Cloud edge terminal cooperative intelligent cabinet management system

Abstract

A cloud-edge-device collaborative intelligent cabinet management system includes a cloud server, an edge computing gateway, a low-level embedded hardware execution unit, and an integrated multi-sensor anti-tampering module. The integrated multi-sensor anti-tampering module is independently fixed in a concealed location within the cabinet and connects to the edge computing gateway via an encrypted independent communication link. It monitors physical tampering behavior within the cabinet in real time and triggers multi-level alarms and emergency locking responses, forming an integrated online and offline physical security protection system. This invention achieves highly reliable dual protection and communication fault tolerance mechanisms through a three-level cloud-edge-device collaborative architecture and offline self-healing architecture, combined with end-to-end remote seamless upgrade technology.

Description

Technical Field

[0001] This application relates to the field of intelligent cabinet technology, and more specifically, to a cloud-edge-device collaborative intelligent cabinet management system. Background Technology

[0002] In the field of intelligent cabinet management, traditional system architectures primarily rely on cloud servers for centralized decision-making and control, with the main control terminal acting as an intermediate layer responsible for interacting with the cloud and coordinating the operation of underlying execution units. However, this design's strong dependence on network connectivity reveals significant shortcomings in real-world applications. When network connectivity is interrupted or cloud services malfunction, the system completely loses its authentication capabilities, preventing users from performing basic operations such as opening and closing doors. This forces critical business processes to be interrupted, severely impacting business continuity, especially in remote areas or environments with weak network coverage. Furthermore, underlying control hardware, such as the microcontroller units in the lock control board, generally lacks direct network connectivity. Firmware updates must be performed manually on-site, requiring technicians to disassemble the cabinet casing and connect dedicated programming equipment to write the program. This process is not only time-consuming and labor-intensive but also significantly increases maintenance costs and system downtime. For large-scale distributed cabinet networks, the problem of low maintenance efficiency is particularly prominent. More seriously, when the system operates unattended or in industrial environments with strong electromagnetic interference, the main program is prone to crashing due to software malfunctions or serial communication deadlocks, resulting in a halted state. Lacking an effective self-recovery mechanism, the equipment remains unresponsive to external commands for extended periods, requiring manual intervention to restart. This not only reduces the system's reliability and availability but may also pose safety hazards. Therefore, existing technologies urgently need improvement to address these issues. Summary of the Invention

[0003] The purpose of this application is to provide a cloud-edge-device collaborative intelligent cabinet management system, which has a three-level cloud-edge-device collaboration and offline self-healing architecture, and combines full-link remote seamless upgrade technology to achieve highly reliable dual protection and communication fault tolerance mechanism.

[0004] A cloud-edge-device collaborative intelligent cabinet management system is characterized by comprising a cloud server, an edge computing gateway, a low-level embedded hardware execution unit, and an integrated multi-sensor anti-tampering module. The edge computing gateway is an Android all-in-one machine with a built-in local database module, biometric recognition module, OTA management module, system protection module, and protocol conversion module. The edge computing gateway is communicatively connected to the cloud server, the low-level embedded hardware execution unit, and the integrated multi-sensor anti-tampering module. The local database module is used to cache and store user permission data, biometric data, and operation log data, and works with the biometric recognition module to achieve local identity verification and business decision-making in offline mode. The OTA management module uses a preset cross-level file transfer protocol. The firmware package distributed from the cloud is divided into packets and transmitted to the underlying embedded hardware execution unit. This, combined with the underlying bootloader module, enables remote, seamless firmware upgrades for the network-free MCU. The system protection module employs a dual-protection mechanism with mutual wake-up and a JobScheduler periodic check mechanism to achieve automatic restart and self-healing after a main program crash. It also incorporates a protocol conversion module with CRC verification and timeout retransmission mechanisms to construct a dual hardware and software communication fault-tolerant system. The integrated multi-sensor anti-tampering module is independently fixed in a concealed location within the cabinet. Connected to the edge computing gateway via an encrypted independent communication link, it monitors physical tampering behavior in the cabinet in real time and triggers multi-level alarms and emergency locking responses, forming an integrated offline and online physical security protection system.

[0005] The biometric recognition module has a built-in face recognition algorithm engine that can extract local features from the collected face images and perform offline comparison with the feature library stored in the local database module. After the comparison is successful, the door opening command is generated directly based on the local permission data without the need for cloud authorization.

[0006] The local database module uses the lightweight SQLite database to achieve persistent storage of user permissions, biometrics, operation logs and security event data. During network outages, offline operation records and tampering alarm information are cached locally first, and automatically synchronized to the cloud server in batches after the network is restored, ensuring data consistency between the cloud and the edge.

[0007] The cross-level file transfer protocol is adapted to serial communication scenarios. The OTA management module divides the firmware package into packets of fixed size and sends them to the underlying embedded hardware execution unit one packet at a time via the serial port. Each time the underlying MCU receives a data packet, it writes it to the memory backup area and completes the verification feedback. After all data packets pass the verification, the Bootloader module updates the boot flag and automatically restarts, moving the new firmware from the backup area to the running area to complete the upgrade. If the upgrade fails, it can automatically roll back to the original firmware version.

[0008] The system protection module includes a dual protection mechanism: a local device security protection mechanism and a remote device security protection mechanism. The two protection mechanisms monitor and wake each other up. At the same time, a periodic wake-up check task is set through the JobScheduler mechanism to actively monitor the running status of the main business program. If the main program is unresponsive or crashes, a restart operation is immediately performed to achieve stable 24 / 7 unattended operation. The protocol conversion module encapsulates cloud JSON instructions into custom RS485 data frames. The data frame includes a frame header, device address, function code, data length, and CRC checksum. After the instruction is sent, a preset timeout waiting period is set. If no hardware response is received, a retransmission is automatically triggered. If the maximum number of retransmissions is exceeded, the device is marked as faulty and an alarm is reported.

[0009] The integrated multi-sensor anti-tampering module is encapsulated in a tamper-proof and robust shell. It integrates a vibration acceleration sensor, a magnetic contact Hall sensor array, an internal ambient light sensor, and a module self-destruct anti-tampering switch. The sensors work together to collect physical status data of the cabinet. The anti-tampering module has a built-in anomaly detection algorithm to distinguish between normal environmental disturbances and malicious physical attacks. When anomalies such as prying, forced entry, or module disassembly are detected, a high-priority tampering alarm signal is immediately generated and transmitted to the edge computing gateway via an encrypted serial port.

[0010] Upon receiving a tampering alarm signal, the edge computing gateway executes a multi-level linkage response: immediately triggering a local high-decibel audible and visual alarm for deterrence; when the network is normal, simultaneously pushing the alarm information to the cloud and notifying the administrator via SMS and APP push; when the network is interrupted, encrypting and storing the alarm data in the local database; and when serious tampering is detected, issuing an emergency lock command to the underlying lock control unit, temporarily locking all departments, which can only be unlocked with administrator authorization.

[0011] The underlying embedded hardware execution unit includes a lock control board and an MCU controller. The MCU has a built-in bootloader module that supports IAP function. After receiving a valid door opening command, the lock control unit drives the electromagnetic lock relay to act and sets a delay power-off mechanism. After a short delay, the coil power supply is automatically stopped to prevent the coil from overheating and being damaged. At the same time, the lock body status is fed back to the edge computing gateway in real time.

[0012] The edge computing gateway communicates with the cloud server via Ethernet or a wireless module, connects to the underlying embedded hardware execution unit via a USB-to-serial port or RS485 interface, and connects to the integrated multi-sensor anti-tampering module via an encrypted bus. The three types of communication links are independent of each other to avoid signal interference and malicious interception.

