Intelligent edge calculation method and system supporting multi-protocol perception and rule-driven processing

Abstract

The invention discloses an edge intelligent computing method and system supporting multi-protocol perception and rule-driven processing, and relates to the technical field of edge intelligent computing, and the method comprises the steps: recognizing protocol features through monitoring communication behaviors, and connecting an animal model structure in parallel to complete semantic registration; collecting a data stream based on a semantic registration field structure, and performing rule judgment on a semantic registration field; and if rule judgment meets the trigger condition, executing the function chain bound with the event, outputting a response result, and feeding back an operation state. According to the method, a complete edge calculation process of protocol feature recognition, field semantic registration, rule-driven calculation and function chain response processing is constructed, so that equipment plug-and-recognize, data structured processing and event intelligent response are realized, and the method has high adaptability, high real-time performance and cloud edge co-evolution capability; and the autonomy and the response efficiency of the edge intelligent system are obviously improved.

Description

Technical Field

[0001] The present invention relates to the field of edge intelligent computing technology, and specifically to an edge intelligent computing method and system that supports multi-protocol perception and rule-driven processing. Background Art

[0002] With the rapid development of the Industrial Internet of Things (IIoT) and edge computing, the edge has placed higher demands on protocol compatibility, data processing, and event response for multi-source heterogeneous devices. Traditional centralized processing architectures are no longer able to meet the application requirements of low latency, high throughput, and local autonomy. In recent years, researchers and engineering practices have gradually promoted the integration of key technologies such as "protocol parsing automation," "physical model-driven management," and "edge rule execution engines," driving the evolution of edge intelligence from static deployment to dynamic perception, rule-driven, and real-time linkage. In practical applications, how to achieve automatic device identification, structured semantic parsing, and complex conditional judgment on the edge has become the core foundation for ensuring industrial real-time control and intelligent response capabilities.

[0003] Although some edge computing systems currently support multi-protocol data access and rule-based judgment, they still suffer from widespread issues such as poor adaptability, delayed triggering mechanisms, and fragmented rule execution processes. Existing systems often rely on manually predefined communication protocol configuration files and lack the ability to identify protocol features, resulting in long new device access cycles and difficulty implementing automatic modeling and semantic recognition. While some platforms have introduced rule-based judgment mechanisms, the structure of rule expressions is generally simplified, lacking the expressive power of time windows, multi-field interactions, and contextual awareness, making it difficult to adapt to complex business scenarios. Existing event response strategies often rely on single rules mapping to single instructions, lacking the ability to chain function execution and state feedback, resulting in coarse response granularity and a lack of connectivity. Summary of the Invention

[0004] In view of the above-mentioned problems, the present invention is proposed.

[0005] Therefore, the technical problems solved by the present invention are: the existing edge data processing and event judgment methods lack an automatic protocol recognition mechanism, the rule expression capability is weak and cannot adapt to complex timing logic, the event response is disconnected from function processing, and how to realize a closed-loop self-driven intelligent computing process from device identification, rule judgment to response execution on the edge side.

[0006] To solve the above technical problems, the present invention provides the following technical solutions: an edge intelligent computing method that supports multi-protocol perception and rule-driven processing, including identifying protocol features by monitoring communication behavior, and completing semantic registration in parallel with the animal model structure; collecting data streams based on the semantically registered field structure, and performing rule judgment on the semantically registered fields; the rule judgment satisfies the trigger conditions, executes the function chain bound to the event, outputs the response result, and feeds back the running status; the object model structure includes the naming rules, numerical types, logical hierarchies and operation method sets of the attribute fields; the rule judgment includes traversing the archived cache data in chronological order, each time window as a rule judgment cycle, executing computing tasks within each cycle, making real-time judgments on the rule conditions and generating judgment results, substituting the field values ​​in the data stream into the corresponding rule expressions, and performing Boolean calculation operations.

[0007] As a preferred solution of the edge intelligent computing method supporting multi-protocol perception and rule-driven processing described in the present invention, the communication monitoring behavior includes extracting the frame start character, field offset structure, encoding format and verification rules in the device communication data, and constructing a protocol feature vector; the protocol feature vector is matched with the preset protocol family to identify the communication protocol type.

