A communication data management system and method based on intelligent fusion terminal

Abstract

This invention belongs to the field of communication technology, specifically a communication data management system and method based on an intelligent converged terminal. The system includes a multi-dimensional resource perception module, a kernel processing bottleneck analysis module, a communication protocol stack dynamic adjustment module, and a computing and communication fusion scheduling module. The multi-dimensional resource perception module monitors the status across layers, the kernel processing bottleneck analysis module quantifies resource contention and outputs reports, the protocol stack dynamic adjustment module intervenes in message strategies based on reports, and the computing and communication fusion scheduling module performs cross-layer resource coordination. This application solves the problems of intra-terminal blocking and communication latency jitter caused by the decoupling of computing and communication resources, and ensures the deterministic transmission of communication data and the dynamic balance of computing tasks under high load conditions.

Description

Technical Field

[0001] This invention belongs to the field of communication technology, specifically a communication data management system and method based on an intelligent converged terminal. Background Technology

[0002] With the deep integration of the energy internet and the industrial internet of things, intelligent converged terminals, as core nodes integrating multiple functions, have become key infrastructure for digital transformation. They need to undertake high computing intensity tasks and evolve from a single communication gateway to a form of coupled computing and communication, posing extremely high challenges to system kernel resource scheduling and communication data management.

[0003] Existing communication management solutions focus on network layer quality of service assurance and solve link congestion through traffic shaping and priority labeling. They assume that the terminal only performs simple transparent forwarding, but this is no longer applicable in the actual working conditions of intelligent converged terminals. The computing tasks will consume a lot of system resources and affect communication processing.

[0004] The existing communication management and terminal computing resource allocation logic are decoupled, which causes intra-terminal blocking and communication latency jitter. Furthermore, there is a lack of a multi-dimensional resource awareness framework across multiple layers. Static priority strategies cannot resolve resource contention conflicts and are difficult to improve communication determinism under high load conditions. Therefore, this invention provides a communication data management system and method based on intelligent converged terminals. Summary of the Invention

[0005] In order to overcome the shortcomings of the prior art, at least one technical problem raised in the background art is solved.