[0013] An offline working method for a cloud-edge-device collaborative intelligent cabinet management system includes the following steps: S1. When the network is interrupted or the cloud service is abnormal, the user triggers a cabinet retrieval operation, and the edge computing gateway calls the camera to capture a facial image; S2. The biometric recognition module extracts image feature values ​​and performs feature comparison in the local SQLite database to verify the user's identity; S3. After successful identity verification, the local user permission data is directly queried. If the permissions are in accordance with the rules, the door opening command is sent to the underlying lock control unit through the protocol conversion module to execute the offline door opening operation; S4. The entire offline operation log and device status data are cached in the local database. After the network is restored, the edge computing gateway automatically synchronizes the cached data to the cloud server to complete the data closure loop; S5. In the offline state, the integrated multi-sensor anti-tampering module continuously monitors the physical security status. When tampering is detected, the local system immediately triggers an audible and visual alarm and an emergency lock. The alarm data is synchronized to the local cache and reported to the cloud after reconnecting to the network.

[0014] The technical highlights of this invention are as follows:

[0015] 1. Cloud-Edge-Device Three-Level Collaboration and Offline Self-Healing Architecture: Innovatively introducing an edge computing layer into Android smart terminals, by building a local database and a face recognition engine, the core business logic (permission verification, identity recognition) in the cloud is moved to the edge, breaking the technical bottleneck of traditional smart cabinets that "paralyze when the network is disconnected".

[0016] 2. Over-the-air (OTA) technology: A cross-level file transfer protocol from the cloud to the embedded underlying layer was designed. Using the Android terminal as a relay gateway, firmware updates and verifications were performed on microcontrollers (MCUs) that do not have network connectivity at the underlying level, solving the pain point of traditional devices requiring manual on-site flashing.

[0017] 3. Highly reliable dual protection and communication fault tolerance mechanism: At the software layer, dual services are used to wake up the daemon process. At the communication layer, a CRC check and automatic retransmission mechanism based on a custom protocol is established, which builds a dual "watchdog" of software and hardware, which significantly improves the anti-interference ability and stability of the system in an unattended environment.

[0018] This application achieves the following significant effects by introducing an Android gateway with edge computing capabilities and a custom hardware communication protocol into the smart cabinet:

[0019] 1. Zero service interruption: Even if users are in environments with poor network signal, such as basements, or during sudden network outages, they can still unlock and retrieve items in milliseconds using local facial recognition. Once the network is restored, the data is automatically synchronized, greatly improving the user experience and business continuity.

[0020] 2. Maintenance costs reduced by more than 90%: Maintenance personnel can remotely complete the underlying firmware repair and function iteration of hundreds of devices by issuing commands with one click through the cloud without having to go to the site, which greatly shortens the fault response time and maintenance cycle.

[0021] 3. Industrial-grade stability: Effectively solves the problem of equipment "fake death" caused by software crashes or on-site electromagnetic interference, realizes long-term stable operation of equipment 24 / 7, and reduces management loopholes caused by equipment failure. Detailed Implementation

[0022] The technical solutions in this application are described clearly and completely below. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0023] This application provides a cloud-edge-device collaborative intelligent cabinet management system, which aims to solve the problems of traditional intelligent cabinet systems failing to work properly in offline states, difficulties in upgrading underlying hardware firmware, and insufficient system stability.

[0024] For ease of understanding, the following explains some key terms in this embodiment:

[0025] The cloud server, as the central management node of the entire system, is responsible for global data storage, user permission management, remote command issuance, and advanced business logic processing.

[0026] An edge computing gateway, as an intelligent terminal deployed in a server rack, undertakes core functions such as local data processing, protocol conversion, security protection, and offline business decision-making. This edge computing gateway is designed as an Android all-in-one machine, possessing powerful local computing capabilities and a rich set of interfaces.

[0027] The underlying embedded hardware execution unit is responsible for the physical operation control of the cabinet, such as driving electromagnetic locks and collecting door status sensor data. It is usually composed of a microcontroller (MCU) and related peripheral circuits.

[0028] The integrated multi-sensor anti-tampering module is independently installed in a concealed location inside the cabinet to monitor the physical security status of the cabinet in real time, such as abnormal behaviors like vibration, opening and closing, and disassembly.

[0029] The local database module, built into the edge computing gateway, is used to cache and store user permission data, biometric data, and operation log data to support local authentication and business decisions in the event of network outages.

[0030] The biometric identification module, which is built into the edge computing gateway, is used to collect users' biometric information and perform local comparison to complete identity verification.

[0031] The OTA management module, built into the edge computing gateway, is responsible for receiving firmware update packages from the cloud server and transmitting them securely and reliably to the underlying embedded hardware execution unit to achieve remote firmware upgrades.

[0032] The system protection module, which is built into the edge computing gateway, is used to continuously monitor the running status of the main business program and automatically perform recovery operations when an anomaly is detected, so as to ensure the stability and availability of the system.

[0033] The protocol conversion module, built into the edge computing gateway, is responsible for converting data of different communication protocol formats to each other, ensuring effective communication between the cloud, the edge, and the underlying hardware.

[0034] The bootloader module, which is built into the microcontroller of the underlying embedded hardware execution unit, is responsible for loading the operating system or application when the device starts up and provides boot functionality for firmware updates.

[0035] This application discloses a cloud-edge-device collaborative intelligent rack management system, designed to achieve efficient and reliable rack management. The system includes a cloud server, an edge computing gateway, a low-level embedded hardware execution unit, and an integrated multi-sensor anti-tampering module. The cloud server can be deployed in a remote data center for centralized management and data analysis. The edge computing gateway can be deployed on-site within the rack as a local intelligent processing unit. The low-level embedded hardware execution unit can be a lock control board or a sensor interface board within the rack. The integrated multi-sensor anti-tampering module can be independently installed inside the rack for monitoring physical security.

[0036] This edge computing gateway is designed as an Android all-in-one device, integrating multiple functional modules. For example, it can function as a general-purpose industrial tablet PC, running the Android operating system and implementing functions such as local database management, biometric recognition, OTA management, system protection, and protocol conversion through the installation of corresponding applications. The gateway exchanges data with the cloud server via a standard network interface, connects to the underlying embedded hardware execution unit via a USB interface, and communicates with the integrated multi-sensor tamper-proof module via a universal serial interface.

[0037] This local database module is used to cache and store user permission data, biometric data, and operation log data. For example, this local database module can be a file system-based storage solution, storing various types of data as structured text files on the storage medium of the edge computing gateway. In the event of network interruption or cloud service anomalies, the biometric recognition module can directly read these local files for authentication and execute business decisions such as opening doors based on the permission information recorded therein.

[0038] This OTA management module is responsible for remote firmware upgrades. For example, it can use an HTTP-based file download method to obtain a complete firmware package from a cloud server. Once obtained, the firmware package is transmitted to the underlying embedded hardware execution unit (HUMU). The bootloader module in the HUMU receives the complete firmware package, performs integrity verification, and completes the upgrade by overwriting the existing firmware.

[0039] The system's guardian module aims to ensure the stable operation of the main program. For example, this guardian module could be a simple process monitoring program that periodically checks the running status of the main business program. When it detects that the main program process is missing, the guardian program will attempt to restart the main program. Meanwhile, the protocol conversion module can employ a simple checksum mechanism during data transmission and perform a fixed number of retransmission attempts in case of communication failure to handle occasional communication errors.

[0040] This integrated multi-sensor anti-tampering module is used to monitor the physical security status of the server rack in real time. For example, the anti-tampering module may contain only a simple vibration sensor. When it detects vibration exceeding a preset threshold, it sends an alarm signal to the edge computing gateway via an unencrypted serial communication interface. Upon receiving the alarm, the edge computing gateway can trigger a local buzzer to sound an audible alarm and record the alarm event.

[0041] This application's cloud-edge-device collaborative intelligent cabinet management system, by deploying an Android all-in-one machine as an edge computing gateway at the edge, and incorporating a local database module and a biometric recognition module, enables local authentication and business decision-making even in the event of network interruption, effectively solving the problem of traditional systems being unavailable offline. Simultaneously, the OTA management module, in conjunction with the underlying bootloader module, enables remote, seamless firmware upgrades for underlying hardware without network connectivity, significantly improving maintenance efficiency. Furthermore, the communication fault-tolerance mechanism of the system protection module and protocol conversion module ensures automatic recovery and stable operation of the system under abnormal conditions. The integrated multi-sensor anti-tampering module constructs an integrated offline and online physical security protection system, comprehensively enhancing the security of the cabinet.