[0008] As a preferred solution of the edge intelligent computing method supporting multi-protocol perception and rule-driven processing described in the present invention, wherein: the identification of the communication protocol type includes calling the object model template corresponding to the communication protocol, the object model template includes the attribute name, data type, logical hierarchical structure and supported operation method definition of the connected device field, and is used to generate a field parsing structure.

[0009] As a preferred solution of the edge intelligent computing method supporting multi-protocol perception and rule-driven processing described in the present invention, the collected data stream includes a field structure based on semantic registration, which collects device communication data according to a preset sampling period, parses it into a structured format, and encapsulates it into a standardized data record containing field name, value, timestamp and device identification.

[0010] As a preferred solution of the edge intelligent computing method supporting multi-protocol perception and rule-driven processing described in the present invention, the rule judgment includes: structured data records are organized into a time window buffer area, each window serves as a rule judgment cycle, and the rule judgment traverses and analyzes the recorded data within each cycle.

[0011] As a preferred solution of the edge intelligent computing method supporting multi-protocol perception and rule-driven processing described in the present invention, each window is used as a rule judgment cycle, including substituting each field value into the rule expression associated with the field, combining the historical average and rate change timing information, and performing a Boolean calculation to determine whether the event triggering condition is met.

[0012] As a preferred solution of the edge intelligent computing method supporting multi-protocol perception and rule-driven processing described in the present invention, the event triggering conditions are met, and the events that meet the triggering conditions are bound to a function set, the function execution order is selected according to the event type, the output of the previous function is used as the input of the next function, and a response result is generated after the chain function is executed.

[0013] As a preferred solution of the edge intelligent computing method supporting multi-protocol perception and rule-driven processing described in the present invention, the response results include event level classification, function call trace, processing time, field labels and response conclusions, and are output in a structured data format for recording and scheduling reference.

[0014] As a preferred solution of the edge intelligent computing method supporting multi-protocol perception and rule-driven processing described in the present invention, the operating status includes response accuracy, event triggering frequency and rule hit rate.

[0015] Another object of the present invention is to provide an edge intelligent computing system that supports multi-protocol perception and rule-driven processing. It can collect data streams through a field structure based on semantic registration and perform rule judgment on the semantically registered fields, solving the problem that the current edge data processing and event judgment methods contain weak rule expression capabilities and cannot adapt to complex temporal logic.

[0016] As an optimal solution for the edge intelligent computing system supporting multi-protocol perception and rule-driven processing described in the present invention, it includes: a communication protocol feature recognition and object model linkage registration module, a structured field-based data collection and rule judgment module, and an event response function execution and operation status feedback module; the communication protocol feature recognition and object model linkage registration module is used to extract communication protocol feature information by monitoring device communication behavior, and match the preset object model template based on the feature; the structured field-based data collection and rule judgment module is used to continuously collect device data streams based on the registered field structure, organize them into cached data blocks according to time windows, and substitute the field values ​​into the rule expression for real-time Boolean logic calculations; the event response function execution and operation status feedback module is used to execute the function chain task bound to the event if the rule conditions are met, output the response conclusion, and construct an operation summary for version feedback and cloud-edge collaborative updates of rules and models.

[0017] A computer device includes a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to implement a step of an edge intelligent computing method that supports multi-protocol perception and rule-driven processing.

[0018] A computer-readable storage medium stores a computer program, which, when executed by a processor, implements the steps of an edge intelligent computing method that supports multi-protocol perception and rule-driven processing.

[0019] Beneficial effects of the present invention: The edge intelligent computing method that supports multi-protocol perception and rule-driven processing provided by the present invention identifies protocol features by monitoring communication behaviors and automatically matches object model structures, breaking through the limitations of traditional reliance on manual configuration of protocols and field rules, and realizing plug-and-play recognition of devices and automatic binding of field semantics, effectively improving the access efficiency and management consistency of heterogeneous devices. Secondly, through a data collection mechanism based on a registered field structure, the collected values ​​are injected into the rule expression, and Boolean judgments are performed in combination with features such as time windows, field offsets and historical trends to construct a rule-driven computing capability that supports context perception, thereby meeting the event recognition requirements under complex conditions. Finally, the response tasks associated with the event are linked and executed through a function chain, and the operating status is fed back in the form of a structured summary to drive cloud-side policy collaboration and version evolution. The overall solution not only improves the real-time decision-making capability and local intelligent response depth of edge computing, but also builds an edge intelligent architecture that is collaborative and adaptively evolved with the cloud and edge, with significant advantages such as strong deployment flexibility, transparent response mechanism, and scalable rule management. BRIEF DESCRIPTION OF THE DRAWINGS