[0006] The technical solution adopted by this invention to solve its technical problem is as follows: A communication data management system based on an intelligent fusion terminal, comprising a multi-dimensional resource perception module, a kernel processing bottleneck analysis module, a communication protocol stack dynamic adjustment module, and a computing and communication fusion scheduling module. The multi-dimensional resource perception module is configured to perform real-time status monitoring across the system kernel layer, computing application layer, and communication link layer. Specifically, this module collects the CPU time slice occupancy rate, system bus bandwidth utilization rate, and concurrent request volume of the memory direct access controller in real time through a resource hook program deployed in the system kernel layer. At the computing application layer, the module synchronously acquires the execution status of edge computing tasks, including the trigger frequency of high-frequency sampling tasks, the computational load of deep learning inference tasks, and the memory page replacement frequency of image recognition tasks. At the communication link layer, the module monitors the link utilization rate, signal strength, and round-trip latency of remote nodes of the physical interface in real time. The multi-dimensional resource perception module spatiotemporally aligns the perceived physical resource parameters with the logical task status to construct a multi-dimensional resource status vector, providing basic data support for subsequent bottleneck analysis. Preferably, the kernel processing bottleneck analysis module is connected to the multi-dimensional resource perception module. This module is configured to deeply analyze the resource contention mechanism within the intelligent fusion terminal. The kernel processing bottleneck analysis module receives the multi-dimensional resource state vector and applies preset bottleneck discrimination logic to quantify the nonlinear interference generated by the calculation task on the communication processing path. The specific process includes: calculating the context switching frequency of the kernel interrupt service routine, determining the impact weight of the soft interrupt delay generated by the system kernel on the processing latency of the communication protocol stack when the switching frequency exceeds a preset interrupt saturation threshold, and simultaneously performing real-time calculation of the conflict probability of the system bus. It analyzes how the competition of physical signals causes logical stagnation of communication management instructions on the internal forwarding path of the system when large amounts of data are transmitted at high speed between the memory direct access controller and the peripheral interface. The kernel processing bottleneck analysis module finally outputs a communication deterministic distortion assessment report, which clarifies the core resource constraints that cause the current communication jitter. Preferably, the communication protocol stack dynamic adjustment module is connected to the kernel processing bottleneck analysis module. This module is configured to intervene in the enqueueing and dequeueing strategies of communication packets in real time based on the communication deterministic distortion assessment report. In a specific implementation, the communication protocol stack dynamic adjustment module reconstructs the buffer allocation logic of the protocol stack. When it detects a shortage of system kernel resources, it automatically allocates protected dedicated soft interrupt processing sequences for high-timeliness management instructions. Furthermore, this module can adjust the depth of the hardware interface cache and adopt an active queue management strategy within the predicted window period of intra-terminal blocking to smoothly discard or delay the enqueueing of non-critical business data packets, thereby ensuring the deterministic pass-through of high-priority communication flows in the terminal's internal processing path. The adjustment mechanism does not rely on the priority labeling of the external network layer, but is based on adaptive feedback of the real-time operating conditions inside the terminal. Preferably, the computing and communication fusion scheduling module is connected to the multi-dimensional resource awareness module, the kernel processing bottleneck analysis module, and the communication protocol stack dynamic adjustment module, respectively. This module is configured to perform cross-layer resource coordination and scheduling, achieving deep integration between computing load and communication load. When the kernel processing bottleneck analysis module determines that intra-end blocking has occurred and the communication determinism index is lower than a preset safety threshold, the computing and communication fusion scheduling module intervenes to perform dynamic task reconstruction. The specific process includes: forcibly adjusting the execution priority of computing tasks through the kernel scheduler interface, suspending or reducing the scheduling frequency of non-real-time high-computation-intensity applications, thereby releasing excessively occupied CPU resources and system bus bandwidth. At the same time, this module, through linkage with the communication protocol stack dynamic adjustment module, synchronously enhances the kernel processing authority of the communication protocol stack, ensuring that during peak computing application load periods, communication management instructions can still be parsed and forwarded within a preset millisecond response period. Preferably, the system further includes a closed-loop feedback compensation module, which is configured to monitor the system response characteristics after scheduling decisions. By comparing the difference in communication message transmission jitter before and after scheduling, the module dynamically updates the bottleneck discrimination threshold in the kernel processing bottleneck analysis module. This closed-loop mechanism ensures that the management strategy can evolve automatically with changes in the operating environment of the intelligent converged terminal, achieving long-term deterministic assurance in heterogeneous communication networking environments. This invention also provides a communication data management method based on an intelligent converged terminal. The method applies the above-mentioned system and includes the following steps: Step 1: Initiate the multi-dimensional resource real-time perception process. The intelligent fusion terminal periodically collects low-level characteristic parameters of computing and communication resources through the monitoring interface at the kernel driver level. These characteristic parameters include at least the instantaneous load of each core of the central processing unit, the arbitration latency of system bus transactions, the conflict frequency of memory access, and the bit error rate of the communication physical layer interface. At the same time, the system obtains the job attributes of the currently running edge computing algorithm through the application programming interface, forming a global state map spanning the hardware resource layer, the kernel operating system layer, and the business logic layer. Step Two: Conduct in-depth analysis of computational communication conflicts. The system inputs the collected multi-dimensional resource data into a preset correlation analysis logic. By establishing a kernel-mode context switching time consumption model, it quantitatively calculates the degree of pressure exerted by the interrupt storm caused by computational tasks on the communication message processing flow. The core of this step is to identify soft blocking phenomena, that is, although the physical communication link bandwidth is sufficient, the abnormal state of communication data in the kernel-mode protocol stack cannot obtain processing time slices in a timely manner because the terminal kernel processing logic is locked by heavy floating-point operation tasks. Step 3: Implement deterministic dynamic intervention for communication. When the jitter trend of communication response latency is detected to exceed the preset safety benchmark, the system immediately initiates an adaptive adjustment strategy for the communication protocol stack. Specific operations include, but are not limited to: dynamically remapping the priority weights of communication tasks in the kernel scheduler, promoting them from ordinary user-space tasks to near real-time task sequences; simultaneously, for control commands to be sent, establishing an emergency processing green channel that bypasses the ordinary buffer queue, and utilizing the priority arbitration mechanism of the direct memory access controller to ensure that critical packets directly interact with the physical communication network card. Step 4: Perform cross-domain resource fusion scheduling. Based on the kernel processing bottleneck analysis results, the system performs time-domain peak-shifting scheduling of the computing task flow and communication data flow within the terminal. During critical communication windows, the system proactively limits the memory bandwidth usage of high-volume computing tasks or migrates the kernel processes of computing tasks to low-load cores, thereby freeing up necessary system resource space for the operation of the communication protocol stack. During idle communication periods, the system restores the full-performance operation mode of computing tasks to maximize terminal resource utilization. Step 5: Complete closed-loop performance evaluation and strategy optimization. After each adjustment cycle, the system automatically measures the end-to-end deterministic indicators of communication services and calculates the attenuation of the impact of adjustment operations on the execution efficiency of edge computing tasks. Through multi-criteria decision-making logic, the trigger thresholds of sensing parameters are finely adjusted to ensure that subsequent scheduling actions are more in line with actual operating condition fluctuations, ultimately achieving a dynamic balance between efficient execution of computing tasks and deterministic transmission of communication data. Preferably, in step two, the method for determining kernel processing bottlenecks is characterized by establishing a non-linear mapping function between bus utilization and message queuing delay. This allows it to identify when the bus request frequency exceeds a preset percentage of the physical limit, triggering the heavy load protection logic of the kernel-mode protocol stack. This logic can predict impending communication performance degradation before actual timeout errors occur. Preferably, in step three, the strategy for adjusting the protocol stack is characterized by introducing a traffic shaping and priority redistribution mechanism based on the token bucket algorithm. This mechanism, tailored to the multi-protocol access scenario unique to intelligent converged terminals, classifies and labels data packets of different communication protocols to ensure that power dispatching instructions, distributed feeder automation signaling, and high-frequency sampling data have different resource acquisition rights in a resource contention environment. Preferably, in step four, the process of performing resource fusion scheduling is characterized by the system implementing a micro-level suspension technique for computing tasks through deep interaction with the operating system kernel. This means that, without interrupting the logical integrity of the computing task, the execution cycle of the computing task is divided into microsecond-level segments using the gaps in communication message processing, thereby resolving resource conflicts between computing and communication at the micro-level. The beneficial effects of this invention are as follows: 1. The communication data management system and method based on intelligent converged terminals described in this invention, by constructing a multi-dimensional resource awareness framework, breaks down the logical barrier between traditional communication management and computing resource allocation. Compared with existing technologies that only focus on network-side congestion, this invention can identify and locate intra-terminal blocking caused by CPU occupancy, bus conflicts, and interrupt handling overhead in real time, thereby providing a more accurate decision-making basis for communication deterministic assurance. 2. The communication data management system and method based on intelligent converged terminals described in this invention effectively solves the communication latency jitter problem of intelligent converged terminals under high load conditions by linking kernel processing bottleneck analysis with dynamic adjustment of the protocol stack. By allocating dedicated kernel resource paths and protected buffers for critical communication tasks, this invention ensures that the terminal's instruction response latency remains within a preset deterministic range when performing computationally intensive tasks such as large-scale floating-point operations, significantly improving the stability of the distributed control system. 3. The communication data management system and method based on intelligent converged terminals described in this invention achieves nonlinear coupling optimization of terminal resources by adopting a strategy of fused computing and communication scheduling. This invention no longer simply addresses performance challenges through hardware stacking, but rather maximizes the parallel processing capabilities of edge computing tasks while ensuring real-time communication through cross-layer resource coordination and task reconfiguration. This deeply integrated architecture not only reduces the system's dependence on hardware specifications but also increases the service carrying capacity of intelligent converged terminals under complex operating conditions. 4. The communication data management system and method based on intelligent fusion terminals described in this invention, through the introduction of a closed-loop feedback compensation mechanism, possesses extremely strong environmental adaptability. The system can automatically correct the management logic according to the actual scheduling effect, effectively cope with the changing task characteristics and communication environment at the end of the distribution network and in industrial sites, and ensure the long-term effectiveness and high reliability of the communication data management strategy, laying a solid technical foundation for supporting the deep integration of the energy internet and the industrial internet of things. Attached Figure Description

