Curtain wall windowing system based on Internet of Things and remote linkage control method thereof

Abstract

The invention relates to an Internet of Things window control system, in particular to a curtain wall window opening system based on the Internet of Things and a remote linkage control method of the curtain wall window opening system. The adaptive windowing control method based on the variable structure fuzzy system is provided for solving the problem that an existing control strategy lacks the multi-target cooperation capability, and dynamic weight adjustment and rule sub-library scheduling of various environment factors are supported. Aiming at the problem that an existing user control path is disjointed with system control, a fusion type control algorithm is designed, a unified control result is calculated according to a user control intention and a system suggestion value, control conflicts are avoided, and humanization and consistency of system response are improved. Meanwhile, in order to solve the problem that traditional remote control lacks safety communication guarantee, the whole process of the curtain wall window opening system, namely end-to-end safety sessions from a user and a sensor to a platform, comprises registration identity binding, secret key rolling updating, MAC verification protection, abnormity tracking response and other mechanisms, and the confidentiality, integrity and traceability of control instructions are guaranteed.

Description

Technical Field

[0001] This invention relates to an Internet of Things (IoT) window control system, specifically an IoT-based curtain wall window opening system and its remote linkage control method. Background Technology

[0002] With the development of smart building and green energy-saving concepts, more and more buildings are introducing Internet of Things (IoT) technology to upgrade their window ventilation systems. In particular, openable window structures are being connected to sensor networks to achieve automatic opening and closing control based on environmental variables (such as temperature, humidity, and carbon dioxide concentration), or remote window opening via user remote operation terminals (such as apps, web platforms, and voice assistants). Currently, most related technologies focus on automatic window systems, with core structures including:

[0003] Environmental sensors are deployed on windows or indoors and outdoors to collect real-time data;

[0004] The decision to open or close a window is generated using static threshold control, simple fuzzy logic, or a neural network prediction model.

[0005] The window is opened or closed by an electric actuator;

[0006] They mostly use wireless sensor networks (WSN) or local communication protocols based on LoRa, ZigBee, and Wi-Fi, combined with edge or cloud gateways for coordination;

[0007] Users can send window control commands via mobile software or voice terminals. Some systems support functions such as time preset and centralized group control.

[0008] Window control rule systems based on fixed environmental thresholds determine whether to open windows by setting preset thresholds for temperature, humidity, or carbon dioxide concentration, such as "open windows when carbon dioxide > 800 ppm, close windows when < 600 ppm". While these static rule-based methods are easy to deploy, they struggle to adapt to situations where multiple environmental factors change simultaneously. For example, in scenarios with high temperature and humidity but low wind speed, single threshold control can lead to frequent window control actions, increased user discomfort, and a lack of adaptability and environmental integration capabilities.

[0009] In addition, existing systems generally adopt two independent mechanisms: automatic system control and remote user control. For example, user control sends direct control commands through an app, while system control is based on sensor inference. The two lack effective integration. By setting a single priority, if user control is given a higher priority, it is difficult to close the window for the user when the environmental conditions are detected to be severe. If system detection is given a higher priority, it is difficult to take into account the user's personal wishes.

[0010] The user's remote control relies solely on traditional interface command issuance mechanisms, employing common protocols such as HTTP or MQTT to send commands. While basic communication encryption (such as TLS) is available, the lack of application-layer identity management mechanisms and fine-grained authentication design makes it difficult to effectively defend against identity spoofing, command tampering, or replay attacks. Furthermore, command transmission lacks context state binding, source traceability, and session key update mechanisms, resulting in insufficient security and auditing capabilities.

[0011] In existing IoT window control systems, sensors or control terminals mostly participate in the communication process using the form of "device number + preset key," which cannot support dynamic key updates, revocation, or multi-identity binding management of devices. Once sensor keys are leaked or devices are hijacked, the overall communication security of the system faces an irrecoverable risk, and there is no ability to track anomalies. Summary of the Invention

[0012] To address the problem that existing building window control systems cannot achieve precise control and remote safety management, this invention provides a clearly structured, rationally controlled, and reliable Internet of Things-based curtain wall window opening system and its remote linkage control method, enabling precise control and remote safety management of window opening in complex building environments.

[0013] This invention is achieved using the following technical solution: an Internet of Things-based curtain wall window system, comprising the following layers:

[0014] Sensing Access Layer: This layer includes a sensor interface module and a data acquisition and scheduling module, which are used to connect the internal and external sensor networks of the building and to collect environmental data in a structured manner;

[0015] Sensor interface module: used to interface with the communication protocols of various sensors;

[0016] Data acquisition and scheduling module: According to the set sampling period, it schedules various sensors to perform data acquisition tasks; the acquired data supports local buffering, and duplicate data is deduplicated through a sliding window algorithm. At the same time, it combines statistical analysis to mark outliers, and finally generates sensing data that conforms to the internal data format and sends it to the data processing layer.

[0017] Data processing layer: This layer includes a context modeling module and a feature encoding module, which are used to perform structured processing on the collected perception data and generate environmental feature encoding vectors that can be used by the control decision layer;

[0018] Context modeling module: Labels the current sensed data based on historical environmental sequences, time labels, and weather state context information to construct an environmental context vector;

[0019] Feature encoding module: merges the perceived data with the environmental context vector, and performs normalization, standard deviation compression and missing value imputation operations on it to generate a unified format environmental feature encoding vector and pass it to the control decision layer;

[0020] Control Decision Layer: This layer includes a fuzzy inference control module, a user intent fusion module, and an execution parameter generation module, which are used to complete the final inference decision on window opening and issue control parameters.

[0021] Fuzzy Inference Control Module: Based on environmental feature encoding vectors, this module invokes a variable-structure fuzzy system for multi-objective inference. It dynamically selects different rule sub-libraries, determines the priority objective based on the current context, and then executes fuzzy logic inference to generate a system-suggested opening value. ;

[0022] User intent fusion module: Receives user control suggestion values ​​from the service interaction layer when a remote control request exists. and execute the fusion algorithm. , among which fusion factor The final target opening value of the system is determined by the current control weight strategy. ;

[0023] Execution parameter generation module: Based on the fused target aperture value By combining the window's historical motion state, current execution position, and damping factor information, a precise window control parameter instruction package is generated and pushed to the execution driver layer for subsequent remote driver calls.

[0024] Service Interaction Layer: This layer includes a remote command access module and a control feedback module, which are used to realize remote two-way interaction and control feedback closed loop between users and the system;

[0025] Remote command access module: This module receives remote control requests, including those from mobile apps, voice terminals, and web platform user control terminals. It first requests command authentication and identity verification from the security management layer. Once verified, it sends the recommended control value. The data is then transmitted back to the control decision-making level.

[0026] Control Receive Feedback Module: Monitors and obtains the state of the window after it is driven by the window control parameter instruction package, obtains the control execution result, encapsulates the control execution result into a control receive report, and pushes it to the user control terminal at a set period. It supports status subscription, anomaly push and remote log persistent query.