[0020] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the following briefly introduces the drawings required for use in the description of the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0021] Figure 1 An overall flow chart of an edge intelligent computing method supporting multi-protocol perception and rule-driven processing provided for the first embodiment of the present invention. DETAILED DESCRIPTION

[0022] To make the above-mentioned objects, features, and advantages of the present invention more clearly understood, the following detailed description of the specific embodiments of the present invention is given in conjunction with the accompanying drawings. It is obvious that the described embodiments are only part of the embodiments of the present invention, not all of them. Based on the embodiments of the present invention, all other embodiments obtained by ordinary persons in this field without creative work should fall within the scope of protection of the present invention.

[0023] Example 1, with reference to Figure 1 , as one embodiment of the present invention, provides an edge intelligent computing method that supports multi-protocol perception and rule-driven processing, including: S1: Identify protocol features by monitoring communication behaviors and complete semantic registration in parallel with animal model structures.

[0024] Furthermore, the data exchange behavior during the device's first communication is monitored, and communication features such as the start of frame character, field offset pattern, data encoding format, and checksum method are extracted to form a protocol feature vector. This vector is used to characterize the type of communication protocol used by the device.

[0025] It should be noted that the extracted protocol feature vectors are matched with predefined protocol feature clusters, and the protocol category to which the communication behavior belongs is identified through similarity measurement, and then classified into a specific protocol family (such as Modbus, MQTT, OPC, etc.). The identification behavior is based on communication patterns rather than protocol identification fields, and is adaptive.

[0026] It should also be noted that based on the identified protocol category, a data structure template matching the protocol family is selected from the locally defined device object model set. This template includes naming rules, value types, logical hierarchies, and a set of operational methods for multiple attribute fields, which are used to standardize the format of subsequent data collection and processing.

[0027] Furthermore, based on the selected template, a unique semantic structure description is generated for the device, including field encoding and semantic tag mapping rules. A unique identifier is constructed based on the device's network attributes (such as IP, port, and communication path) and entered into the list of pending devices as a registration entry. It should be noted that after completing the above identification and modeling, the device identifier, field structure, data format, and the model structure of the attached object are packaged into a unified data record format and stored in the logical processing list for direct reference in the data collection and rule calculation steps, without the need for human intervention.

[0028] S2: Collect data streams based on the semantically registered field structure and perform rule judgment on the semantically registered fields.

[0029] Furthermore, after completing device protocol identification and object model binding, the system continuously collects communication data in real time, parses the protocol, and normalizes the format to support standard edge data access in high-frequency scenarios. Based on the generated device identifier and field structure definition, the system continuously receives uplink data frame content from the device through the corresponding communication channel according to the set sampling period. The acquisition frequency can reach seconds or milliseconds to meet the needs of high-frequency signal processing.

[0030] It should be noted that protocol-level parsing is performed on the received data frame to identify the address offset, data type, starting position, and parsing method corresponding to the field content. Parsing is performed based on the field template defined in the binding model, ensuring semantic consistency during the extraction process.

[0031] It should also be noted that the field values ​​parsed from each data frame are associated and mapped with their logical labels, and the results are organized into a unified structured format (such as key-value pairs or standard JSON structure), with the collection timestamp, device ID and sampling batch number attached.

[0032] Furthermore, the normalized data is archived and temporarily buffered according to a set time window. The time window can be set as a sliding window, a jumping window, or an event-based dynamic window. The buffered content is used as input for subsequent rule triggering and streaming calculation steps.

[0033] It should be noted that the constructed window cache data is marked with available tags and declared as an input data stream that can be called by the rule calculation engine, supporting direct reference of functional modules such as event triggering and expression condition judgment to establish a data-driven processing chain.