[0042] In some of the embodiments described above in this application, although the edge computing gateway is proposed to have a built-in biometric identification module and a local database module in order to achieve local authentication and business decision-making in the offline state, in practical applications, how to ensure the efficiency and security of the biometric identification process, and directly and quickly respond to core business needs (such as door opening operations), while completely eliminating the dependence on cloud authorization, is the key challenge to achieve truly efficient offline operation.

[0043] In this regard, this application further proposes that the biometric recognition module has a built-in face recognition algorithm engine, which can extract local features from the collected face images and perform offline comparison with the feature library stored in the local database module. After the comparison is successful, the door opening command is generated directly based on the local permission data without the need for cloud authorization intervention.

[0044] Specifically, the biometric recognition module is a hardware or software unit responsible for collecting, processing, and recognizing biometric information. Its built-in face recognition algorithm engine refers to the software logic and computational capabilities integrated within the module specifically for processing facial images, extracting facial features, and performing comparisons. This engine can be implemented based on a deep learning model (such as a convolutional neural network, CNN), with a pre-trained model deployed on the module's processor. When the camera captures a facial image, the engine can preprocess the image in real time (such as grayscale conversion, normalization, and liveness detection), and then extract a high-dimensional feature vector through a neural network. This ensures that the face recognition process is completed locally, without uploading the original image or feature data to the cloud, thereby improving recognition speed and data privacy.

[0045] The biometric recognition module can extract features locally from the captured facial images, meaning that the facial image data captured by the camera is directly input into the built-in facial recognition algorithm engine. The engine utilizes its computing power to perform a series of image processing and pattern recognition operations on the image, generating a unique digital feature vector. This process typically includes steps such as face detection, key point localization, and feature encoding, ensuring that the feature extraction process is entirely performed locally on the edge computing gateway, rather than relying on the cloud.

[0046] Furthermore, the extracted facial feature vectors are compared offline with a pre-stored database of legitimate user features in the local database module. The comparison algorithm can be cosine similarity, Euclidean distance, etc., calculating the similarity between the feature to be identified and the features in the database to determine if it belongs to the same person. The local database module uses a lightweight database such as SQLite to store user IDs, permission information, and corresponding facial feature vectors, thus ensuring that the authentication process is completely offline.

[0047] Once facial recognition is successful, the biometric identification module queries the local database for permission data associated with the user ID. This permission data defines the racks and time periods the user can access. If the permission check passes, the edge computing gateway immediately generates an opening command that conforms to the communication protocol of the underlying embedded hardware execution unit (especially the lock control unit), enabling rapid and autonomous offline business decisions.

[0048] The entire authentication and authorization decision-making chain is completed within the edge computing gateway, without any real-time interaction with the cloud server, meaning no cloud authorization intervention is required. This means that even if the cloud server fails or the network is completely interrupted, the system can still provide door opening services normally. The permission data stored in the local database module is the latest version pre-synchronized from the cloud, ensuring the accuracy of offline decisions, thus emphasizing the system's autonomy and high availability, especially in the event of network instability or interruption.

[0049] Through the above technical solution, the biometric recognition module incorporates a built-in facial recognition algorithm engine, enabling it to extract features from captured facial images locally and perform offline comparison with the feature library stored in the local database module. Upon successful comparison, it directly generates an opening command based on local access control data, without requiring cloud authorization. This allows the system to quickly and accurately complete user authentication and door opening operations even during network outages or cloud service failures, significantly improving the system's offline availability and response speed. Simultaneously, local processing of biometric data effectively protects user privacy and data security, providing the smart cabinet with stable and reliable access control management capabilities.

[0050] In some embodiments described above, the local database module built into the edge computing gateway is used to cache and store user permission data, biometric data, and operation log data. This, combined with the biometric recognition module, enables local authentication and business decision-making in offline states. The biometric recognition module has a built-in face recognition algorithm engine that can extract local features from captured face images and perform offline comparison with the feature library stored in the local database module. Upon successful comparison, it directly generates an access control command based on the local permission data, without requiring cloud authorization. However, in practical applications, ensuring the reliable storage of this critical data (including user permissions, biometrics, operation logs, and potential security event data) during network outages, efficient synchronization after network recovery, and ultimately ensuring end-to-end data consistency between the cloud and the edge is a significant challenge in achieving high system reliability and data integrity.

[0051] This application further proposes a specific implementation method for the local database module. The local database module uses the lightweight SQLite database. SQLite is an embedded relational database management system characterized by its lightweight nature, high performance, zero configuration, and ease of integration. It does not require a separate server process, storing the database directly in files, making it ideal for resource-constrained edge device environments. By integrating the SQLite library into the operating system of the edge computing gateway, applications can directly access and manage database files through the SQLite application programming interface (API), thus providing an efficient, reliable, and resource-efficient local data storage solution.

[0052] This local database module provides persistent storage for user permissions, biometric data, operation logs, and security event data. This means that even if the edge computing gateway loses power or restarts, this critical data will not be lost. Specifically, user permission data can be stored in a dedicated permission table, biometric data in a feature table, operation logs in a log table, and security event data generated by the integrated multi-sensor tamper-proof module in an event table. The transaction mechanism and file storage features of the SQLite database ensure reliable writing and saving of this data.

[0053] During network outages, offline operation logs and tampering alarm information are prioritized for local caching. When the edge computing gateway detects a network connection interruption with the cloud server, all newly generated operation logs (e.g., door opening records and authentication attempts in offline mode) and tampering alarm information detected by the integrated multi-sensor anti-tampering module will no longer attempt to upload them to the cloud server immediately. Instead, they will be directly written to a pre-defined offline cache table in the local SQLite database or marked as "not synchronized." This mechanism ensures that all important operation logs and security alarm information can be reliably saved when the network is unavailable, without losing any critical data. Furthermore, priorities can be set to ensure that high-priority security data, such as tampering alarms, are recorded promptly.

[0054] Once the network is restored, the local database module automatically synchronizes to the cloud server in batches. The edge computing gateway's system daemon module or dedicated data synchronization service periodically checks the network connection status. Once the network is restored, this service automatically scans the local SQLite database for all offline operation records and tampering alarms marked as "not synchronized." This data is packaged into batches and uploaded to the cloud server in batches using a preset communication protocol. After successful upload, the corresponding data in the local database is marked as "synchronized" or deleted from the cache table, ensuring that the cloud server can promptly obtain all data generated by the edge during its offline period.

[0055] The above mechanisms ensure end-to-end data consistency between the cloud and the edge. Through offline caching and automatic batch synchronization, combined with data version management or timestamp mechanisms, it is ensured that data copies on the cloud and edge are synchronized at any given time, avoiding data conflicts and information silos. For example, during synchronization, the cloud can handle potential conflicts based on timestamps or data versions, ensuring that the latest data is retained. Simultaneously, the cloud can periodically push the latest permission data and configuration information to the edge to achieve bidirectional synchronization and consistency.

[0056] Through the above technical solution, this application effectively addresses the challenges of reliable storage, data consistency maintenance, and efficient offline data synchronization for critical data (such as user permissions, biometrics, operation logs, and security event data) in a cloud-edge-device collaborative intelligent cabinet management system when network connectivity is interrupted. The local database module uses the lightweight SQLite database, providing the edge computing gateway with efficient, stable, and low-resource-consumption local data storage capabilities. This ensures persistent storage of user permissions, biometrics, operation logs, and security event data, preventing data loss even after power outages or restarts. Particularly during network interruptions, offline operation records and tampering alarm information are prioritized for local caching, which is crucial for ensuring business continuity and complete security event recording when the system is offline. Once the network is restored, the system automatically synchronizes this cached data to the cloud server in batches, ensuring end-to-end data consistency between the cloud and the edge. This mechanism not only avoids the risk of data loss due to network instability but also ensures that the cloud can promptly obtain all operation and security event information from the edge, providing a complete and reliable data foundation for subsequent data analysis, auditing, and decision-making. This significantly improves the reliability, data integrity, and security of the entire intelligent cabinet management system.