[0007] The invention will now be further described with reference to the accompanying drawings.

[0008] Figure 1 This is a structural block diagram of a communication data management system based on an intelligent converged terminal in this invention; Figure 2 This is a flowchart of a communication data management method based on an intelligent converged terminal in this invention. Detailed Implementation

[0009] To make the technical means, creative features, objectives and effects of this invention easier to understand, the invention will be further described below in conjunction with specific embodiments.

[0010] like Figure 1 As shown in the embodiment of the present invention, a communication data management system based on an intelligent converged terminal is based on the construction of a management architecture that can deeply perceive the allocation of micro-resources within the terminal and perform cross-layer scheduling in real time. In the hardware architecture of the intelligent converged terminal, the central processing unit, memory controller, system bus, and various communication physical interfaces together constitute the infrastructure supporting edge computing and data communication. However, when performing high-intensity edge computing tasks (such as deep learning inference, large-scale power topology analysis, or high-frequency data sampling), existing technical solutions often lead to serious resource contention within the terminal due to the logical decoupling of computing resources and communication resources. The embodiment of the present invention achieves comprehensive monitoring of the underlying state of the system through a multi-dimensional resource perception module.

[0011] The multi-dimensional resource awareness module is configured to perform real-time status monitoring across the system kernel layer, computing application layer, and communication link layer. In specific engineering implementations, the multi-dimensional resource awareness module utilizes a resource hook program deployed in the operating system kernel layer. This program is mounted in the scheduler engine as a kernel module (LKM) and can collect the CPU time slice utilization rate in real time with microsecond-level precision. Furthermore, the module obtains the system bus bandwidth utilization by reading the status register of the system bus controller and monitors the concurrent request volume of the direct memory access controller (DMA) in real time. At the computing application layer, the multi-dimensional resource awareness module synchronously obtains the execution status of edge computing tasks through the application layer feedback mechanism, specifically including the trigger frequency and execution status of high-frequency sampling tasks. The computational load during deep learning inference tasks and the frequency of memory page replacements due to large-scale data exchange are addressed at the communication link layer. This module monitors the link utilization of the physical interface, the physical layer signal strength, and the round-trip latency of the remote node in real time by calling the statistical interface of the network card driver. The multi-dimensional resource perception module aligns the physical resource parameters from different levels with the logical task status in time and space, and uses a unified clock reference to stamp all sampled data with high precision, thereby constructing a multi-dimensional resource status vector that can describe the current all-round operating status of the terminal. This multi-dimensional resource status vector not only includes the absolute value of hardware load, but also covers the relative temporal relationship between task flows, providing a solid foundation of data support for subsequent bottleneck analysis.