[0027] Security Management Layer: This layer includes a node identity registration module, a communication authentication and key negotiation module, and a session trace chain management module, which are used to implement identity verification, security encryption, and attack tracing for the entire system control process;

[0028] Node identity registration module: Based on the centralized registration center-master password generator architecture, it registers each user control terminal. With sensor nodes Register;

[0029] Communication authentication and key negotiation module: It adopts a two-way challenge-response mechanism to authenticate remote control requests and data upload processes. It combines AES symmetric encryption and HMAC message digest to realize the integrity of command content and the validity of identity. After each round of authentication negotiation, a new session key is generated for that interaction.

[0030] Session Trajectory Chain Management Module: Generates a unique trajectory tag (Session_Tag) for each control session and constructs a trajectory chain structure based on time chain and identity chain for anomaly tracking, attack location and command tracing. It supports tracking revocation mechanism and key rolling update.

[0031] The remote linkage control method for the above-mentioned IoT-based curtain wall window opening system includes the following steps in the sensor node registration process within the safety management layer:

[0032] Step 1: Identity initialization for each deployed sensor node. Generate a globally unique identifier This identifier can be generated based on the device's factory code, geographic location information, and device type hash, ensuring that it is unique in the network.

[0033] Step 2: Master Key Binding, Gateway Node For each sensor node Allocate an initial master key As the source key for identity authentication, the gateway node GN calculates it using a hash function within the secure channel. , generated This serves as the master key for the sensor in subsequent communications, used for data encryption and authentication.

[0034] Step 3: Local storage and destruction of the master key Stored separately in the gateway node With sensor nodes Gateway node Will the master key An index mapping is established for subsequent authentication, while the initial master key... The key is destroyed immediately to ensure its one-way nature and prevent subsequent leakage that could lead to a network-wide data breach.

[0035] Step 4: Registration Confirmation and Track Tag Initialization. After registration is complete, the gateway node... To sensor nodes Initial communication trajectory tags are issued. As the starting point for the sensor node to participate in system communication, this initial communication trajectory label will be updated as the session progresses, and is used for communication behavior tracking and key update indexing.

[0036] Step 5: Write the registration log and attach it to the anomaly audit chain. The registration event is written to the system registration log and attached to the system's anomaly audit chain to ensure that the access time, key status and registration behavior of the sensor node can be traced at any point in the future, thus building a complete security lifecycle management mechanism.

[0037] The remote linkage control method for the above-mentioned IoT-based curtain wall window opening system includes the following steps in the user registration process within the security management layer:

[0038] Step 1: User registration request submission. The user initiates a registration request through the user control terminal and sends it to the gateway node. Submit its unique user identifier. With custom password Security management Perform uniqueness verification to prevent duplicate registrations, and verify passwords. Validate the complexity and input specifications for legality.

[0039] Step 2: Key derivation and identity binding, using a salted hash function to perform irreversible processing on the password. ,in, It is a random salting factor generated by the system for each user, which is the identity key. Used for authentication and HMAC signing of all control messages between the user control terminal and the system; , identity key after salting hash It is stored in the authentication database, and a user identity status table is established to record registration time, initial device, login IP and auxiliary metadata;

[0040] Step 3: Control terminal binding and authentication credential configuration. The system supports users binding multiple user control terminals simultaneously. Each user control terminal is assigned an independent terminal session key by the platform upon initial login. and based on Derived from: This key is used to establish a TLS or symmetric encrypted channel with the terminal; the terminal credential configuration package contains the device ID, terminal type, binding timestamp, and signature credential to ensure the legitimacy and identifiability of the user control terminal's communication identity;

[0041] Step 4: Track tag and access control initialization, the system is... Initialize session trajectory tags And establish a user access control list to record the user's permission level; trajectory tags It will serve as the context reference point for all subsequent communication behaviors and be linked to the system's anomaly audit chain to identify abnormal instructions, authenticate forged behavior, or illegal instruction replay security risks.

[0042] Step 5: Client security prompts and key rotation mechanism activation. After successful user registration, the user's control terminal receives a registration success notification and terminal authentication credentials, prompting the user to configure a second layer of security verification to enhance security. Simultaneously, the system activates the key rotation policy, triggering the identity key according to a set period or specific events. and session key The update process enhances forward security and resistance to attacks during long-term use.

[0043] The aforementioned remote linkage control method for an IoT-based curtain wall window opening system includes the following steps in the user login authentication process within the security management layer:

[0044] Step 1: Login request submission. The user initiates the login operation on the user control terminal and submits the request. With custom password After local preprocessing, the user control terminal sends the authentication request to the security management layer. The request message includes: user ID, salted password digest, and client device identifier. Timestamp and requested random number;

[0045] Step Two: Password Verification and Replay Protection Check, Search and Associated random salting factor With identity key And calculate the hash value of the login input. ,like If the login request fails, the login attempt will be rejected and the failed attempt will be logged. It is supported to set a login failure threshold and temporary blocking. If they are consistent, the timestamp and nonce will be checked to prevent replay attacks.

[0046] Step 3: Terminal session key negotiation. After authentication, the system will grant the device... Generate current session key And establish a TLS channel or a secure channel based on symmetric encryption, the session key is used for data encryption and instruction signing when the user sends control commands;

[0047] Step 4: Session Track Update and Behavior Audit Initialization, the system updates the user's session track tags. It also enables login session logging, including device, IP, location, login time, etc., and writes it into the security audit chain to support abnormal behavior analysis and post-event accountability.

[0048] The remote linkage control method for the above-mentioned IoT-based curtain wall window opening system includes the following steps in the sensor data uploading and processing process at the sensing access layer:

[0049] Step 1: Data Acquisition and Local Packaging, Sensor Nodes After periodically collecting raw environmental data such as temperature, light intensity, CO2 concentration, PM2.5, noise, and wind speed, the data is packaged into a data packet.

[0050] Step 2: Encrypted upload. Data is encrypted using the AES-CBC symmetric encryption algorithm. The data packet is transmitted using a key-encrypted method and includes an HMAC checksum. After decryption, the platform verifies the MAC value to determine whether the data packet has been tampered with.

[0051] Step 3: Processing by the perception access layer. The perception access layer is responsible for:

[0052] Authentication: Confirm Valid and the key is effective;

[0053] Local caching: If data arrives faster than the processing capacity, it is first stored in the buffer;

[0054] Deduplication: Remove duplicate data within the time window;

[0055] Outlier labeling: Detecting sudden changes or outliers using a sliding statistical window and attaching anomaly labels;

[0056] Data scheduling: Pack and upload data to the data processing layer based on task priority and system load.

[0057] The aforementioned remote linkage control method for an IoT-based curtain wall window system includes the following steps in context modeling and feature encoding at the data processing layer:

[0058] Step 1: The context modeling module establishes a set of context tags based on multi-dimensional information such as environmental time, historical weather, and historical user behavior.

[0059] Step 2: The feature encoding module normalizes the collected values ​​from multiple types of sensors and constructs a unified environmental feature vector by combining it with context labels.

[0060] The aforementioned remote linkage control method for an IoT-based curtain wall window opening system includes the following steps in the fuzzy reasoning process of the control decision layer:

[0061] Step 1: Design a family of membership functions for each environmental factor. The numerical inputs in the environmental feature vector F are mapped to the corresponding membership space to form a fuzzy input set.