[0034] It should also be noted that the predefined set of rule expressions is read. Each rule consists of a Boolean condition, a threshold setting, comparison logic, and a trigger action identifier. The referenced field names and data structures are derived from the cached data format; each rule is bound to a unique device identifier to ensure the calculation target is clear and the context is consistent.

[0035] Furthermore, the archived cache data is traversed in chronological order according to the defined time window division method. Each time window is regarded as a rule judgment cycle; within each cycle, parallel rule calculation tasks for single devices and multiple fields are executed.

[0036] It should be noted that substituting the field values ​​in the data stream into the corresponding rule expression and performing the Boolean calculation operation is expressed as: ; in, Represents a regular expression function based on the linkage of multiple fields in a data stream. is the sampling value of the main field in the current data stream, is the second field value (such as temperature or current) in the same cycle, For fields Sliding history sampling sequence, For sequence The sliding mean of For sequence The sliding standard deviation of is the sampling value of the second field in the previous time period, For fields The rate of change of is the first-order difference value, To prevent small constants with zero denominators (such as ), This is the lower threshold constant for rule triggering, usually set to 0.85 to control the sensitivity of the event.

[0037] The value range is ,when When , it is determined that the rule expression is satisfied and the event is triggered; when When , only logs are recorded and no events are triggered; a value close to 1 indicates that the current data state is highly abnormal or highly consistent with the rules, and a value close to 0 indicates deviation from the rules.

[0038] If the judgment result is "true", that is, the condition is met, a structured event judgment result is generated immediately; if the condition is not met, no event is output in the current cycle, and only the log status is retained.

[0039] The event determination result includes: event type, trigger time, trigger field, trigger value and corresponding device ID, which are used for subsequent execution processing.

[0040] It should also be noted that, based on the event type specified in the event determination result, a preset set of function identifiers is searched and associated. Functions can be diagnostic functions, anomaly label generators, predictive reasoning models, early warning strategy functions, and so on. Functions are not specifically executed in this step; only call preparation and parameter encapsulation are completed here. The encapsulated function call task is packaged into an instruction stream and output along with the event for use in the next step. The event result and the bound function instruction stream are encapsulated together as an "event-function pair" and pushed to subsequent processing stages in the form of a data structure, achieving a logical transition between rule-based judgment and intelligent processing.

[0041] S3: The rule determines whether the trigger condition is met, executes the function chain bound to the event, outputs the response result, and feedbacks the running status.

[0042] Furthermore, for events that meet the conditions, the corresponding function instruction structure is determined by the event type and parameter domain. The matching response function set is selected based on the field label, value intensity, trigger time and other information contained in the event structure. , where each function is defined as; ; in, Represents a set of matching response functions, Indicates the absolute value of the main field in the event judgment Indicates the deviation of the field from the historical average. Indicates the rate of change when an event occurs.

[0043] It should be noted that if the event falls into the category of high priority chain response (such as prediction control record), the current function The output of the next function Each level of function can transform dimensions or perform inference and classification, ultimately generating a comprehensive response result.

[0044] It should also be noted that the final first-level function chain output value is compared with the preset response threshold: if it is greater than the judgment threshold , the event is marked as severe and a control command is generated; if it is in the middle range, a medium-level record label is generated; if it is less than the lower threshold, it is marked as minor or ignored.

[0045] After the local event response is completed, in order to improve the long-term adaptability of rule configuration and model strategy, a collaborative method based on version tracking and policy synchronization is adopted to achieve continuous consistency and secure evolution between edge processing logic and remote control strategy.

[0046] The generated event response information, function call traces, output labels, and response levels are aggregated to construct periodic run summary data, which includes field statistics, event counts, response success rates, and data coverage. This summary is used to evaluate the effectiveness of current rules and the execution status of model behaviors.

[0047] Hash the currently deployed rule expression structure and function chain logic to generate a globally unique version fingerprint number; at the same time, record its loading time, the most recent modification record, and the version number of the cloud source file to form a complete version triple: local version number, local hash code, and the corresponding cloud original version number.

[0048] Based on the performance data reported in the run summary, the system determines whether the current rules are experiencing response delays, increased false positives, or redundant events. If any preset thresholds are triggered, the system requests a remote update recommendation through the cloud-edge communication channel and receives the new version of the rule expression, function structure, or policy parameter set.