[0012] The kernel processing bottleneck analysis module is electrically connected to the multi-dimensional resource perception module or logically connected through inter-process communication. This module is configured to deeply analyze the resource contention mechanism within the intelligent fusion terminal. After receiving the multi-dimensional resource state vector, the kernel processing bottleneck analysis module immediately applies a preset bottleneck discrimination logic. This logic is based on a nonlinear interference model and aims to quantify the negative impact of computational tasks on the communication processing path. In specific operation, this module first calculates the context switching frequency of the kernel interrupt service routine (ISR). Since computationally intensive tasks are often accompanied by frequent memory accesses and complex logical judgments, this causes the kernel scheduler to frequently trigger interrupts to coordinate resource allocation. When the switching frequency exceeds a preset interrupt saturation threshold, the soft interrupt latency generated by the system kernel will increase sharply, and the kernel processing bottleneck... The bottleneck analysis module calculates the impact weight of soft interrupt latency on communication protocol stack processing latency, thereby determining what proportion of communication latency is caused by kernel scheduling overhead. At the same time, the module performs real-time calculation of the system bus conflict probability. When large amounts of data (such as high-definition image data or high-frequency sampling sequences) are transmitted at high speed between the memory direct access controller and the peripheral interface, competition for physical signals on the bus can cause logical stalls in the forwarding path of communication management instructions. The kernel processing bottleneck analysis module analyzes the waiting period of the bus arbiter to quantitatively assess the degree of damage to communication determinism caused by this physical-level competition, and finally outputs a communication deterministic distortion assessment report. This report clarifies the core resource constraints that cause the current communication jitter, such as whether it is due to the CPU core being fully occupied or the system bus bandwidth reaching its physical limit.

[0013] The communication protocol stack dynamic adjustment module is connected to the kernel processing bottleneck analysis module. Its core function is to intervene in the enqueue and dequeue strategies of communication packets in real time based on the communication deterministic distortion assessment report. In a specific implementation, the communication protocol stack dynamic adjustment module achieves this goal by reconstructing the buffer allocation logic of the protocol stack. When the system kernel resources are detected to be scarce due to high load computing, the module will automatically allocate protected dedicated soft interrupt processing sequences (Tasklet or Workqueue) for management instructions with high timeliness requirements, ensuring that these critical packets can bypass the regular task queue and obtain faster processing response. Furthermore, the module can dynamically adjust the depth of the hardware interface cache. During the window period when it is predicted that the internal blocking will occur, the module adopts an active queue management strategy to smoothly discard or delay the enqueue processing of non-critical business data packets (such as routine status reporting data), freeing up valuable buffer resources. This adjustment mechanism has a high degree of adaptability. It does not rely on the priority labeling of the external network layer (such as DSCP or VLAN priority), but is entirely based on the feedback of the real-time working conditions inside the terminal, ensuring the deterministic pass-through of high-priority communication flows in the internal processing path of the terminal.

[0014] The computational-communication fusion scheduling module is connected to the multi-dimensional resource awareness module, the kernel processing bottleneck analysis module, and the communication protocol stack dynamic adjustment module, respectively. As the core command unit of the system, this module is configured to perform cross-layer resource coordination and scheduling, aiming to achieve deep integration between computational and communication loads. When the kernel processing bottleneck analysis module determines that intra-end blocking has occurred and the communication determinism index is lower than a preset safety threshold, the computational-communication fusion scheduling module immediately intervenes and performs dynamic task reconfiguration. Specifically, this module forcibly adjusts the execution priority of computational tasks through the kernel scheduler interface. For applications with low real-time requirements but extremely high computational intensity (such as historical trend analysis algorithms), the module will perform time-slice suspension operations or reduce its scheduling frequency by adjusting its weight value in the Completely Fair Scheduler (CFS). This operation can quickly release excessively occupied CPU resources and system bus bandwidth. At the same time, through linkage with the communication protocol stack dynamic adjustment module, the module synchronously enhances the kernel processing authority of the communication protocol stack, ensuring that communication management instructions can still be parsed and forwarded within a preset millisecond response period when the computing application load is at its peak, thereby eliminating the preemptive interference of computing tasks on communication tasks.