[0062] Step 2: Based on the context labels, select and activate the rule sub-library that best matches the current scenario;

[0063] Step 3: Use fuzzy inference to fuse the outputs of multiple rules, and finally generate a window target opening value between 0 and 100%.

[0064] The aforementioned remote linkage control method for an IoT-based curtain wall window opening system includes the following steps for user control command access and verification in the service interaction layer remote command access module:

[0065] Step 1: The user sends a window opening control request on the user control terminal, and the request is encapsulated as a remote control command message;

[0066] Step Two: The remote control command message from the remote command access module undergoes the following security verification process: 1. Extraction And obtain its currently valid device key. 2. Verify if the timestamp is within the valid time window; 3. Verify if the Nonce is appearing for the first time to prevent replay; 4. Recalculate the signature HMAC′ and verify:

[0067] HMAC′=HMAC( , If HMAC′≠Sig, then discard the instruction and record the security event;

[0068] Step 3: After authentication, the system converts the control command into a standard format, writes it into the current user behavior trajectory chain, and attaches it to the current... This is to support subsequent tracking, auditing, and intent fusion processes.

[0069] The remote linkage control method for the above-mentioned IoT-based curtain wall window opening system includes the following steps in the design and implementation of the fusion mechanism for the user intent fusion module at the control decision level:

[0070] Step 1: This module receives input from two sources: DesiredOpenRatio: the user-target opening from the remote control command; Open_ratio: the system-recommended opening output by the fuzzy inference result of the control decision layer; and uses Open_ratio as the system-recommended opening value. By combining the user's recent overall opening tendency, opening preferences over different time periods, and the range of user preferred opening ratios, a user behavior preference vector is established to obtain the user control suggestion value. Alternatively, DesiredOpenRatio can be used directly as the user-controlled suggested value. ;

[0071] Step 2: Design an algorithm to obtain the final opening degree. ,in, ∈[0.1,0.9] represents the dynamic fusion weight coefficient, indicating the credibility of the system;

[0072] Step 3: The merged Open fused It will be written into the Session behavior trajectory and flow to the execution driver layer as formal control parameters, supporting subsequent debugging analysis and control correction modeling.

[0073] The aforementioned remote linkage control method for an IoT-based curtain wall window opening system includes the following steps in the position calibration and damping compensation mechanism within the parameter generation module of the control decision layer:

[0074] Step 1: The system periodically, or before receiving control commands, acquires the actual opening degree of the current window via an angle encoder or Hall sensor. This value is related to the expected opening after the last command was executed. If the deviation exceeds δ=±2%, it is marked as an offset drift, and the execution offset record table is updated. This is used to generate the error inverse compensation term for the next action;

[0075] Step Two: Based on the window load characteristics, motor inertia, and environmental disturbances, design a damping algorithm to dynamically adjust the motion speed: Change to openness The process includes B+ offset drift; the initial segment -10%, slow start; 10-90% in the middle, stable and uniform speed; 90% to target in the final stage. The speed decreases;

[0076] Step 3: Finally, the execution parameters are encapsulated and flow into the drive control interface. The actuator directly drives the window opening motor to complete the actual window status update.

[0077] This invention addresses the lack of multi-objective collaborative capabilities in existing control strategies by proposing an adaptive windowing control method based on a variable structure fuzzy system. This method supports dynamic weight adjustment of various environmental factors and rule sub-library scheduling to adapt to the comprehensive trade-offs between ventilation, safety, and energy conservation objectives in different scenarios. To address the disconnect between existing user control paths and system control, a fusion control algorithm is designed. This algorithm calculates a unified control result based on the user's control intent and the system's suggested values ​​without requiring user intervention, avoiding control conflicts and improving the humanization and consistency of system response. Furthermore, to solve the problem of insufficient secure communication in traditional remote control, the entire curtain wall windowing system, from user and sensor to platform, features end-to-end secure sessions, including registration and identity binding, rolling key updates, MAC verification protection, and anomaly tracking and response mechanisms, ensuring the confidentiality, integrity, and traceability of control commands.

[0078] The system features a sensing access layer, including a sensor interface module and a data acquisition and scheduling module. It supports the access of multiple types of environmental sensors, periodic data acquisition, deduplication, and anomaly marking. It can simultaneously sense multi-dimensional environmental conditions (such as temperature, light intensity, CO2, PM2.5, and wind speed), improving the comprehensiveness and responsiveness of environmental monitoring. Simultaneously, the anomaly marking mechanism can proactively eliminate failed or deviating data, enhancing control stability and data reliability.

[0079] The system employs a data processing layer, including a context modeling module and a feature encoding module. Based on historical environmental, time, and weather information, it performs context labeling modeling on the current perceived data and generates standardized encoded feature vectors. This enhances the system's ability to understand dynamic environmental changes, enabling control strategies to adapt to different scenarios and preventing the failure of fixed strategies. Simultaneously, it performs unified formatting and missing value compensation on input data, improving the robustness of the overall control chain.

[0080] The control decision layer employs a variable-structure fuzzy control mechanism, dynamically activating different rule sub-libraries based on context, and performing collaborative reasoning to generate system-recommended opening values ​​for multiple environmental targets. By introducing a variable-structure fuzzy system, it can dynamically adapt to different ventilation targets and still output highly robust control results under complex environmental changes, effectively improving the system's stable control capability in non-ideal environments.

[0081] A user intent fusion module is implemented, employing a fusion algorithm to directly combine system-suggested values ​​and user-controlled values ​​without interrupting prompts or decision confirmations. This significantly improves human-machine collaboration efficiency, avoids response delays caused by user prompts or manual interventions in traditional control paths, ensures continuous and consistent system control, and enhances the user experience.

[0082] The control parameter generation module combines the current window state, damping coefficient, and physical constraints to generate precise control commands and provides a position calibration function to correct the window's current execution state. This effectively solves the actual opening error caused by window friction, wind pressure, or mechanical lag, achieving refined control of the opening. It is particularly suitable for addressing the problem of unstable window movement in high-rise curtain wall structures, improving execution consistency and operational safety.

[0083] The service interaction layer supports control command access from heterogeneous terminals such as voice terminals and mobile apps through a remote command access module, while the control feedback module pushes the control status to the user terminal in real time. It supports cross-platform remote linkage control, enabling efficient collaboration between the window control system and various user terminals; the feedback mechanism allows users to perceive the current window status, avoiding black-box control and improving operational transparency and system reliability.

[0084] The security management layer employs a node identity registration mechanism and a dynamic key negotiation mechanism to configure unique identifiers and dynamic keys for user terminals and sensor nodes, and uses a trajectory chain structure to track command behavior. This prevents communication security issues such as unauthorized device impersonation, command forgery, and data hijacking, and supports forward security, backward security, anomaly tracking, and rolling key updates, effectively improving the security and traceability of the control process and meeting the high security standards required in smart buildings.