[0057] In some embodiments described above, an OTA management module is proposed to transmit firmware packages sent from the cloud to the underlying embedded hardware execution unit in packets using a preset cross-level file transfer protocol. This, in conjunction with the underlying bootloader module, enables remote, seamless firmware upgrades for network-free MCUs. However, in practical applications, especially in scenarios where the underlying embedded hardware execution unit may communicate via non-network methods (such as serial ports), ensuring the reliability and integrity of firmware transmission and the security of the upgrade process, and preventing device startup failure due to transmission interruption or firmware corruption, remains a key technical problem that needs to be solved.

[0058] In this regard, this application further proposes that the cross-level file transfer protocol is adapted to serial communication scenarios. The OTA management module divides the firmware package into packets of fixed size and sends them to the underlying embedded hardware execution unit one packet at a time via the serial port. Each time the underlying MCU receives a data packet, it writes it to the memory backup area and completes the verification feedback. After all data packets pass the verification, the Bootloader module updates the boot flag and automatically restarts, moving the new firmware from the backup area to the running area to complete the upgrade. If the upgrade fails, it can automatically roll back to the original firmware version.

[0059] Specifically, the cross-level file transfer protocol is designed to run efficiently and stably on a serial communication interface. The protocol defines the packet frame format, error detection mechanisms (e.g., checksums or cyclic redundancy check CRC), flow control, and retransmission strategies to accommodate the characteristics of serial communication, such as lower bandwidth and potential noise interference.

[0060] The OTA management module is responsible for processing the firmware package to be upgraded, dividing it into multiple fixed-size data packets. This packet segmentation strategy helps optimize serial port transmission efficiency, simplifies the management of the underlying MCU's receive buffer, and facilitates error detection and recovery during transmission. For example, each data packet can be set to 128 bytes or 256 bytes, with the specific size adjustable according to the MCU's memory resources and serial communication rate.

[0061] The segmented firmware data packets are sent one by one to the underlying embedded hardware execution unit via serial port. Upon receiving each data packet, the underlying MCU writes it to a reserved backup area in its internal memory. Simultaneously, the MCU performs integrity checks on the received data packets, such as calculating a checksum or CRC, and sends the verification result back to the OTA management module. If the verification fails, the OTA management module retransmits the data packet based on the feedback information until it is successfully received and verified.

[0062] The bootloader module will only proceed to the next step after all firmware data packets have been successfully received by the underlying MCU, written to the backup area, and verified. At this point, the bootloader module will update a specific boot flag indicating that a complete and valid firmware to be upgraded exists in the backup area. The system will then automatically reboot.

[0063] After the system restarts, the bootloader module first starts and checks the boot flag. If new firmware is detected waiting for upgrade in the backup area, the bootloader module will completely transfer the new firmware from the backup area to the runtime area, i.e., the MCU's main program storage area, thus completing the firmware upgrade process.

[0064] To ensure the security and reliability of the upgrade process, this application also provides an automatic rollback mechanism for upgrade failures. If an error occurs during firmware transfer, or if the new firmware fails to function correctly on first boot (e.g., determined by watchdog timer timeout or specific status feedback mechanisms), the bootloader module can detect the upgrade failure. In this case, the bootloader module will automatically roll back to the original firmware version, i.e., reload or start the previously working firmware, thereby avoiding the risk of the device becoming "bricked" due to upgrade failure and ensuring the continuous availability of the device.

[0065] Through the above technical solution, the OTA management module can divide the firmware package into fixed-size packets and reliably deliver the firmware to the underlying embedded hardware execution unit packet by packet through a cross-level file transfer protocol adapted to serial communication scenarios. The underlying MCU verifies each received data packet and writes it to the memory backup area, ensuring the integrity of the transmitted data. Only when all data packets have passed verification does the bootloader update the boot flag and automatically restart, moving the new firmware from the backup area to the runtime area, thus achieving atomicity and security in the firmware upgrade process. More importantly, when any abnormality or failure occurs during the upgrade process, the system can automatically roll back to the original firmware version, effectively avoiding the risk of the device failing to boot due to upgrade failure. This greatly improves the reliability and fault tolerance of remote firmware upgrades, ensuring the stable operation of the intelligent cabinet management system, and is especially suitable for offline upgrade scenarios with no network-connected MCUs.

[0066] In some of the embodiments described above in this application, although a dual-guardian mechanism with mutual wake-up mechanism and JobScheduler periodic check mechanism is proposed for the system guardian module, and CRC check and timeout retransmission mechanism of the protocol conversion module are used to build a dual hardware and software communication fault tolerance system, in practical applications, how to ensure that the main program of the system remains stable during long-term operation, and how to further improve the reliability of communication between the edge computing gateway and the underlying hardware execution unit to meet the stringent requirements of 24 / 7 unattended operation, still require more specific and complete technical solutions.

[0067] To address this, this application further proposes that the system guardian module includes a dual guardian mechanism: a local device security protection mechanism and a remote device security protection mechanism. These two mechanisms monitor and wake each other up. Simultaneously, a periodic wake-up check task is set through the JobScheduler mechanism to actively monitor the running status of the main business program. If the main program is detected as unresponsive or crashing, a restart operation is immediately performed, achieving stable 24 / 7 unattended operation. Specifically, the local device security protection mechanism is typically deployed as an independent process or service running within the edge computing gateway. Its main responsibility is to monitor the running status of the main business program, for example, by checking the existence of the main program process ID, CPU utilization, memory usage, or periodically sending heartbeat packets. When the main program is detected as abnormally terminated or unresponsive for an extended period, the local device security protection mechanism immediately attempts to restart the main program. The remote device security protection mechanism can be a higher-level guardian process or a task scheduling service provided by the operating system (such as JobScheduler). It runs independently of the local device security protection mechanism and can monitor all critical processes, including the local device security protection mechanism. When the local device security protection mechanism itself malfunctions, the remote device security protection mechanism can wake it up or restart it. The dual-daemon mechanism forms a closed-loop self-healing system through mutual monitoring and wake-up, ensuring that the system can automatically recover even in extreme situations. The JobScheduler mechanism, as a system-level task scheduling service, can be configured to periodically wake up and check tasks. Even if both daemon processes terminate unexpectedly, JobScheduler can wake up one or both daemon processes at preset times (e.g., every 5 minutes) to restart monitoring the main program's status, further enhancing the system's robustness and ensuring stable, unattended operation 24 / 7.

[0068] Meanwhile, the protocol conversion module encapsulates cloud-based JSON commands into custom RS485 data frames. These data frames include a frame header, device address, function code, data length, and CRC checksum. A preset timeout period is set after command transmission; if no hardware response is received, a retransmission is automatically triggered. If the maximum number of retransmissions is exceeded, a device fault is flagged and an alarm is reported. Specifically, commands issued by the cloud server typically use JSON (JavaScript Object Notation) format, which is convenient for cloud processing and transmission. However, the underlying embedded hardware execution unit usually communicates via serial communication interfaces such as RS485 and has strict requirements for data format. The protocol conversion module is responsible for parsing the JSON-formatted commands, extracting key information (such as device ID, operation type, parameters, etc.), and then encapsulating it into custom RS485 data frames according to predefined protocol rules. This custom data frame typically includes the following fields: a frame header, identifying the start of the data frame; a device address, specifying the target underlying embedded hardware execution unit for the instruction; a function code, indicating the specific operation type of the instruction (e.g., opening a door, querying status); a data length, representing the number of bytes in subsequent data fields; and a CRC (Cyclic Redundancy Check) checksum, used at the receiving end to verify the integrity and accuracy of data transmission. After the instruction is sent, the protocol conversion module starts a timer with a preset timeout period. If no response is received from the underlying embedded hardware execution unit within this period, the communication is considered a failure, and the protocol conversion module will automatically trigger a retransmission of the instruction. To avoid infinite retransmissions that could exhaust system resources, the system sets a maximum number of retransmissions. Once the number of retransmissions exceeds this maximum, the protocol conversion module will no longer attempt to retransmit, but will instead mark the underlying embedded hardware execution unit as faulty and generate an alarm message to be reported to the edge computing gateway, which can then be synchronized to the cloud server for timely intervention by the administrator.