[0015] In a preferred embodiment of the present invention, the system further includes a closed-loop feedback compensation module. This module is configured to monitor the actual response characteristics of the system after the scheduling decision is executed. By comparing the difference in communication message transmission jitter before and after scheduling, this module can evaluate the effectiveness of the current scheduling strategy. If the jitter improvement is not significant, the closed-loop feedback compensation module will dynamically update the bottleneck discrimination threshold in the kernel processing bottleneck analysis module, so that the system can trigger scheduling earlier or more accurately when facing similar working conditions in the future. This closed-loop mechanism ensures that the management strategy can evolve with the migration of the intelligent fusion terminal operating environment, the update of software algorithms, and the performance fluctuations caused by hardware aging. It achieves long-term deterministic guarantee in heterogeneous communication networking environments (such as a hybrid environment combining low-power wireless, carrier communication and Ethernet).

[0016] like Figure 2 As shown, this embodiment of the invention also provides a communication data management method based on an intelligent converged terminal, which applies the above-mentioned system and includes the following specific steps: In step one, the system initiates a multi-dimensional resource real-time perception process. The intelligent fusion terminal utilizes the monitoring interface at the kernel driver level to periodically collect low-level characteristic parameters of computing and communication resources at a preset sampling period (e.g., once every 100 microseconds). These characteristic parameters not only include the instantaneous load of each core of the central processing unit, but also delve into the hardware abstraction layer to obtain key indicators such as the arbitration latency of system bus transactions, the conflict frequency of memory access, and the bit error rate of the communication physical layer interface. At the same time, the system obtains the job attributes of the currently running edge computing algorithm through the application programming interface (API), including the algorithm's expected execution time, peak memory usage, and real-time performance level. In this way, the system forms a global state map spanning the hardware resource layer, the kernel operating system layer, and the business logic layer, laying the information foundation for accurate decision-making.

[0017] In step two, the system performs an in-depth analysis of computational communication conflicts. In this step, the system inputs the collected multi-dimensional resource data into a preset correlation analysis logic. This logic, by establishing an accurate kernel-mode context switching time consumption model, can quantitatively calculate the specific degree of pressure exerted by the interrupt storm caused by the computational task on the communication message processing flow. The key to this step is to identify soft blocking phenomena. Soft blocking refers to a situation where, although the bandwidth of the physical communication link is very sufficient and the signal quality is good, the processing logic of the terminal kernel is locked for a long time by heavy floating-point operation tasks, causing communication data packets to be unable to obtain the required processing time slice in time after entering the kernel-mode protocol stack, thus generating huge latency jitter in the internal queuing. This invention can accurately identify this hidden performance bottleneck by establishing a nonlinear mapping function between bus utilization and message queuing delay.

[0018] In step three, the system implements deterministic dynamic intervention in communication. When the multi-dimensional resource perception module detects that the jitter trend of communication response latency exceeds the preset safety benchmark, the system immediately activates the adaptive adjustment strategy of the communication protocol stack. Specifically, the system modifies the Task Control Block (TCB) of the kernel scheduler to dynamically remap the priority weight of communication tasks in the kernel scheduler, promoting them from ordinary user-space tasks to a near-real-time task sequence with higher real-time attributes. At the same time, for critical control instructions to be sent, the system establishes an emergency processing green channel that bypasses the ordinary buffer queue. This channel utilizes the priority arbitration mechanism of the direct memory access controller to enable critical messages to directly interact with the physical communication network card, greatly shortening the data retention time inside the protocol stack.

[0019] In step four, the system performs cross-domain resource fusion scheduling. Based on the kernel processing bottleneck analysis results, the system performs fine-grained time-domain peak-shifting scheduling of the computing task flow and communication data flow within the terminal. During critical communication windows, the system actively limits the memory bandwidth usage of high-volume computing tasks, or utilizes the multi-core affinity feature of the kernel to forcibly migrate the computing task process to the auxiliary processor core with a relatively low load, thereby freeing up the necessary resource space for the main processor core running the communication protocol stack. During the gaps when communication services are idle, the system automatically restores the full-performance running mode of the computing tasks. Through this dynamic push-pull scheduling, the terminal maximizes resource utilization while ensuring real-time communication. In particular, this invention introduces a micro-suspending technology for computing tasks. By dividing the execution cycle of computing tasks at the microsecond level, computing instructions are inserted during the tiny gaps in communication message processing, thereby resolving resource conflicts without destroying the integrity of the computing logic.

[0020] In step five, the system completes closed-loop performance evaluation and strategy optimization. After each adjustment cycle, the system automatically measures the end-to-end deterministic indicators of communication services and calculates the attenuation of the impact of the above adjustment operations on the execution efficiency of edge computing tasks. The system's built-in multi-criteria decision logic will finely adjust the trigger threshold of the perception parameters based on the evaluation results. For example, if it is found that a certain type of image recognition algorithm occupies the bus in a highly regular manner, the system will adjust the size of the prediction window accordingly. This continuous optimization ensures that the scheduling actions can accurately match the fluctuations of the actual working conditions, and finally achieve a dynamic balance between the efficient execution of computing tasks and the deterministic transmission of communication data.