[0085] The present invention has the following technical advantages:

[0086] (1) A multi-objective cooperative control method based on variable structure fuzzy system, which supports dynamic weighting of sensor variables and scheduling of fuzzy rule sub-library;

[0087] (2) User intent fusion algorithm, which automatically merges system suggestions and user preferences without prompting user interaction;

[0088] (3) A dynamic communication security mechanism based on session trace chain, combined with identity evolution, key update and MAC protection, to achieve forward and backward security and traceability;

[0089] (4) Supports fine-grained percentage control of window status, and optimizes the execution action by combining position calibration and damping algorithm;

[0090] (5) It has an integrated control process from remote request generation, local control inference, to encrypted control parameter generation and receipt. Detailed Implementation

[0091] The Internet of Things-based curtain wall window system includes the following levels:

[0092] 1. Perception Access Layer

[0093] This layer includes a sensor interface module and a data acquisition and scheduling module, which are used to connect the building's internal and external sensor networks and to collect environmental data in a structured manner.

[0094] Sensor Interface Module: Used to interface with the communication protocols of devices such as temperature sensors, light sensors, carbon dioxide concentration sensors, noise sensors, PM2.5 concentration sensors, and wind speed sensors. This module is configured with driver adaptation logic according to the communication standard of each sensor (such as I²C, UART, RS485, or ZigBee) to achieve low-latency retrieval and standardized encapsulation of sensor data.

[0095] Data acquisition and scheduling module: Based on the set sampling period, it schedules various sensors to perform data acquisition tasks; the acquired data supports local buffering, and duplicate data is removed using a sliding window algorithm, while outliers (such as mutations or missing values) are marked using statistical analysis. Finally, it generates sensing data conforming to the internal data format and sends it to the data processing layer.

[0096] 2. Data Processing Layer

[0097] This layer includes a context modeling module and a feature encoding module, which are used to perform structured processing on the perceived data and generate high-quality environmental feature encoding vectors that can be used by the control decision layer.

[0098] Context modeling module: Labels the current perceived data based on contextual information such as historical environmental sequences, time labels (e.g., day / night), and weather conditions (e.g., sunny / rainy) to construct an environmental context vector.

[0099] Feature encoding module: merges the perceived data with the environmental context vector, and performs normalization, standard deviation compression and missing value imputation on it to generate a unified format environmental feature encoding vector and pass it to the control decision layer.

[0100] 3. Control the decision-making level

[0101] This layer includes a fuzzy inference control module, a user intent fusion module, and an execution parameter generation module, which are used to complete the final inference decision and control parameter distribution for window opening.

[0102] The fuzzy inference control module, based on environmental feature encoding vectors, invokes a variable-structure fuzzy system for multi-objective inference. This module dynamically selects different rule sub-libraries, determines the priority objective based on the current context, and then executes fuzzy logic inference to generate a system-suggested opening value. .

[0103] User intent fusion module: Receives user control suggestion values ​​from the service interaction layer when a remote control request exists. and execute the fusion algorithm. , among which fusion factor The final target opening value of the system is determined by the current control weight strategy. .

[0104] Execution parameter generation module: Based on the fused target aperture value By combining information such as the window's historical motion state, current execution position, and damping factor, a precise window control parameter instruction package is generated and pushed to the execution driver layer for subsequent driver calls.

[0105] 4. Service Interaction Layer

[0106] This layer includes a remote command access module and a control feedback module, which are used to realize remote two-way interaction and control feedback closed loop between users and the system.

[0107] Remote command access module: This module receives remote control requests via HTTPS, MQTT, or WebSocket protocols, including those from user control terminals such as mobile apps, voice terminals, and web platforms. It first requests command authentication and identity verification from the security management layer. Once verified, it sends the recommended control values. It is then transmitted back to the control decision-making level.

[0108] Control Receive Feedback Module: Monitors and obtains the window status after the window is driven by the window control parameter instruction package, obtains the control execution results (target opening degree, current opening degree, execution time, etc.), encapsulates the control execution results into a control receive report, and pushes it to the user control terminal at a set period. It supports status subscription, anomaly push and remote log persistent query.

[0109] 5. Security Management Team

[0110] This layer includes a node identity registration module, a communication authentication and key negotiation module, and a session trace chain management module, which are used to implement identity verification, security encryption, and attack tracking for the entire system control process.

[0111] Node identity registration module: Based on the centralized registration center-master password generator architecture, it registers each user control terminal. With sensor nodes Register.

[0112] The communication authentication and key negotiation module employs a two-way challenge-response mechanism for authenticating remote control requests and data uploads. It combines AES symmetric encryption and HMAC message digests to verify command content integrity and identity validity. A new session key is generated after each round of authentication negotiation for that interaction.

[0113] Session Trajectory Chain Management Module: Generates a unique trajectory tag (Session_Tag) for each control session and constructs a trajectory chain structure based on time chain and identity chain for anomaly tracking, attack location and command tracing. It supports tracking revocation mechanism and key rolling update.

[0114] [Security Management Layer] Sensor Node Registration Process

[0115] In the system architecture of this invention, the sensor node registration process is used in the initial stage of sensor deployment or during sensor replacement to ensure that each physical sensor has a globally unique identity and to establish a connection with the master control node, i.e., the gateway node. This forms the basis for encrypted communication between systems. This process is crucial for ensuring secure data transmission and traceability, with its core objectives being: preventing unauthorized devices from accessing the system; establishing a foundation for key negotiation; and building a tracking system for subsequent communication.

[0116] Step 1: Identity Initialization

[0117] For each deployed sensor node Generate a globally unique identifier This identifier can be generated based on the device's factory code, geographic location information, and device type hash, ensuring that it is unique across the network.

[0118] Step 2: Master Key Binding

[0119] Gateway Node For each sensor node Allocate an initial master key This serves as the source key for identity authentication. In the secure channel, the gateway node... Calculated using a hash function , generated This serves as the master key for the sensor in subsequent communications, used for data encryption and authentication.

[0120] Step 3: Local Storage and Destruction

[0121] Master key Stored separately in the gateway node With sensor nodes Gateway node Will the master key An index mapping is established for subsequent authentication, while the initial master key... The key is destroyed immediately to ensure its one-way nature and prevent subsequent leakage that could lead to a network-wide data breach.

[0122] Step 4: Registration Confirmation and Track Tag Initialization

[0123] After registration is complete, the gateway node To sensor nodes Initial communication trajectory tags are issued. This serves as the starting point for the sensor node's participation in system communication. This initial communication trajectory label will be updated as the session progresses, and is used for communication behavior tracking and key update indexing.

[0124] Step 5: Registration Log Writing and Anomaly Audit Chain Connection

[0125] The registration event is written to the system registration log and linked to the system's anomaly audit chain, ensuring that the access time, key status, and registration behavior of the sensor node can be traced at any point in the future, thus building a complete security lifecycle management mechanism.

[0126] [Security Management Layer] User Registration Process

[0127] The user registration process establishes a two-way authentication foundation between the user and the control platform, supporting the verification of the source of remote control commands, the establishment of encrypted channels, and behavior tracking. This process is applicable to various user interaction interfaces, including mobile apps, web terminals, and voice devices.

[0128] Step 1: Submit User Registration Request

[0129] Users initiate registration requests through user control terminals (such as apps or web applications) and by sending requests to the gateway node. Submit its unique user identifier. Such as mobile phone number, third-party account, username, etc., and custom password. .

[0130] Security management Perform uniqueness verification to prevent duplicate registrations, and verify passwords. The complexity and input specifications are validated for legality.

