on the equivalent relative load characteristics. XR terminals that are qualified to process responses, as determined by the joint assessment of response eligibility, generate response messages and perform cross-comparison of historical responses to filter response messages; The XR terminal with the source fault performs a completeness check on the response message and cross-confirms the fault classification with the neighboring machine, and performs disaster recovery backhaul based on the fault classification result.
2. The method for disaster recovery and backhaul of XR simulation experimental data based on edge caching according to claim 1, characterized in that: The method for persistently storing XR simulation experiment data packets based on local link events includes: When the source XR terminal detects a local link failure and is unable to directly connect to the remote virtual simulation platform, it performs atomic transaction persistent writing on the XR simulation experiment data packets that have not yet been uploaded, along with the unique idempotent key of the data packets, the unique identifier of the source device, the unique identifier of the source session, and the fragmentation index metadata. After obtaining the transaction confirmation receipt, the original simulation experiment data packet is completely preserved locally, and the neighbor machine search and detection process is executed.
3. The method for disaster recovery and backhaul of XR simulation experimental data based on edge caching according to claim 1, characterized in that: The equivalent relative load characteristics of each XR terminal obtained by combining normalization processing and fusion mapping include: Collect the historical service dataset of XR terminals across the entire domain within the local area network cluster registry. Select the upper limit of the overall hardware performance of a single XR terminal in the historical service dataset as the reference benchmark. Each historical service record corresponds to a complete neighbor detection and data transmission long-term transaction. Extract the average overall load observation result of the transaction within the limited load statistics window. Perform device-wide normalization transformation as a complete whole on a single historical service record to obtain the overall relative load characteristics of the machine relative to the upper limit of the machine's overall hardware performance. After statistical analysis of all valid historical service records of a single XR terminal after normalization, the data is compared with the pre-configured lower limit boundary and upper limit boundary of discrete fluctuation. When the statistical measurement result reaches or exceeds the acceptable upper limit of discrete fluctuation, all contributions are allocated to the sample mode level; When the statistical metric reaches or falls below the acceptable lower bound of discrete fluctuation, all contributions are allocated to the arithmetic mean. When the statistical measure result lies between two sets of configuration boundaries, linear interpolation is performed based on the position of the statistical measure relative to the boundary to obtain the contribution ratio corresponding to the arithmetic mean level and the sample mode level.
4. The method for disaster recovery and backhaul of XR simulation experimental data based on edge caching according to claim 3, characterized in that: The method of combining normalization processing and fusion mapping to obtain the equivalent relative load characteristics of each XR terminal also includes: Based on the contribution ratio obtained from the solution, a weighted fusion calculation is performed on the arithmetic average level and the sample mode level to obtain a comprehensive inherent processing background index that characterizes the inherent service carrying capacity of a single XR terminal. Based on the comprehensive inherent processing background index of the entire XR terminal, a cluster service reference benchmark is constructed. Combined with the observation results of the average overall load of the business tasks to be executed reported by the XR terminal, cross-terminal mapping and conversion are performed through the comprehensive inherent processing background index of the corresponding terminal to obtain the equivalent relative load characteristics of the business task acting on the corresponding XR terminal. The persistent storage integrates the inherent processing background indicators and the mapping and conversion parameters corresponding to the equivalent relative load characteristics to the local area network cluster registry.
5. The method for disaster recovery and backhaul of XR simulation experimental data based on edge caching according to claim 1, characterized in that: The joint determination of response eligibility based on equivalent relative load characteristics includes: The XR terminal with the source fault binds the unique identifier of the local source device, the unique identifier of the source session, the resource requirement metadata of the simulation data packet to be uploaded, the source end return address information, and the unique request identifier, generates a neighbor reachability detection request message, and broadcasts it to logical neighbors in the local area network. The XR terminal that receives the neighbor reachability detection request message makes a determination through the sliding time analysis window, and calculates the average overall load and load fluctuation amplitude of the sliding time analysis window by combining the local whole machine load sampling data; Based on the inherent processing background index of the machine and the average overall load of the machine, the equivalent relative load characteristics are obtained by mapping and conversion. The qualification for response is jointly determined based on the equivalent relative load characteristics, load fluctuation amplitude, and resource requirement metadata of the simulation data packet to be uploaded. The XR terminal that is deemed qualified to respond generates a response message and unicasts the response message back to the source faulty XR terminal based on the source return address information.