[0021] Furthermore, in step two of the above method, the method for determining the kernel processing bottleneck is characterized by establishing a non-linear mapping function between bus utilization and message queuing delay. When the bus request frequency exceeds a preset percentage (such as 85%) of the physical limit, the heavy load protection logic of the kernel-mode protocol stack is immediately triggered. This forward-looking logic can predict the upcoming communication performance degradation before the actual timeout error or physical packet loss occurs, thereby leaving sufficient response buffer time for the system.

[0022] In step three, the strategy for adjusting the protocol stack specifically introduces a traffic shaping and priority redistribution mechanism based on the token bucket algorithm. This mechanism is specifically optimized for the multi-protocol access scenarios unique to intelligent converged terminals (such as simultaneous access to Modbus, IEC 61850 and MQTT protocols). The system classifies and labels data packets of different communication protocols to ensure that power dispatch instructions and distributed feeder automation signaling, which are crucial to the power system, have the highest token acquisition rights. For ordinary high-frequency sampling data, reasonable queuing or rate limiting is performed based on the current token balance in a resource contention environment.

[0023] In step four, the process of performing resource fusion scheduling is achieved through deep interaction with the operating system kernel. The micro-suspending technology uses the operating system's clock interrupt mechanism to divide the originally continuously executed computing tasks into several tiny execution units. Between each execution unit, the system checks whether there are any high-priority communication tasks to be processed. If so, the communication messages are processed first during the microsecond-level gap. In this way, it appears that the computing tasks are still running on a macroscopic level, but in reality, a highway has been cleared for communication tasks at the micro level, fundamentally solving the communication congestion problem under high load.

[0024] To further demonstrate the technical advantages of this invention, this embodiment constructs a specific application scenario for experimental verification. In this embodiment, the intelligent fusion terminal simultaneously runs a device status recognition algorithm based on a convolutional neural network (CNN) (high computational load) and a distribution network real-time control protocol based on IEC 61850. Example: In this example, the intelligent fusion terminal adopts the communication data management system described in this invention. The multi-dimensional resource perception module detects that when the CNN inference task starts, the CPU utilization rate rapidly rises to 92% and the system bus bandwidth utilization rate reaches 88%. The kernel processing bottleneck analysis module immediately identifies the risk that the queuing delay of control protocol messages will increase from the usual 2 milliseconds to 15 milliseconds due to frequent switching of interrupt service routines. Subsequently, the computing and communication fusion scheduling module starts, lowers the priority of the CNN task in the kernel scheduler, and applies micro-suspending technology, reserving a communication blank period of 500 microseconds for every 10 milliseconds of computing task execution. At the same time, the communication protocol stack dynamic adjustment module opens a DMA green channel for IEC 61850 messages.

[0025] Comparative Example: In the comparative example, the intelligent fusion terminal adopts the traditional independent management mode of communication and computing. The computing tasks run according to the default fair scheduling policy of the operating system, and the communication protocol stack relies only on the QoS label of the network layer for priority management. When the same CNN inference task is started, the kernel-mode protocol stack experiences severe soft blocking because the computing task preempts a large amount of system bus bandwidth and CPU time slice.

[0026] The table below shows a comparison of the communication determinism metrics of the embodiments of the present invention and the comparative examples under the above-mentioned high-load conditions: Note: Although the embodiments of the present invention slightly increase the computation task completion time, this is a necessary trade-off to ensure communication determinism. Experimental results show that the present invention improves communication determinism by an order of magnitude while sacrificing only a small amount of computational efficiency, and completely solves the risk of control failure caused by intra-terminal blocking.

[0027] Furthermore, the communication data management system based on intelligent converged terminals described in this invention exhibits excellent robustness when handling multi-protocol convergence scenarios. Since intelligent converged terminals typically need to process data from different physical interfaces (such as RS-485, CAN bus, and Ethernet) simultaneously, and the drivers for these interfaces have different interrupt priorities in kernel mode, the communication protocol stack dynamic adjustment module of this invention can perform unified resource view mapping on these data from heterogeneous interfaces. For example, when data accumulation is detected in the RS-485 serial interface due to a low baud rate, the module will automatically allocate a larger circular buffer for it in memory and synchronously adjust its weight in the kernel task queue to prevent interrupt cascading delays in the entire system kernel due to the blocking of low-speed interfaces. This adaptive capability for heterogeneous communication environments is unparalleled by traditional solutions based on a single network layer management.