[0131] Step Two: Key Derivation and Identity Binding

[0132] Use salted hash functions (such as PBKDF2, bcrypt, or scrypt) to perform irreversible processing on the password. ,in, It is a random salting factor generated by the system for each user, ensuring that even if multiple users set the same password, the system-derived identity key remains the same. They are also different from each other. This identity key is used for authentication and HMAC signing of all subsequent control messages between the client and the system.

[0133] Will , identity key after salting hash It is stored in the authentication database, and a user identity status table is established to record auxiliary metadata such as registration time, initial device, and login IP.

[0134] Step 3: Control Terminal Binding and Authentication Credential Configuration

[0135] The system supports users binding multiple user control terminals (such as mobile phones, tablets, and voice boxes) simultaneously. Each user control terminal is assigned an independent terminal session key by the platform upon initial login. and based on identity key Derived from: This key is used to establish a TLS or symmetric encrypted channel with the terminal. The terminal credential configuration package contains the device ID, terminal type, binding timestamp, and signature credential, ensuring the legitimacy and identifiability of the client's communication identity.

[0136] Step 4: Track Label and Access Control Initialization

[0137] The system is Initialize session trajectory tags In addition, a User Access Control List (ACL) is established to record the user's permission levels, including but not limited to: viewing window status only, controlling window opening and closing (setting opening degree), configuring window control policy preferences, executing commands such as forcibly opening / closing windows, and managing family sharing permissions;

[0138] The trajectory tags will serve as context references for all subsequent communication activities and will be linked to the system's anomaly audit chain to identify security risks such as abnormal commands, authentication forgery, or illegal command replay.

[0139] Step 5: Activate Client Security Prompts and Key Rotation Mechanism

[0140] After successful user registration, the user's control terminal receives a registration success notification and terminal authentication credentials, prompting the user to configure a second layer of security verification (such as SMS verification code, device authentication code, etc.) to enhance security. Simultaneously, the system activates a key rotation policy, triggering the identity key based on a set period (e.g., 30 days) or specific events (e.g., multi-terminal login, sensitive operations, etc.). and session key The update process enhances forward security and resistance to attacks during long-term use.

[0141] [Security Management Layer] User Login Authentication Process

[0142] The user login authentication process is a prerequisite for remote control. It aims to verify the legitimacy of the user's identity, check the integrity of communication credentials, and establish a secure communication channel for the encrypted transmission and response feedback of window control commands.

[0143] Step 1: Submit Login Request

[0144] The user initiates a login process on the user control terminal and submits the request. With entering password After local preprocessing, the user control terminal sends the authentication request to the security management layer. The request message includes: user ID, salted password digest (not plaintext), and client device identifier. Timestamp and request random number (Nonce).

[0145] Step 2: Password Verification and Replay Protection Check

[0146] Search and Associated random salting factor With identity key And calculate the hash value of the login input. ,like If the login request fails, the login attempt will be rejected and the failed attempt will be logged. Setting a login failure threshold and temporary blocking is supported.

[0147] If they match, check if the timestamp and nonce have been used to prevent replay attacks.

[0148] Step 3: Terminal Session Key Negotiation

[0149] After authentication, the system assigns a client device identifier to the device. Generate current session key It establishes a TLS channel or a secure channel based on symmetric encryption, and the session key is used for data encryption and instruction signing when the user sends control commands.

[0150] Step 4: Session Tracking Update and Behavior Audit Initialization

[0151] The system updates the user's session trajectory tags. It also enables login session logging, including device, IP, location, login time, etc., and writes it into the security audit chain to support abnormal behavior analysis and post-event accountability.

[0152] [Security Management Layer] User Node Recovery Process

[0153] This process is used to restore a user's remote control capabilities after the user's device is lost, the terminal is replaced, or the user exits the system due to expired / abnormal key access. The goal is to prevent unauthorized recovery while ensuring data integrity and session continuity.

[0154] Step 1: Submit the recovery request

[0155] Users initiate recovery requests through trusted channels (such as SMS verification, email identity links, original device verification, etc.) and submit them. Restore verification information (mobile verification code, security questions, trusted device authorization); if it is a new device, provide the new device identifier. .

[0156] Step Two: Multi-Factor Identity Verification

[0157] The system compares and restores the verification information and checks. If an active device exists, and the original device still exists and is trusted, cross-device verification can be triggered (such as a pop-up confirmation on the old device); otherwise, an alternative verification path will be activated.

[0158] After successful verification, the system generates a new terminal session key: .

[0159] Step 3: Reset Session Tracking and Authorization Synchronization

[0160] The system clears the session history state of the old terminal and updates the user's current active terminal to point to the new one. and reinitialize it. Use Access Control Lists (ACLs) to prevent older devices from retaining valid authorizations without revocation.

[0161] Step 4: Security Prompts and Multi-device Freezing Mechanism

[0162] After a user successfully recovers their account, the system pushes a recovery notification to all old devices. It also supports options such as freezing old devices or limiting login to a single device to enhance proactive security protection capabilities in account recovery scenarios.

[0163] [Security Management Layer] Sensor Node Recovery Process

[0164] When sensor node When a user is disconnected from system communication due to network loss, key loss, sensor replacement, or attack, and needs to rejoin, a controlled recovery process must be used to establish a new communication key and session tag, ensuring the continuity of historical data and identity.

[0165] Step 1: Device Identity Verification

[0166] The system uses information such as physical port, device ID (e.g., MAC address), and original deployment records to... Perform identity verification; if a match fails, refuse to restore access.

[0167] Step Two: Key Recovery and Update

[0168] System search for original and Record (if the key is recoverable), otherwise generate a new one. And calculate the new key. The platform will distribute the new key to Controlled write area, and old key Marked as invalid, all historical session traces are closed.

[0169] Step 3: Track Chain Recovery and Reconstruction

[0170] The system is after recovery Assign new It generates a recovery trajectory chain, which is linked with the historical session trajectory chain to form a complete link record, supporting subsequent security auditing and anomaly detection.

[0171] Step 4: Communication Channel Renegotiation

[0172] System notification control decision layer and data processing layer resubscribe The data stream is synchronized with the recovered identity status to avoid inference errors or control interruptions caused by inconsistent status.

[0173] [Sensing Access Layer] Sensor Data Upload and Sensing Access Layer Process

[0174] All deployed sensor nodes The system needs to report current environmental data periodically or on an event-triggered basis. To ensure data confidentiality and integrity, the entire upload process is based on the registered identity master key. Encryption and authentication are implemented. The perception access layer is responsible for local caching of data, anomaly marking, and scheduling uplink processing.

[0175] Step 1: Data Acquisition and Local Packaging

[0176] sensor nodes After periodically collecting raw environmental data such as temperature, light intensity, CO2 concentration, PM2.5, noise, and wind speed, the data package is encapsulated according to the following structure:

[0177] ={

[0178] ,

[0179] timestamp,

[0180] sensor_type is one of the following: temperature, light intensity, CO2 concentration, PM2.5, noise, or wind speed.