[0069] Through the above technical solutions, the system protection module adopts a dual protection mechanism of local device security protection and remote device security protection, supplemented by a JobScheduler mechanism for periodic wake-up checks. This effectively addresses situations where the main program crashes or becomes unresponsive, ensuring that the system can automatically recover and continue to operate stably even in unattended conditions, greatly improving the system's reliability and availability. Simultaneously, the protocol conversion module encapsulates cloud-based JSON commands into custom RS485 data frames containing frame headers, device addresses, function codes, data lengths, and CRC checksums, and introduces a timeout retransmission mechanism. This effectively ensures the accuracy and robustness of communication between the edge computing gateway and the underlying embedded hardware execution unit. When communication anomalies occur, the system can promptly detect and report alarms, thus constructing a more complete and reliable hardware and software dual fault-tolerant system, ensuring the stability and security of the intelligent cabinet management system under various complex operating conditions.

[0070] In some of the embodiments described above in this application, although an integrated multi-sensor anti-tampering module is proposed for monitoring physical tampering behavior of the cabinet, in practical applications, the module itself may face the risk of being maliciously damaged or evaded. Moreover, for complex and diverse physical attack behaviors, how to accurately identify and effectively distinguish between normal environmental disturbances and malicious attacks is a key challenge to ensure the reliability of the cabinet physical security protection system.

[0071] In this regard, this application further proposes that the integrated multi-sensor anti-tampering module is encapsulated in a tamper-proof and robust shell, and integrates a vibration acceleration sensor, a magnetic contact Hall sensor array, an internal ambient light sensor, and a module self-destruct anti-tampering switch. The sensors work together to collect physical status data of the cabinet. The anti-tampering module has a built-in anomaly detection algorithm to distinguish between normal environmental disturbances and malicious physical attacks. When anomalies such as prying, forced entry, or module disassembly are detected, a high-priority tampering alarm signal is immediately generated and transmitted to the edge computing gateway through an encrypted serial port.

[0072] Specifically, the housing of the integrated multi-sensor tamper-proof module is constructed using high-strength materials (e.g., aluminum alloy, stainless steel, or high-strength engineering plastics) through integral molding, riveting, or welding to ensure a robust structure that can effectively resist physical damage such as external impacts, compression, or drilling. Simultaneously, the housing is designed with anti-tampering mechanisms, such as using special screws (e.g., one-way screws, internal Torx screws) or triggering an internal microswitch during disassembly to prevent unauthorized physical disassembly.

[0073] The vibration acceleration sensor integrated inside this module is typically a microelectromechanical system (MEMS) accelerometer, which can monitor minute vibrations, impacts, or displacement changes in the cabinet or module body in real time. When the cabinet door is pried open, the module is struck, or an attempt is made to remove it from its fixed position, the sensor generates specific vibration or acceleration signals, which are then converted into digital signals for analysis by the subsequent processing unit.

[0074] A magnetic contact Hall sensor array consists of multiple Hall sensors, typically used in conjunction with permanent magnets, and deployed at the contact surfaces between the module and the cabinet, at the joints of the module housing, or at critical structural connection points. When the module housing is illegally opened or the fixed connection between the module and the cabinet is damaged, changes in the magnetic field will be detected by one or more Hall sensors in the array, thus indicating an abnormality in the physical contact state.

[0075] An internal ambient light sensor is installed inside the anti-tampering module to monitor changes in light intensity inside the module or in its concealed space. When the cabinet door is opened, the module is disassembled, or an external light source (such as a flashlight) shines into the module, the ambient light sensor will detect a significant change in light intensity, which can serve as an auxiliary basis for judging physical tampering.

[0076] A module self-destruct tamper switch is typically a microswitch or spring contact switch that is pressed in or held open when the module is normally installed. When the module is forcibly removed or taken from its fixed position, the switch state changes, immediately triggering a preset self-destruct or alarm mechanism, such as sending a high-priority tamper alarm signal to the edge computing gateway.

[0077] The aforementioned sensors collaboratively collect physical status data of the cabinet. The microcontroller or dedicated processing chip inside the anti-tamper module is responsible for receiving raw data from the vibration acceleration sensor, magnetic contact Hall sensor array, internal ambient light sensor, and module self-destruct tamper switch. This data is synchronized in time and forms a multi-dimensional physical status dataset, providing comprehensive input for subsequent anomaly detection algorithms.

[0078] The anti-tampering module's built-in anomaly detection algorithm can be based on rule engines, threshold comparisons, pattern recognition, or machine learning models. This algorithm distinguishes between normal environmental disturbances and malicious physical attacks through comprehensive analysis and pattern matching of data from different sensors. For example, slight vibrations might indicate an environmental disturbance, but if accompanied by drastic changes in the Hall sensor's state and a sudden increase in ambient light, it is highly likely to be a forced entry attempt. The algorithm can set multiple threshold levels and logical judgments, and combine inputs from multiple sensors to build a more complex judgment model to avoid false alarms and missed detections.

[0079] When the anomaly detection algorithm determines that a malicious attack has occurred, the microcontroller of the anti-tampering module immediately encapsulates a specific data packet. This packet contains the alarm type (e.g., prying open, forced entry, module disassembly), the time of occurrence, a snapshot of relevant sensor data, and a high-priority flag. This alarm signal is designed to interrupt or take precedence over other regular data transmissions, ensuring that it can be quickly received and processed by the edge computing gateway. This high-priority tampering alarm signal is transmitted to the edge computing gateway via an encrypted serial port. The encrypted serial communication link ensures the confidentiality and integrity of the alarm signal during transmission, preventing the alarm information from being eavesdropped on, tampered with, or forged. Encryption can employ either symmetric or asymmetric encryption algorithms to establish a secure session between the anti-tampering module and the edge computing gateway.

[0080] The above technical solution encapsulates an integrated multi-sensor anti-tampering module within a robust, tamper-proof housing. Internally, it integrates a vibration acceleration sensor, a magnetic contact Hall sensor array, an internal ambient light sensor, and a self-destruct tamper-proof switch, achieving physical protection for the module itself and effectively resisting external damage and unauthorized disassembly attempts. The sensors collaboratively collect rack physical status data, providing multi-dimensional and more comprehensive information input for the anomaly detection algorithm. The built-in anomaly detection algorithm accurately distinguishes between normal environmental disturbances and malicious physical attacks, significantly reducing false alarm rates and improving alarm accuracy. When anomalies such as prying, forced entry, or module disassembly are detected, a high-priority tampering alarm signal is immediately generated and transmitted to the edge computing gateway via an encrypted serial port, ensuring the timeliness, security, and reliability of alarm information. This constructs a more robust and intelligent integrated offline and online physical security protection system, effectively solving the problems of module vulnerability and difficulty in accurately identifying attack behaviors.

[0081] In some of the embodiments described above in this application, although an integrated multi-sensor anti-tampering module is proposed to detect physical tampering of the cabinet and generate alarm signals, relying solely on alarm signal generation may be insufficient to cope with complex and ever-changing physical attack scenarios. Especially when network connections are unstable or the nature of the attack is severe, the lack of a comprehensive, flexible, and reliable response mechanism to effectively prevent attacks and promptly notify relevant personnel may lead to an expansion of security risks and delayed response.

[0082] In response to this, the present application further proposes that the edge computing gateway executes a multi-level linkage response after receiving a tampering alarm signal. Specifically, as the core processing unit of the entire system, after receiving the alarm signal sent by the integrated multi-sensor anti-tampering module, the edge computing gateway is no longer a single passive receiver, but actively initiates a series of collaborative actions according to preset policies to cope with threats of different levels. This "multi-level linkage response" means that the response measures are not single, but hierarchical and multi-dimensional, and can be dynamically adjusted according to the nature of the alarm and the system status. For example, an alarm processing engine or state machine can be preset in the software logic inside the edge computing gateway, and this engine triggers different response processes according to the type, severity of the received alarm signal, and the current network connection status.