[0049] The update operation is controlled by the version identifier and is only executed when the cloud version number is higher than the local version number, or when the local version is marked as "pending update status".

[0050] The received new version undergoes a structural consistency check and instruction specification conversion, then replaces the current local version. A record of the update is generated, including a summary of the update content, the previous and next version numbers, the replacement time, the execution result, and whether rollback protection was triggered. If the update fails or is incompatible with the local data structure, a rollback is triggered, restoring the previous valid version.

[0051] After the update is complete, the new rule configuration and function chain will be registered in the edge configuration list, along with the version identifier and loading information. This information will be used as a historical baseline for performance evaluation in subsequent run summaries, supporting future multi-version comparisons and dynamic adjustments.

[0052] Example 2 is an embodiment of the present invention, which provides an edge intelligent computing system that supports multi-protocol perception and rule-driven processing, including a communication protocol feature recognition and object model linkage registration module, a structured field-based data collection and rule judgment module, and an event response function execution and operation status feedback module.

[0053] Among them, the communication protocol feature identification and object model linkage registration module is used to extract communication protocol feature information by monitoring device communication behavior, and construct a protocol feature vector based on the extraction results; compare the protocol feature vector with the preset protocol family to identify the communication protocol type; then match the corresponding object model template according to the identification result to complete the registration of device field structure, field naming rules, numerical type, logical hierarchy and operation method; generate a unique identifier based on the device network attributes, and write it into the semantic structure list for downstream data collection and rule calculation process to call.

[0054] It should also be noted that after extracting device communication features and identifying protocol types, the Communication Protocol Feature Identification and Object Model Linkage Registration module loads the corresponding object model template based on the identification results and generates a complete semantic description structure containing field structures, data types, naming tags, and operation methods. This structure, combined with the device's network attributes, generates a unique identifier and encapsulates it as standard structured registration information for direct reference by subsequent modules. This enables the automatic connection of field templates, protocol parsing methods, and rule parameter configuration, eliminating manual configuration operations and improving the system's automatic deployment capabilities and device adaptability.

[0055] The data collection and rule judgment module based on structured fields is used to continuously collect the data frame content uploaded by the device based on the communication protocol feature recognition and semantic registration results; parse the field values ​​according to the set sampling period and time window mode, perform protocol-level decoding, unit conversion and type normalization processing, and output structured data records; further, substitute the field values ​​in the data stream into the preset rule expression, perform Boolean logic calculations based on the field change trend within the time window, determine whether the event triggering conditions are met, and construct an event-function binding structure for subsequent response processing.

[0056] It should also be noted that the structured field-based data collection and rule judgment module utilizes the semantic structure and field configuration provided by the registration module for high-frequency data collection. The collected data stream is organized according to a time window cache to form a standardized field data structure. During the rule judgment phase, the system dynamically injects each field value into the rule expression, performs Boolean logic calculations, and determines whether the event triggering conditions are met. When the rule conditions are met, the system establishes an event-function binding relationship, binding the event type, field label, trigger strength, and other content to the function identifier, encapsulating it into an event-function pair structure and pushing it to the response processing module.

[0057] The event response function execution and operation status feedback module is used to receive the event-function binding structure generated based on rule judgment, select the matching function set according to the event type, execute the response function tasks in sequence in a chain order, and output the response result and processing label; at the same time, it collects the function trajectory, processing time, event level and trigger field during the response execution process, builds an operation status summary, and feeds it back to the remote management end; when the system receives a new version of the policy issued by the cloud, it executes policy replacement, function chain upgrade or automatic rollback operations based on the version fingerprint and verification mechanism to ensure that local rules and models are continuously optimized and coordinated with the cloud and edge.

[0058] It should also be noted that after receiving the event-function pair, the event response function execution and operation status feedback module selects and executes the corresponding function chain task according to the event type, and outputs the processing result, response label and call path. At the same time, the system records the response delay, event level and field participation information during the function execution process, and builds an operation status summary for the cloud system to perform rule performance evaluation and policy adjustment. When the system receives a new version of the rule configuration or function set issued by the cloud, the module performs a difference check based on the local version fingerprint information to ensure structural compatibility before performing version replacement or rollback, completing the dynamic evolution of the edge node's policy and the cloud-edge collaborative update closed loop.