[0181] value, numerical value

[0182] nonce,

[0183] MAC=HMAC( , ||value||timestamp||nonce)

[0184] }

[0185] Step 2: Encrypt Upload

[0186] Data is encrypted using the symmetric encryption algorithm AES-CBC. The data packet is transmitted using a key-encrypted method and includes an HMAC checksum. After decryption, the platform verifies the MAC value to determine whether the data packet has been tampered with.

[0187] Step 3: Sensing Access Layer Processing

[0188] The perception access layer is responsible for:

[0189] Authentication: Confirm Valid and the key is effective;

[0190] Local caching: If data arrives faster than the processing capacity, it is first stored in the buffer;

[0191] Deduplication: Remove duplicate data within the time window;

[0192] Outlier labeling: Detecting sudden changes or outliers using a sliding statistical window and attaching anomaly labels;

[0193] Data scheduling: Pack and upload data to the data processing layer based on task priority and system load.

[0194] [Data Processing Layer] Context Modeling and Feature Encoding

[0195] The core objective of the data processing layer is to perform semantic enhancement, structural unification, and numerical normalization on the perceptual data input from the perceptual access layer, combining historical states and scene context, so as to provide structured input for the control reasoning process.

[0196] Step 1: Context Modeling Module

[0197] This module establishes a set of contextual tags based on multi-dimensional information such as environmental time (time period, holidays), historical weather (sunny, rainy, windy), and historical user behavior (whether anyone opened a window).

[0198] Ctx={

[0199] time_type: weekday / weekend / night,

[0200] weather_type: Sunny / Cloudy / Rainy

[0201] occupancy_flag: Occupancy / No one is present.

[0202] building_mode: Pollution protection priority / Ventilation priority

[0203] }

[0204] Each sensor data point is labeled with a corresponding context tag to dynamically adjust the adaptability of the control strategy.

[0205] Step 2: Feature Encoding Module

[0206] The collected sensor values ​​of various types are normalized and combined with contextual semantic labels to construct a unified environmental feature vector representation:

[0207] F=[Temp_norm,CO2_norm,Light_norm,PM2.5_norm,Noise_norm,Wind_norm,Ctx_vec]

[0208] Ctx_vec is the one-hot encoding or embedding vector of the context label, used to participate in the inference logic judgment.

[0209] [Control Decision Layer] Fuzzy Reasoning Process

[0210] A variable-structure fuzzy system is used to perform analytical reasoning on the environmental state vector to generate window opening control values. Compared with fixed-structure fuzzy systems, this scheme can activate different rule subsets according to the context, improving the adaptability and responsiveness of the control.

[0211] Step 1: Fuzzy Membership Function Mapping

[0212] The system designs a family of membership functions for each environmental factor. For example, temperature has membership functions such as "low temperature, medium temperature, high temperature" (which maps a certain temperature to low temperature, medium temperature, and high temperature) such as μ_T_low(x), μ_T_mid(x), and μ_T_high(x).

[0213] The numerical inputs in the environmental feature vector F are mapped to the corresponding membership space, forming a fuzzy input set.

[0214] Step 2: Dynamically activate the rule sub-library

[0215] Based on the context labels in the current Ctx_vec, select and activate the rule sub-library that best matches the current scenario. For example, in a summer / daytime / ventilation-priority scenario, the activated rule would be:

[0216] IF High temperature AND High CO2 concentration AND Low PM2.5 THEN Opening degree = 90% (opening degree is expressed as a percentage).

[0217] Step 3: Fuzzy Reasoning and Clear Output

[0218] The outputs of multiple rules are fused using fuzzy inference methods (such as weighted averaging or centroid method) to ultimately generate a window target opening value between 0 and 100%.

[0219] Open_ratio=∑( × ) / ∑( )

[0220] in Let i be the suggested opening value corresponding to the i-th rule. Its matching degree.

[0221] [Service Interaction Layer] User control command access and verification mechanism of remote command access module

[0222] This module supports users in issuing control commands to the opening of curtain wall windows via remote terminals (such as App / Web / voice devices). Since it involves an entry path from the outside into the control core, it requires two-way authentication, command integrity verification, and communication session tracking in conjunction with the user node key system to ensure that the control commands are from a reliable source, their content is not tampered with, and their transmission is not replayed.

[0223] Step 1: User instruction encapsulation

[0224] When a user sends a window opening control request on the client side, such as "set the opening to 65%", this request is encapsulated as a remote control command message:

[0225] CtrlReq={

[0226] ,

[0227] ,

[0228] ,

[0229] DesiredOpenRatio,###Openness

[0230] Nonce,

[0231] Sig=HMAC( , ||DesiredOpenRatio|| ||Nonce)

[0232] }

[0233] : The identity identifier of the initiating user

[0234] Initiating device identifier

[0235] Temporary session key bound to the device

[0236] Nonce: a random number to prevent replay.

[0237] Sig: Symmetric Key Message Authentication Code (HMAC)

[0238] Step Two: Platform Reception and Identity Verification

[0239] The remote command access module performs the following security verification process on this message:

[0240] 1. Extract And obtain its currently valid device key. ;

[0241] 2. Verify that the timestamp is within the valid time window (e.g., ±5 seconds);

[0242] 3. Verify if the nonce is appearing for the first time to prevent replay;

[0243] 4. Recalculate and verify the signature HMAC:

[0244] HMAC′=HMAC( , ||DesiredOpenRatio||Timestamp||Nonce)

[0245] If HMAC′≠Sig, the instruction is discarded and a security event is logged.

[0246] Step 3: Instruction Standardization and Trajectory Recording

[0247] After authentication, the system converts the control command into a standard format, writes it into the current user behavior trajectory chain, and attaches it to the current... This is to support subsequent tracking, auditing, and intent fusion processes.

[0248] [Control Decision Layer] Design and Implementation of the User Intent Fusion Mechanism for the User Intent Fusion Module

[0249] The user intent fusion module is used to coordinate the system's autonomously calculated opening suggestion value Open_ratio with the user-submitted target opening value DesiredOpenRatio to form a fused control value.

[0250] Step 1: Receive fused input

[0251] This module receives input from two sources:

[0252] DesiredOpenRatio: User-targeted opening degree from remote control commands (percentage value, 0–100).

[0253] Open_ratio: The system-recommended opening degree output by the fuzzy inference results of the control decision layer.

[0254] Use Open_ratio as the system's recommended opening value. ;

[0255] By combining DesiredOpenRatio with the user's recent overall opening tendency, opening preferences over different time periods, and the range of user preferred opening levels, a user behavior preference vector can be established to obtain the user control suggestion value. Alternatively, DesiredOpenRatio can be used directly as the user-controlled suggested value. .

[0256] Step 2: Design of the fusion strategy function

[0257] To ensure control flexibility and user experience, an algorithm was designed to obtain the final opening / closing degree. ,in, ∈[0.1,0.9] represents the dynamic fusion weight coefficient, indicating the system's credibility, which can be adjusted according to the following...

[0258] User equipment control frequency (high-frequency operation weight reduced);

[0259] Are there any risks in the environmental context (e.g., if the outside wind speed is high, then prevent the user from forcibly turning on all the power);

[0260]

[0261] in, Here is an example function for normalizing the magnitude of the deviation between the two. As a regulating factor. When the conflict between the two is small, Approaching 0.1; when the conflict is large, It increases, especially in harsh environments (which can be obtained from context labels), approaching 0.9.