[0028] In the specific implementation process, the computing and communication fusion scheduling module also adopts a scheduling algorithm based on prediction weights. This algorithm learns from historical data to establish a feature library between different computing tasks and resource consumption. When the multi-dimensional resource perception module identifies a specific edge computing task (such as transient waveform analysis) being triggered, the scheduling module can sense in advance the instantaneous impact that the task will have on the bus, thereby making advance adjustments to the priority of the communication protocol stack. This prediction-based scheduling method further reduces the system's response latency at the moment of load change, realizing the transformation from passive response to active prevention.

[0029] Furthermore, the system architecture described in this invention has extremely high flexibility in hardware implementation. Although the above description is mostly based on a general processor architecture, the logic of this invention is also applicable to heterogeneous system-on-chips (SoCs) that integrate field-programmable gate arrays (FPGAs) or neural network acceleration units (NPUs). In an intelligent fusion terminal containing an FPGA accelerator, the multi-dimensional resource perception module can obtain the data exchange frequency between the accelerator and the main processing core by monitoring the bus transactions of the high-performance extended interface (AXI) inside the SoC. The computing and communication fusion scheduling module can achieve collaborative optimization between hardware acceleration tasks and communication tasks by dynamically adjusting the computing frequency or DMA transfer priority of the FPGA.

[0030] In summary, the communication data management system and method based on intelligent converged terminals provided in this invention break down the logical barriers between traditional communication management and computing resource allocation by constructing a multi-dimensional resource awareness framework. Compared to existing technologies that only focus on network-side congestion, this invention can identify and locate intra-terminal blocking caused by CPU occupancy, bus conflicts, and interrupt handling overhead in real time, thus providing a more accurate decision-making basis for communication determinism. Through the linkage of kernel processing bottleneck analysis and dynamic adjustment of the protocol stack, this invention effectively solves the communication latency jitter problem of intelligent converged terminals under high load conditions. By allocating dedicated kernel resource paths and protected buffers for critical communication tasks, this invention ensures stable communication even when performing computationally intensive tasks such as large-scale floating-point operations. During operation, the terminal's command response latency remains within a preset deterministic range. By adopting a computing and communication fusion scheduling strategy, the nonlinear coupling optimization of terminal resources is achieved. This invention no longer simply addresses performance challenges through hardware stacking, but rather through cross-layer resource coordination and task reconfiguration. While ensuring real-time communication, it maximizes the parallel processing capability of edge computing tasks. Finally, through the introduction of a closed-loop feedback compensation mechanism, this invention possesses strong environmental adaptability and can automatically correct management logic based on actual scheduling effects. It effectively copes with the changing task characteristics and communication environments at the end of the distribution network and in industrial sites, ensuring the long-term effectiveness and high reliability of the communication data management strategy. This lays a solid technical foundation for supporting the deep integration of the energy internet and the industrial internet of things.

[0031] In actual production deployments, this invention can be widely applied to power distribution automation terminals, power load management terminals, and various industrial IoT edge gateways. In these scenarios, terminals often need to maintain high-speed and reliable communication with the master station system or adjacent terminals while performing complex local logical operations (such as load forecasting and fault location). The system and method provided by this invention can ensure that the response to communication commands is always in a controllable and deterministic state during the parallel execution of these critical tasks, which greatly improves the overall security and operating efficiency of the distributed system.

[0032] The foregoing has shown and described the basic principles, main features, and advantages of the present invention. Those skilled in the art should understand that the present invention is not limited to the above embodiments. The embodiments and descriptions in the specification are merely illustrative of the principles of the invention. Various changes and modifications can be made to the invention without departing from its spirit and scope, and all such changes and modifications fall within the scope of the present invention as claimed. The scope of protection of the present invention is defined by the appended claims and their equivalents.

Claims

1. A communication data management system based on an intelligent converged terminal, characterized in that, The system includes: The multi-dimensional resource awareness module is used to perform real-time status monitoring across the system kernel layer, computing application layer and communication link layer. It collects underlying physical resource parameters by deploying resource hook programs in the system kernel layer, and obtains the logical task status of the computing application layer and the link characteristics of the communication link layer. The collected physical resource parameters and logical task status are spatiotemporally aligned to construct a multi-dimensional resource status vector. The kernel processing bottleneck analysis module, connected to the multi-dimensional resource perception module, is used to receive the multi-dimensional resource state vector, establish a kernel state context switching time consumption model, calculate the delay increment when the context switching frequency of the kernel interrupt service routine exceeds a preset threshold, quantify the interference of the calculation task on the communication processing path, analyze the impact weight of the soft interrupt delay generated by the system kernel on the processing delay of the communication protocol stack, and output a communication deterministic distortion assessment report to clarify the core resource constraints of communication jitter. The communication protocol stack dynamic adjustment module is connected to the kernel processing bottleneck analysis module. It is used to intervene in the enqueue and dequeue strategies of communication messages according to the communication deterministic distortion assessment report. By reconstructing the buffer allocation logic of the protocol stack, it allocates a dedicated soft interrupt processing sequence for high-timeliness management instructions and executes an active queue management strategy according to the prediction window of intra-terminal blocking. The computation and communication fusion scheduling module is connected to the multi-dimensional resource perception module, the kernel processing bottleneck analysis module, and the communication protocol stack dynamic adjustment module, respectively. It is used to perform cross-layer resource coordination and scheduling. When the communication determinism index is lower than the preset safety threshold, it modifies the vruntime weight value of the computing process in the Completely Fair Scheduler (CFS) through the API interface provided by the kernel scheduler, or limits the CPU quota of the computing process through the cgroups subsystem. It suspends the execution time slice of high-computation-intensive applications or reduces their scheduling frequency, and simultaneously improves the kernel processing privileges of the communication protocol stack, thereby realizing the fusion scheduling of computing load and communication load.