Claims

1. An edge intelligent computing method that supports multi-protocol perception and rule-driven processing, characterized in that: include: By monitoring communication behaviors to identify protocol features, semantic registration is completed in parallel with the animal model structure; Collect data streams based on the semantically registered field structure and perform rule judgment on the semantically registered fields; The rule determines whether the trigger condition is met, executes the function chain bound to the event, outputs the response result, and feedbacks the running status; The object model structure includes the naming rules, value types, logical levels, and operation method sets of attribute fields; Rule judgment includes traversing the archived cache data in chronological order, with each time window as a rule judgment cycle, executing calculation tasks within each cycle, making real-time judgments on rule conditions and generating judgment results, substituting field values ​​in the data stream into corresponding rule expressions, and performing Boolean calculation operations.

2. The edge intelligent computing method supporting multi-protocol perception and rule-driven processing according to claim 1, characterized in that: The communication monitoring behavior includes: Extract the frame start character, field offset structure, encoding format and verification rules in the device communication data, and construct the protocol feature vector; The protocol feature vector is matched with the preset protocol family to identify the communication protocol type.

3. The edge intelligent computing method supporting multi-protocol perception and rule-driven processing according to claim 2, characterized in that: The identification of the communication protocol type includes: The object model template corresponding to the communication protocol is called. The object model template includes the attribute name, data type, logical hierarchical structure and supported operation method definition of the connected device field, and is used to generate a field parsing structure.

4. The edge intelligent computing method supporting multi-protocol perception and rule-driven processing according to claim 3, characterized in that: The collected data stream includes: Based on the semantically registered field structure, device communication data is collected according to the preset sampling period, parsed into a structured format, and encapsulated into standardized data records containing field name, value, timestamp and device identification.

5. The edge intelligent computing method supporting multi-protocol perception and rule-driven processing according to claim 1, characterized in that: The rule judgment includes: Structured data records are organized into time window buffers, with each window serving as a rule judgment cycle. Rule judgments traverse and analyze the recorded data within each cycle.

6. The edge intelligent computing method supporting multi-protocol perception and rule-driven processing according to claim 5, characterized in that: Each window as a rule judgment cycle includes: Substitute each field value into the rule expression associated with the field, combine the historical average and rate change timing information, and perform a Boolean calculation to determine whether the event triggering conditions are met.

7. The edge intelligent computing method supporting multi-protocol perception and rule-driven processing according to claim 6, characterized in that: The event triggering conditions are met, Events that meet the triggering conditions are bound to a set of functions. The function execution order is selected according to the event type, the output of the previous function is used as the input of the next function, and a response result is generated after the chain function is executed.

8. The edge intelligent computing method supporting multi-protocol perception and rule-driven processing according to claim 1, characterized in that: The response result includes: Event level classification, function call trace, processing time, field labels, and response conclusions are output in a structured data format for recording and scheduling reference.

9. The edge intelligent computing method supporting multi-protocol perception and rule-driven processing according to claim 1, characterized in that: The operating status includes: Response accuracy, event triggering frequency and rule hit rate.

10. An edge intelligent computing system supporting multi-protocol perception and rule-driven processing, characterized by: It includes a module for identifying communication protocol features and registering object models, a module for collecting data and judging rules based on structured fields, and a module for executing event response functions and providing feedback on the running status. The communication protocol feature recognition and object model linkage registration module is used to extract communication protocol feature information by monitoring device communication behavior and match the preset object model template based on the features; The structured field-based data acquisition and rule judgment module is used to continuously collect device data streams based on the registered field structure, organize them into cached data blocks according to time windows, and substitute field values ​​into rule expressions to perform real-time Boolean logic calculations; The event response function execution and operation status feedback module is used to execute the function chain task bound to the event if the rule conditions are met, output the response conclusion, and build an operation summary for version feedback and cloud-edge collaborative updates of rules and models.

11. A computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: When the processor executes the computer program, the steps of the edge intelligent computing method supporting multi-protocol perception and rule-driven processing described in any one of claims 1 to 9 are implemented.

12. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the edge intelligent computing method supporting multi-protocol perception and rule-driven processing described in any one of claims 1 to 9 are implemented.

Images