6. The method for disaster recovery and backhaul of XR simulation experimental data based on edge caching according to claim 1, characterized in that: The process of performing historical response cross-comparison to filter response messages includes: The XR terminal with response qualification is determined by joint judgment of response qualification, the reachability detection of the remote virtual simulation platform is carried out, the remaining available resources of the local machine that can be used for response and proxy services are counted, and a neighboring machine reachability detection response message is generated. Before sending a neighbor reachability probe response message to the XR terminal with the source fault, query the local response cache to retrieve historical response records corresponding to the unique identifier of the same source device and the unique identifier of the same request. The remaining available resources carried in the current response message to be sent are cross-compared with the remaining available resources carried in the historical response messages returned by this machine; based on the cross-comparison result, a neighbor reachability probe response message is sent to the XR terminal with the source fault.
7. The method for disaster recovery and backhaul of XR simulation experimental data based on edge caching according to claim 1, characterized in that: The step of performing disaster recovery backhaul based on the fault classification results includes: After the XR terminal with the source fault sends a neighbor reachability detection request message, it starts a preset waiting time and receives neighbor reachability detection response messages within the preset waiting time. When the preset waiting time expires, it stops receiving messages. Based on the received valid neighbor reachability probe response messages, perform neighbor cross-confirmation to obtain fault classification results: local link available, local link abnormal, platform or public link abnormal, and indeterminate. Based on the fault classification results, one of the following methods is selected for processing: taking over the neighboring machine to perform proxy upload, performing local delayed return, or performing direct upload, and idempotency verification is performed throughout the process.
8. A disaster recovery and backhaul system for XR simulation experimental data based on edge caching, which is applied to the disaster recovery and backhaul method for XR simulation experimental data based on edge caching as described in any one of claims 1-7, characterized in that: The XR simulation experiment data disaster recovery and backhaul system includes a local transaction persistence module, a load feature parsing module, a neighboring machine response qualification determination module, and a fault classification disaster recovery and backhaul module. The local transaction persistence module is used to complete the local transaction persistence storage of XR simulation experiment data packets based on local link events; the load feature parsing module is used to obtain the historical service dataset of the entire XR terminal and perform mapping processing to obtain the equivalent relative load features of each XR terminal; the neighbor response qualification determination module is used for the source fault XR terminal to construct a neighbor reachability detection request message to detect and compare and filter response messages; the fault classification disaster recovery backhaul module is used to perform disaster recovery backhaul based on the fault classification results.
9. The XR simulation experiment data disaster recovery and backhaul system based on edge caching according to claim 8, characterized in that: The local transaction persistence module includes a link failure monitoring unit and an atomic transaction writing unit; the link failure monitoring unit is used to monitor and identify local link failure events of the source XR terminal; the atomic transaction writing unit is used to perform atomic transaction persistence writing. The load feature analysis module includes a historical service data acquisition unit and an equivalent load conversion unit; the historical service data acquisition unit is used to collect the historical service dataset of XR terminals in the local area network cluster registry; the equivalent load conversion unit is used to solve the comprehensive inherent processing background index and the equivalent relative load feature.
10. The XR simulation experiment data disaster recovery and backhaul system based on edge caching according to claim 8, characterized in that: The neighboring machine response qualification determination module includes a probe message sending and receiving unit and a historical response cross-comparison unit; the probe message sending and receiving unit is used to generate and broadcast neighboring machine reachability probe request messages and determine response qualification; the historical response cross-comparison unit is used to control the sending of neighboring machine reachability probe response messages; The fault classification and disaster recovery backhaul module includes a response message monitoring unit and a disaster recovery backhaul execution unit; the response message monitoring unit is used for cross-confirmation with neighboring machines to obtain fault classification results; the disaster recovery backhaul execution unit is used to select the corresponding processing method to execute backhaul based on the fault classification results.