[0083] First, the system will immediately trigger a local high-decibel audible and visual alarm deterrence. This is the first level and the most direct physical response, aiming to immediately deter potential attackers through strong auditory and visual stimuli, making them abandon or interrupt the attack behavior and attracting the attention of surrounding personnel. The edge computing gateway communicates with the connected audible and visual alarm devices (such as high-pitched buzzers, flashing warning lights) through its I / O interface or through a protocol conversion module, sending instructions to make them start immediately.

[0084] Second, when the network is normal, the system will synchronously push the alarm information to the cloud server and notify the administrator by means of SMS and APP push. This move aims to ensure that the alarm information can be transmitted to the remote administrator in a timely and reliable manner, so that they can understand the situation and take further remote intervention measures. Multiple notification methods increase the reliability of information delivery. The edge computing gateway establishes communication with the cloud server through its Ethernet or wireless module. When receiving a tampering alarm signal and the network status is normal, the gateway encapsulates the alarm data (including time, location, alarm type, sensor data, etc.) into a specific data packet and sends it to the cloud server through a preset communication protocol. After receiving the alarm, the cloud server sends an SMS through the SMS gateway or sends a notification to the administrator through the push service of the mobile application (APP) according to the preset administrator contact information and notification strategy.

[0085] Furthermore, when the network is interrupted, the system will encrypt and store the alarm data in the local database. This measure aims to ensure that important alarm data will not be lost in the case of network unavailability or disconnection, providing a basis for subsequent event analysis and tracing. Encrypted storage guarantees the security of the data. The local database module built into the edge computing gateway is used to store data. When the edge computing gateway detects a network connection interruption, it will encrypt all newly generated tampering alarm signals and their related data, and then write them into a specific table in the local database.

[0086] Finally, when serious tampering is detected, the system issues an emergency lock command to the underlying lock control unit, temporarily locking all doors, which can only be unlocked with administrator authorization. This is the highest level of response, addressing severe physical attacks by physically preventing attackers from further entering or damaging the equipment, thus maximizing the protection of assets within the cabinet. The temporary lock and administrator-authorized unlock mechanism ensures security and controllability. The anomaly detection algorithm built into the edge computing gateway analyzes sensor data to determine the severity of the tampering. When such behavior is detected, the edge computing gateway sends an emergency lock command to the underlying embedded hardware execution unit (especially the lock control board) via the protocol conversion module. Upon receiving the command, the lock control unit activates all connected electromagnetic lock relays, locking them and preventing them from responding to regular door opening commands. Unlocking requires a high level of authentication and authorization from the administrator through a cloud or local management interface.

[0087] Through the aforementioned technical solution, this application enables a comprehensive, flexible, and reliable multi-level linkage response mechanism after the integrated multi-sensor anti-tampering module detects physical tampering of the server rack. First, the immediate triggering of local high-decibel audible and visual alarms effectively deters attackers and draws attention to the site, buying time for subsequent processing. Second, when the network is normal, alarm information is promptly notified to the administrator through multiple channels (SMS, APP push), ensuring the effectiveness of remote monitoring and intervention. More importantly, even in the extreme case of network interruption, alarm data is encrypted and stored in a local database, ensuring the integrity and traceability of critical information. When serious tampering is detected, the system can immediately issue an emergency lock command, physically preventing further action by the attacker, and through the administrator-authorized unlocking mechanism, ensure the highest level of security protection for the server rack assets. This significantly improves the physical security protection capabilities and response efficiency of the intelligent server rack in various complex environments.

[0088] To address this issue, this application proposes a cloud-edge-device collaborative intelligent cabinet management system, in which the underlying embedded hardware execution unit is responsible for receiving and executing instructions from the edge computing gateway, such as controlling the opening and closing of the cabinet door lock. However, in practical applications, if an electromagnetic lock is used as the actuator, continuously supplying power to the electromagnetic lock coil to maintain the unlocked state may cause the coil to overheat, leading to problems such as electromagnetic lock damage, increased energy consumption, and decreased system reliability.

[0089] To address the aforementioned issues, this application further proposes the specific structure and working mechanism of a low-level embedded hardware execution unit. This low-level embedded hardware execution unit includes a lock control board and an MCU controller. The lock control board is a circuit board specifically designed to drive and control the door lock mechanism. It serves as the interface between the MCU controller and the electromagnetic lock and other execution components, responsible for converting the low-level control signals output by the MCU controller into the higher voltage and current required to drive the electromagnetic lock. The MCU controller, or microcontroller unit, is the intelligent core of this execution unit. It is responsible for receiving instructions from the edge computing gateway, parsing the instruction content, and controlling the relays and other execution components on the lock control board according to the instruction logic, thereby achieving precise control of the cabinet door lock.

[0090] This MCU controller integrates a bootloader module that supports IAP (In-Application Programming). The bootloader is a small program that executes before the main application runs, its primary function being to initiate the application's startup. The IAP function allows the MCU to receive new firmware programs via a specific communication interface during its own operation and write them to its own flash memory, thus enabling remote updates to the main application. This means that even if the underlying embedded hardware execution unit lacks direct network connectivity, firmware upgrades can be performed through the edge computing gateway's OTA management module without manual intervention or device disassembly, significantly improving system maintainability and flexibility.

[0091] The lock control unit, as the logical entity responsible for door lock control within the underlying embedded hardware execution unit, parses a verified and valid door opening command sent by the edge computing gateway. Subsequently, the MCU controller drives the electromagnetic lock relay on the lock control board via control signals. The relay, acting as an electronic switch, connects or disconnects the electromagnetic lock's power supply, causing the lock tongue to retract or extend, thus opening or closing the cabinet door. To address the overheating issue that may result from prolonged power supply to the electromagnetic lock, this application introduces a delayed power-off mechanism in the lock control unit. When the lock control unit receives the door opening command and drives the electromagnetic lock relay to the unlocked state, the MCU controller starts a preset timer. After a short delay sufficient to ensure the door is fully open, the MCU controller automatically cuts off the power supply to the electromagnetic lock coil. This means the electromagnetic lock only receives power briefly during the unlocking process and then ceases continuous power consumption, effectively preventing heat accumulation in the coil due to prolonged power supply and fundamentally preventing the risk of coil overheating and damage. This not only extends the service life of the electromagnetic lock, but also reduces system maintenance costs and safety hazards caused by hardware failures.

[0092] To ensure accurate monitoring of the rack door lock status, the underlying embedded hardware execution unit also possesses the ability to provide real-time feedback on the lock's status. After the electromagnetic lock activates, the lock control unit uses built-in sensors to detect the actual position of the bolt, determining whether the lock is open, closed, or locked. This status information is collected by the MCU controller and sent back to the edge computing gateway in real-time via the communication interface. Upon receiving this feedback, the edge computing gateway updates the current status of the rack door lock and reports it to the cloud server as needed, thereby enabling remote monitoring and management of the rack door lock.

[0093] Through the above technical solution, this application effectively solves the problem of overheating and damage caused by prolonged power supply to electromagnetic locks, significantly improving the reliability and service life of the cabinet door lock system. Specifically, the underlying embedded hardware execution unit works in collaboration with the MCU controller through the lock control board. Upon receiving a valid door opening command, it not only accurately drives the electromagnetic lock relay to unlock, but more importantly, its designed delayed power-off mechanism ensures that the electromagnetic lock coil is only briefly energized when necessary, and then immediately cuts off the power supply. This mechanism eliminates the risk of coil overheating at the source, avoiding performance degradation or even complete damage to the electromagnetic lock due to overheating, thereby reducing maintenance costs and ensuring the physical security of the cabinet. Simultaneously, the MCU has a built-in Bootloader module supporting IAP functionality, enabling the underlying hardware to work with the OTA management module of the edge computing gateway to achieve remote, seamless firmware upgrades, greatly simplifying the system maintenance and function iteration process. Furthermore, the function of real-time feedback of the lock body status to the edge computing gateway allows the system to accurately grasp the actual physical state of the door lock, providing reliable data support for cloud-edge-device collaborative management, further enhancing the system's intelligence and security. Overall, this solution significantly improves hardware durability, system maintainability, and real-time status monitoring while ensuring the rack door lock function is implemented.