[0262] Step 3: Write the fused output into the trajectory chain

[0263] The merged Open fused It will be written into the Session behavior trajectory and flow to the execution driver layer as formal control parameters, supporting subsequent debugging analysis and control correction modeling.

[0264] [Control Decision Layer] Execution Parameter Generation Module's Position Calibration and Damping Compensation Mechanism

[0265] Before generating window execution parameters, the control system must consider the dynamic difference between the actual window opening and the motor response behavior. Therefore, this module introduces a position calibration algorithm and a damping compensation mechanism to perform error correction and soft-start adjustment during the generation of opening control commands, ensuring accurate, smooth, and safe physical actions.

[0266] Step 1: Current position calibration

[0267] The system periodically, or before receiving control commands, obtains the actual opening degree of the current window via an angle encoder or Hall sensor. This value is related to the expected opening after the last command was executed. If the deviation exceeds δ=±2%, it is marked as an offset drift, and the execution offset record table is updated. This is used to generate the error inverse compensation term for the next action.

[0268] Step 2: Construction of Damping Compensation Model

[0269] Based on the window load characteristics, motor inertia, and environmental disturbances (such as wind speed), this invention designs a damping algorithm to dynamically adjust the motion speed:

[0270] exist Change to openness The process involves B+ offset drift;

[0271] Pre-start segment ( -10%), slow start;

[0272] Mid-range (10-90%), stable and uniform speed;

[0273] End paragraph (90% - target) ), the speed decreases.

[0274] Step 3: Final Parameter Encapsulation

[0275] The generated control parameters are encapsulated in the following data packet:

[0276] CtrlExec={

[0277] TargetOpenRatio:Open fused ,

[0278] OffsetCorrection: Δθ, offset drift

[0279] Damping: D, Damping Control

[0280] PWM_Signal,

[0281] Timestamp, time

[0282] }

[0283] This parameter flows into the drive control interface, whereby the actuator directly drives the window opening motor to complete the actual window status update.

[0284] [Control Decision Level] Position Calibration

[0285] After each system cold start or long-term operation

[0286] The window slowly closes to zero, and the current position is recorded as POS_0;

[0287] Drive at a constant speed to the maximum open position and record the maximum position encoding value pos_max;

[0288] Calculate the calibration interval Δ = pos_max - pos_0.

Claims

1. A curtain wall opening system based on Internet of Things, characterized in that: Comprise the following hierarchy: Perception access layer: this layer includes sensor interface module and data collection scheduling module, for connecting the internal and external sensor network in the building and structuring the collection of environmental data; Sensor interface module: for interfacing various types of sensor communication protocols; Data collection scheduling module: according to the set sampling period, scheduling various types of sensors to perform data collection tasks; the obtained data supports local buffering, and through the sliding window algorithm, the repeated data is de-duplicated, combined with statistical analysis to mark outliers, and finally generates perception data conforming to the internal data format, and sends to the data processing layer; Data processing layer: this layer includes context modeling module and feature encoding module, for structuring the collected perception data, generating environment feature encoding vector for control decision layer; Context modeling module: according to the historical environment sequence, time label, weather state context information, the current perception data is labeled and processed, and the environment context vector is constructed; Feature encoding module: merge the perception data and the environment context vector, and perform normalization, standard deviation compression and missing value interpolation processing operations to generate uniform format environment feature encoding vector and pass to the control decision layer; Control decision layer: this layer includes fuzzy reasoning control module, user intention fusion module and execution parameter generation module, for completing the final reasoning decision and control parameter issuing of window opening degree; Fuzzy inference control module: based on the environment feature coding vector, call variable structure fuzzy system to carry out multi-objective inference, this module dynamically selects different rule sub-bases, determines the priority target according to the current context, and then executes fuzzy logic inference to generate system recommended opening degree value ; User intention fusion module: receiving user control suggestion value from service interaction layer under the condition that remote control request exists , and performing fusion algorithm , wherein the fusion factor is determined by the current control weight strategy, and finally obtaining the system execution target opening degree value ; An execution parameter generation module: according to the fused target opening value , combined with the window historical motion state, the current execution position, and the damping factor information, generate accurate window control parameter instruction packages, and push to the execution driving layer for subsequent remote driving call; Service interaction layer: this layer includes remote instruction access module and control feedback module, for realizing the remote bidirectional interaction and control feedback closed loop between user and system; Remote command access module: This module receives remote control requests, including those from mobile apps, voice terminals, and web platform user control terminals. It first requests command authentication and identity verification from the security management layer. Once verified, it sends the recommended control value. The data is then transmitted back to the control decision-making level. Control feedback module: supervises and obtains the state of the window driven by the window control parameter instruction package, obtains the control execution result, encapsulates the control execution result as a control feedback report, and pushes it to the user control terminal according to the set period, supports state subscription, abnormal push and remote log persistent query; Security management layer: this layer includes node identity registration module, communication authentication and key negotiation module, session trajectory chain management module, for realizing identity authentication, security encryption and attack tracking of the whole system control process; Node identity registration module: according to the centralized registration center - master password generator architecture, for each user control terminal with the sensor node registration; Communication authentication and key negotiation module: adopts two-way challenge response mechanism for authentication of remote control request and data upload process, combines AES symmetric encryption and HMAC message digest to realize command content integrity and identity validity check, and generates a new session key for this interaction after each round of authentication negotiation; Session trajectory chain management module: generates a unique trajectory label Session_Tag for each control session, and constructs a trajectory chain structure based on time chain and identity chain, for abnormal tracking, attack positioning and instruction tracing, supports tracking revocation mechanism and key rolling update. 2.The remote linkage control method of the IoT-based curtain wall fenestration system according to claim 1, characterized in that: The sensor node registration process in the security management layer comprises the following steps: Step one: identity initialization, for each deployed sensor node Generate a globally unique identifier The identifier can be generated based on the device factory code, geographic location information and device type hash, ensuring no duplication in the network; Step two: master key binding, gateway node For each sensor node Assign an initial master key , as the source key of identity authentication, in the secure channel, the gateway node GN calculates , the generated is the master key of the sensor in subsequent communication, used for data encryption and authentication; Step three: local storage and destruction, master key are stored in the gateway node and the sensor node respectively The gateway node will establish an index mapping for subsequent authentication, while the initial master key is immediately destroyed, ensuring the unidirectionality of the key to avoid subsequent leakage leading to network leakage Step four: registration confirmation and trajectory label initialization, after completing the registration, the gateway node issues an initial communication trajectory label to the sensor node as the starting point for the sensor node to participate in system communication, which will be updated subsequently with session rolling for communication behavior tracking and key update index; ​ Step five: registration log writing and abnormal audit chain hanging, this registration event is written into the system registration log and hung in the system abnormal audit chain, ensuring that at any future time point, the access time, key state and registration behavior of the sensor node can be traced, and a complete security life cycle management mechanism is constructed. 3.The remote linkage control method of the IoT-based curtain wall fenestration system according to claim 2, characterized in that: The user registration process in the security management layer comprises the following steps: Step one: user registration request submission, the user initiates a registration request through a user control terminal, and submits the request to the gateway node by submitting its unique user identification and custom password ; the security management layer performs uniqueness verification to prevent duplicate registration, and performs legality verification on password complexity and input specification ​ Step two: key derivation is bound to identity, using salted hash function to make irreversible processing on password wherein, is a random salt factor generated by the system for each user, the identity key is used for subsequent user control terminal and system between all control message authentication and HMAC signature; the , and the identity key after salting hash is saved in the authentication database, and a user identity state table is established to record registration time, initial device, login IP auxiliary metadata; Step three: control terminal binding and authentication credential configuration, the system supports users to bind multiple user control terminals at the same time, and each user control terminal is allocated an independent terminal session key by the platform when it logs in for the first time , and based on derivation: The key is used to establish a TLS or symmetric encryption channel with the terminal; the terminal credential configuration package contains device ID, terminal type, binding timestamp, and signed credentials, ensuring the legality and identifiability of the user control terminal communication identity; Step 4: Track tag and access control initialization, the system is... Initialize session trajectory tags And establish a user access control list to record the user's permission level; trajectory tags It will serve as the context reference point for all subsequent communication behaviors and be linked to the system's anomaly audit chain to identify abnormal instructions, authenticate forged behavior, or illegal instruction replay security risks. Step five: client security prompt and key rotation mechanism activation, after the user registers successfully, the user control terminal receives the registration success notification and terminal authentication credentials, and prompts the user to configure the second security verification to enhance security, at the same time, the system activates the key rotation strategy, and triggers the identity key and session key updating process according to the set period or specific events, to improve the forward security and attack resistance ability during long-term use. 4.The remote linkage control method of the IoT-based curtain wall fenestration system according to claim 3, characterized in that: The user login authentication process in the security management layer includes the following steps: Step one: login request submission, the user initiates a login operation at the user control terminal, and submits with a custom password , and the user control terminal sends an authentication request to the security management layer after local preprocessing, and the request message includes: user ID, salted password digest, client device identification , timestamp and request random number; Step two: password verification and anti-replay check, find the associated random salt factor with identity key and calculate the hash value of login input , if , reject the login request and record the failed attempt, support to set the login failure threshold and temporary lock; if consistent, check whether the timestamp and nonce have been used to prevent replay attacks;​ Step three: terminal session key negotiation, after authentication, the system generates a current session key for the device and establishes a TLS channel or a secure channel based on symmetric encryption, which is used for data encryption and instruction signature when the user sends control instructions; ​ Step four: Session track update and behavior audit initialization, the system updates the user's session track label and starts login session logging, including device, IP, location, login time, etc., written to the security audit chain to support abnormal behavior analysis and post-factum responsibility tracking. 5.The remote linkage control method of the IoT-based curtain wall fenestration system according to claim 3, characterized in that: The sensor data uploading in the perception access layer and the perception access layer processing process include the following steps: Step one: data collection and local encapsulation, sensor node After collecting the original environmental data of temperature, light, CO2 concentration, PM2.5, noise, and wind speed periodically, the data packet is encapsulated; Step two: encryption upload, data is transmitted by symmetric encryption algorithm AES-CBC with a key, with an accompanying HMAC check code, and the platform judges whether the data packet has been tampered with by verifying the MAC value after decryption. Step three: perception access layer processing, the perception access layer is responsible for: Identity verification: Confirmation Legit, key valid; Local cache: if the data arrives faster than the processing capacity, it is stored in the buffer first; De-duplication processing: repeated data in the time window is removed; Outlier marking: use a sliding statistical window to detect mutations or outliers, and attach an exception label; Data scheduling: according to the task priority and system load, the data is packaged and uploaded to the data processing layer. 6.The remote linkage control method of the IoT-based curtain wall fenestration system according to claim 5, characterized in that: Context modeling and feature encoding in the data processing layer include the following steps: Step one: the context modeling module establishes a set of context labels based on environmental time, historical weather, and historical user behavior multi-dimensional information: Step two: the feature encoding module normalizes the collected multi-type sensor values and constructs a unified environmental feature vector combined with the context labels. 7.The remote linkage control method of the IoT-based curtain wall fenestration system according to claim 6, characterized in that: The fuzzy reasoning process of the control decision layer includes the following steps: Step one: design a membership function family for each environmental factor, the numerical input in the environmental feature vector F is mapped to the corresponding membership degree space, forming a fuzzy input set; Step two: according to the context label, select the rule sub-library that best matches the current scene; Step three: use fuzzy reasoning method to fuse the outputs of multiple rules, finally generate a window target opening value between 0-100%. 8.The remote linkage control method of the IoT-based curtain wall fenestration system according to claim 7, characterized in that: User control instruction access and verification of the remote instruction access module in the service interaction layer include the following steps: Step one: the user sends a window opening control request on the user control terminal, which is packaged as a remote control instruction message; Step two: the remote instruction access module remotely controls the instruction message to execute the following security check process:

1. Extract and obtain its current valid device key ; 2. Check if the timestamp is within the valid time window; 3. Check if the nonce appears for the first time to prevent replay; HMAC′=HMAC( , If HMAC′≠Sig, then discard the instruction and record the security event; Step three: After authentication, the system converts the control instruction into a standard format, writes it into the current user behavior track chain, and hooks the current to support subsequent tracking, auditing, and intention fusion processes. 9.The remote linkage control method of the IoT-based curtain wall fenestration system according to claim 8, characterized in that:

4. Recalculate the signature HMAC' and verify: Step 1: This module receives inputs from two sources: DesiredOpenRatio: the user's target opening degree from remote control instructions; Open_ratio: the system's recommended opening degree output by the fuzzy inference results of the control decision layer; Open_ratio is taken as the system's recommended opening degree value ; DesiredOpenRatio is combined with the user's recent overall opening degree tendency, opening degree preferences in different time periods, and the user's preferred opening degree amplitude range to establish a user behavior preference vector to obtain the user control recommendation value , or directly take DesiredOpenRatio as the user control recommendation value ; Step two: design algorithm to get the final opening degree wherein, ∈ [0.1, 0.9] is a dynamic fusion weight coefficient, representing the credibility of the system; Step three: fusion of Open fused The write Session behavior trajectory is written into the formal control parameter flow to the execution driver layer, supporting subsequent debugging analysis and control correction modeling. 10.The remote linkage control method of the IoT-based curtain wall fenestration system according to claim 9, characterized in that: The fusion mechanism design and implementation of the user intention fusion module in the control decision layer include the following steps: Step 1: The system acquires the actual opening of the current window body through an angle encoder or a Hall sensor periodically or before receiving a control command The value is compared with the expected opening after the last command execution If the deviation exceeds δ = ± 2%, it is marked as a drift, and the execution offset record table is updated for generating an error back compensation term for the next action Step 2: According to the window load characteristics, motor inertia and environmental disturbance, design damping algorithm to dynamically adjust the action speed: in the process of opening degree B+ offset drift; the front starting section -10%, slow start; middle 10-90%, stable uniform speed; end section 90%-target , speed reduction; The position calibration and damping compensation mechanism in the execution parameter generation module of the control decision layer includes the following steps: Step three: final execution parameter packaging, the parameter flows into the drive control interface, and the actuator directly drives the window motor to act, completing the actual window state update.