[0094] In some of the embodiments described above in this application, the cloud-edge-device collaborative intelligent cabinet management system involves the collaborative operation between a cloud server, an edge computing gateway, a low-level embedded hardware execution unit, and an integrated multi-sensor anti-tampering module. However, in actual deployment, there may be various communication needs between these devices at different levels. If a single or indiscriminate communication method is used, it may lead to low communication efficiency, unstable data transmission, and even the risk of signal interference and security vulnerabilities, thereby affecting the reliability and security of the entire system.

[0095] In response, this application further proposes that the edge computing gateway communicates with the cloud server via Ethernet or wireless module, connects to the underlying embedded hardware execution unit via USB to serial port or RS485 interface, and connects to the integrated multi-sensor anti-tampering module via encrypted bus. The three types of communication links are independent of each other to avoid signal interference and malicious interception.

[0096] Specifically, edge computing gateways communicate with cloud servers via Ethernet or wireless modules. Ethernet communication typically uses a standard RJ45 interface and leverages the TCP / IP protocol stack for data transmission, offering high transmission speeds, good stability, and strong anti-interference capabilities, making it suitable for scenarios requiring large data transmission volumes and high reliability. Wireless module communication can utilize technologies such as Wi-Fi, cellular networks (e.g., 4G / 5G), or low-power wide-area networks (e.g., LoRa), providing flexible deployment options and mobility, suitable for environments where cabling is inconvenient or remote access is required.

[0097] Edge computing gateways connect to underlying embedded hardware execution units via USB-to-serial or RS485 interfaces. The USB-to-serial interface, implemented with a built-in USB-to-UART converter chip, converts USB data into serial data for transmission, suitable for short-range, point-to-point device connections and debugging. The RS485 interface, an industrial-grade serial communication standard, uses differential signal transmission and features long transmission distances, strong noise immunity, and support for multi-point communication, making it ideal for connecting multiple underlying embedded hardware execution units in industrial environments.

[0098] The edge computing gateway connects to the integrated multi-sensor tamper-proof module via an encrypted bus. An encrypted bus is a dedicated communication channel that encrypts communication data at the physical or data link layer. For example, it can be based on physical interfaces such as SPI, I2C, or UART, and can ensure the confidentiality and integrity of data transmission by integrating hardware encryption chips or implementing software encryption protocols, effectively preventing sensitive tamper-proof data from being maliciously intercepted or tampered with.

[0099] The three types of communication links described above are physically or logically independent, meaning they use different physical interfaces, communication protocols, or independent communication channels. This independence ensures that the links do not interfere with each other; even if one link fails or is attacked, it will not directly affect the normal operation of other links.

[0100] Through the above technical solutions, the edge computing gateway can select the most suitable communication method based on the characteristics and security requirements of the communication objects to address the communication needs between devices at different levels in the cloud-edge-device collaborative intelligent cabinet management system. For example, high-speed and stable Ethernet or flexible wireless modules are used for data interaction with the cloud server; industrial-grade USB-to-serial or RS485 interfaces are used for reliable command issuance and status feedback with the underlying embedded hardware execution unit; and a highly secure encrypted bus is used for high-sensitivity sensitive data transmission with the integrated multi-sensor anti-tampering module. This layered and independent communication link design effectively avoids signal crosstalk and resource contention between different communication protocols, significantly improving the stability and efficiency of each communication channel. At the same time, the encrypted bus protects the sensitive data of the anti-tampering module, greatly enhancing the system's ability to resist malicious interception and tampering, thereby constructing a highly reliable, highly secure, and highly efficient cloud-edge-device collaborative communication architecture, ensuring the stable operation and data security of the entire intelligent cabinet management system in complex and ever-changing environments.

[0101] In some embodiments described above, a cloud-edge-device collaborative intelligent cabinet management system is proposed. This system achieves intelligent cabinet management through the collaborative operation of a cloud server, an edge computing gateway, a low-level embedded hardware execution unit, and an integrated multi-sensor anti-tampering module. However, in real-world applications, network outages or cloud service anomalies occur frequently, which may cause the system to be unable to provide services normally at critical moments. For example, users may be unable to authenticate or open doors. Furthermore, the physical security status of the cabinets may not be monitored and responded to in a timely and effective manner, thus severely impacting the system's availability, reliability, and security.

[0102] In response, this application further proposes an offline working method for a cloud-edge-device collaborative intelligent cabinet management system, which includes the following steps:

[0103] First, in the event of a network outage or cloud service failure, when a user triggers a cabinet retrieval operation, the edge computing gateway will use its camera to capture a facial image. The determination of network outage or cloud service failure is achieved by the edge computing gateway periodically sending heartbeat packets or service requests to the cloud server. If multiple consecutive heartbeat packets fail to receive a response, or the service request returns an error code, the edge computing gateway can determine that a network outage or cloud service failure is in effect. User-triggered cabinet retrieval operations are typically achieved through a touchscreen, proximity button, or Near Field Communication (NFC) module on the cabinet door. Once the user triggers the operation, the edge computing gateway immediately activates its built-in or external camera and initiates the image acquisition process to obtain the user's facial image data.

[0104] Secondly, the biometric recognition module extracts image feature values ​​and performs feature comparison in a local SQLite database to verify the legitimacy of the user's identity. Specifically, after receiving a facial image captured by the camera, the biometric recognition module built into the edge computing gateway processes the image using its built-in facial recognition algorithm engine to extract facial biometric values, such as key point coordinates and texture features. Subsequently, the module compares the extracted feature values ​​with a pre-registered user biometric database stored in the local database module. The local database module uses the lightweight SQLite database, which can efficiently store and retrieve user biometric data. The comparison process typically involves calculating the similarity between feature vectors and comparing them with a preset threshold to determine the legitimacy of the user's identity.

[0105] Next, after successful authentication, the system directly queries local user permission data. If the permissions are valid, the system sends an opening command to the underlying lock control unit via the protocol conversion module to execute an offline opening operation. Once the biometric identification module verifies the user's identity, the edge computing gateway immediately queries the user permission data stored in the local database module. This permission data records in detail each user's access permissions to a specific cabinet or a specific time period. If the query results show that the user has valid opening permissions, the edge computing gateway will use its built-in protocol conversion module to encapsulate the standard opening command into a specific communication protocol data frame, such as an RS485 data frame, that can be recognized by the underlying embedded hardware execution unit. Subsequently, this data frame is sent to the underlying lock control unit via a physical connection, driving the electromagnetic lock relay to actuate, thereby realizing the offline opening operation of the cabinet.

[0106] Simultaneously, offline operation logs and device status data are cached in a local database. Once the network is restored, the edge computing gateway automatically synchronizes the cached data to the cloud server, completing the data loop. Throughout the offline operation, all relevant data, including user authentication, door opening command issuance, lock status feedback, and other device operating statuses, are recorded in real-time and cached in the edge computing gateway's local database module. This data forms detailed offline operation logs and device status records. When the edge computing gateway detects that the network connection has returned to normal, it automatically initiates a data synchronization mechanism, uploading all cached offline data from the local database module to the cloud server in batches. This process typically employs breakpoint resume and data consistency verification technologies to ensure complete and accurate data synchronization, thereby completing the entire data loop and guaranteeing data consistency between the cloud and the edge.

[0107] Furthermore, in offline mode, the integrated multi-sensor anti-tampering module continuously monitors the physical security status. Upon detecting tampering, it immediately triggers audible and visual alarms and an emergency lockout. Alarm data is synchronously cached locally and reported to the cloud upon network connection. Even offline, the integrated multi-sensor anti-tampering module independently and continuously monitors the physical security status of the cabinet. The integrated vibration acceleration sensor, magnetic contact Hall sensor array, internal ambient light sensor, and self-destruct tamper switch work together to collect real-time physical status data of the cabinet. When its built-in anomaly detection algorithm detects tampering such as prying, forced entry, or module disassembly, it immediately generates a high-priority tampering alarm signal and transmits it to the edge computing gateway via an encrypted independent communication link. Upon receiving the alarm, the edge computing gateway immediately triggers a local high-decibel audible and visual alarm for deterrence and issues an emergency lockout command to the underlying locking unit, temporarily locking all doors. Simultaneously, the alarm data is synchronously cached in the local database module and automatically reported to the cloud server after network recovery.

[0108] By employing the aforementioned offline operating method, this application effectively solves the technical challenge of the intelligent cabinet system being unable to provide normal service and security protection in the event of network interruption or cloud service anomalies. Specifically, the edge computing gateway can independently complete user authentication and door opening operations, ensuring that users can access the cabinet normally under any network conditions, greatly improving system availability and user experience. Simultaneously, the integrated multi-sensor anti-tampering module can continuously monitor physical security even offline, and promptly trigger local alarms and emergency locking responses, effectively protecting the physical security of the cabinet and avoiding security blind spots caused by network failures. Furthermore, the local caching of offline operation logs and alarm data, along with the automatic synchronization mechanism after network recovery, ensures data integrity and consistency, achieving closed-loop data management and preventing data loss, thereby significantly enhancing the robustness, reliability, and security of the entire cloud-edge-device collaborative intelligent cabinet management system.

[0109] The above description is merely an embodiment of this application and is not intended to limit the scope of protection of this application. Various modifications and variations can be made to this application by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this application should be included within the scope of protection of this application.

Claims

1. A cloud-edge-device collaborative intelligent cabinet management system, characterized in that, The system includes a cloud server, an edge computing gateway, an underlying embedded hardware execution unit, and an integrated multi-sensor anti-tampering module. The edge computing gateway uses an Android all-in-one device and integrates a local database module, a biometric recognition module, an OTA management module, a system protection module, and a protocol conversion module. The edge computing gateway communicates with the cloud server, the underlying embedded hardware execution unit, and the integrated multi-sensor anti-tampering module. The local database module caches and stores user permission data, biometric data, and operation log data, working with the biometric recognition module to achieve local identity verification and business decision-making in offline mode. The OTA management module uses a preset cross-level file transfer protocol to divide the firmware package distributed from the cloud into... The packet is transmitted to the underlying embedded hardware execution unit, which, together with the underlying bootloader module, enables remote, seamless firmware upgrades for the network-free MCU. The system protection module employs a dual-protection mechanism with mutual wake-up and a JobScheduler periodic check mechanism to achieve automatic restart and self-healing after the main program crashes abnormally. It also incorporates the CRC check and timeout retransmission mechanism of the protocol conversion module to construct a dual hardware and software communication fault-tolerant system. The integrated multi-sensor anti-tampering module is independently fixed in a concealed structure within the cabinet and connected to the edge computing gateway via an encrypted independent communication link. It monitors physical tampering behavior in the cabinet in real time and triggers multi-level alarms and emergency locking responses, forming an integrated offline and online physical security protection system.

2. The cloud-edge-device collaborative intelligent cabinet management system according to claim 1, characterized in that, The biometric recognition module has a built-in face recognition algorithm engine that can extract local features from the collected face images and perform offline comparison with the feature library stored in the local database module. After the comparison is successful, the door opening command is generated directly based on the local permission data without the need for cloud authorization.

3. The cloud-edge-device collaborative intelligent cabinet management system according to claim 2, characterized in that, The local database module uses the lightweight SQLite database to achieve persistent storage of user permissions, biometrics, operation logs and security event data. During network outages, offline operation records and tampering alarm information are cached locally first, and automatically synchronized to the cloud server in batches after the network is restored, ensuring data consistency between the cloud and the edge.

4. The cloud-edge-device collaborative intelligent cabinet management system according to claim 1, characterized in that, The cross-level file transfer protocol is adapted to serial communication scenarios. The OTA management module divides the firmware package into packets of fixed size and sends them to the underlying embedded hardware execution unit one packet at a time via the serial port. Each time the underlying MCU receives a data packet, it writes it to the memory backup area and completes the verification feedback. After all data packets pass the verification, the Bootloader module updates the boot flag and automatically restarts, moving the new firmware from the backup area to the running area to complete the upgrade. If the upgrade fails, it can automatically roll back to the original firmware version.

5. The cloud-edge-device collaborative intelligent cabinet management system according to claim 1, characterized in that, The system protection module includes a dual protection mechanism: a local device security protection mechanism and a remote device security protection mechanism. The two protection mechanisms monitor and wake each other up. At the same time, a periodic wake-up check task is set through the JobScheduler mechanism to actively monitor the running status of the main business program. If the main program is unresponsive or crashes, a restart operation is immediately performed to achieve stable 24 / 7 unattended operation. The protocol conversion module encapsulates cloud JSON instructions into custom RS485 data frames. The data frame includes a frame header, device address, function code, data length, and CRC checksum. After the instruction is sent, a preset timeout waiting period is set. If no hardware response is received, a retransmission is automatically triggered. If the maximum number of retransmissions is exceeded, the device is marked as faulty and an alarm is reported.

6. The cloud-edge-device collaborative intelligent cabinet management system according to claim 1, characterized in that, The integrated multi-sensor anti-tampering module is encapsulated in a tamper-proof and robust shell. It integrates a vibration acceleration sensor, a magnetic contact Hall sensor array, an internal ambient light sensor, and a module self-destruct anti-tampering switch. The sensors work together to collect physical status data of the cabinet. The anti-tampering module has a built-in anomaly detection algorithm to distinguish between normal environmental disturbances and malicious physical attacks. When anomalies such as prying, forced entry, or module disassembly are detected, a high-priority tampering alarm signal is immediately generated and transmitted to the edge computing gateway via an encrypted serial port.

7. The cloud-edge-device collaborative intelligent cabinet management system according to claim 6, characterized in that, Upon receiving a tampering alarm signal, the edge computing gateway executes a multi-level linkage response: it immediately triggers a local high-decibel audible and visual alarm to deter tampering; when the network is normal, it simultaneously pushes the alarm information to the cloud and notifies the administrator via SMS and APP push; when the network is interrupted, it encrypts and stores the alarm data in the local database; when serious tampering is detected, it issues an emergency lock command to the underlying lock control unit, temporarily locking all departments, which can only be unlocked with administrator authorization.

8. The cloud-edge-device collaborative intelligent cabinet management system according to claim 1, characterized in that, The underlying embedded hardware execution unit includes a lock control board and an MCU controller. The MCU has a built-in bootloader module that supports IAP function. After receiving a valid door opening command, the lock control unit drives the electromagnetic lock relay to act and sets a delay power-off mechanism. After a short delay, the coil power supply is automatically stopped to prevent the coil from overheating and being damaged. At the same time, the lock body status is fed back to the edge computing gateway in real time.

9. The cloud-edge-device collaborative intelligent cabinet management system according to claim 1, characterized in that, The edge computing gateway communicates with the cloud server via Ethernet or a wireless module, connects to the underlying embedded hardware execution unit via a USB-to-serial port or RS485 interface, and connects to the integrated multi-sensor anti-tampering module via an encrypted bus. The three types of communication links are independent of each other to avoid signal interference and malicious interception.

10. An offline working method of the cloud-edge-device collaborative intelligent cabinet management system according to any one of claims 1-9, characterized in that, Includes the following steps: S1. When the network is interrupted or the cloud service is abnormal, the user triggers the cabinet retrieval operation, and the edge computing gateway calls the camera to capture the face image; S2. The biometric recognition module extracts the image feature values, completes feature comparison in the local SQLite database, and verifies the user's identity; S3. After the identity verification is successful, the local user permission data is directly queried. If the permissions are in accordance with the rules, the door opening command is sent to the underlying lock control unit through the protocol conversion module to execute the offline door opening operation; S4. The entire offline operation log and device status data are cached in the local database. After the network is restored, the edge computing gateway automatically synchronizes the cached data to the cloud server to complete the data loop. S5. In the offline state, the integrated multi-sensor anti-tampering module continuously monitors the physical security status. When tampering is detected, the local system immediately triggers an audible and visual alarm and an emergency lock. The alarm data is synchronized to the local cache and reported to the cloud after the network is connected.