2. The communication data management system based on an intelligent converged terminal according to claim 1, characterized in that, The multi-dimensional resource awareness module collects low-level physical resource parameters in real time at the system kernel layer through the resource hook program, including at least the following: CPU time slice utilization, system bus bandwidth utilization, and concurrent requests to the memory direct access controller; The execution status of the edge computing tasks obtained by the multi-dimensional resource perception module at the computing application layer includes at least: the trigger frequency of high-frequency sampling tasks, the computational load of deep learning inference tasks, and the memory page replacement frequency of image recognition tasks. The characteristic parameters monitored by the multi-dimensional resource awareness module at the communication link layer include at least: the link utilization of the physical interface, the signal strength, and the round-trip latency of the remote node. The multi-dimensional resource perception module uses a unified clock reference to stamp all sampled data with high precision, so as to achieve spatiotemporal alignment between the physical resource parameters and the logical task status.

3. The communication data management system based on an intelligent converged terminal according to claim 1, characterized in that, The kernel processing bottleneck analysis module is configured as follows: Calculate the context switching frequency of the kernel interrupt service routine. When the context switching frequency exceeds a preset interrupt saturation threshold, determine the impact weight of the soft interrupt delay generated by the system kernel on the processing latency of the communication protocol stack. The probability of collisions on the system bus is calculated, and the logical stalls caused by physical signal competition during the transmission of large amounts of data between the memory direct access controller and the peripheral interface are analyzed. By analyzing the waiting period of the bus arbiter, the extent to which physical layer competition disrupts communication determinism is quantitatively assessed, and a communication deterministic distortion assessment report is generated.

4. A communication data management system based on an intelligent converged terminal according to claim 1, characterized in that, The communication protocol stack dynamic adjustment module is configured as follows: When a shortage of system kernel resources is detected, time-sensitive management instructions are redirected to a protected dedicated soft interrupt handling sequence, which is selected from a task assistant (Tasklet) or a work queue (Workqueue). The depth of the hardware interface cache is dynamically adjusted. During the predicted window period of intra-terminal blocking, non-critical business data packets are smoothly dropped or delayed in queuing according to the active queue management strategy to free up buffer resources. The active queue management strategy is based on adaptive feedback of real-time operating conditions within the terminal and does not rely on priority labeling of the external network layer.

5. A communication data management system based on an intelligent converged terminal according to claim 1, characterized in that, The computing and communication fusion scheduling module is configured as follows: By intervening in the kernel scheduling process through the kernel scheduler interface, for applications with a real-time requirement lower than the preset value and computational intensity exceeding the preset load, their scheduling frequency is reduced by modifying their weight value in the Completely Fair Scheduler, thereby releasing the occupied CPU resources and system bus bandwidth. In conjunction with the communication protocol stack dynamic adjustment module, during peak computing application load periods, the processing authority of the communication protocol stack is elevated to a near real-time task sequence, ensuring that communication management commands are parsed and forwarded within a preset millisecond-level response period.

6. A communication data management system based on an intelligent converged terminal according to claim 1, characterized in that, The system also includes a closed-loop feedback compensation module, which is configured as follows: Monitor the system response characteristics after the execution of scheduling decisions, and evaluate the effectiveness of the current scheduling strategy by comparing the difference in communication message transmission jitter before and after the execution of scheduling actions; Based on the jitter difference in the communication message transmission, the bottleneck discrimination threshold in the kernel processing bottleneck analysis module is dynamically updated, enabling the system to evolve its strategies for environmental migration and hardware aging in heterogeneous communication network environments.

7. A communication data management method based on an intelligent converged terminal, applicable to the communication data management system based on an intelligent converged terminal as described in any one of claims 1-6, characterized in that, The method includes the following steps: Step 1: Initiate the multi-dimensional resource real-time perception process. The intelligent fusion terminal periodically collects the underlying characteristic parameters of computing and communication resources through the monitoring interface at the kernel driver level. The underlying characteristic parameters include the instantaneous load of the central processing unit core, the arbitration latency of system bus transactions, the conflict frequency of memory access, and the bit error rate of the communication physical layer interface. It also obtains the job attributes of the currently running edge computing algorithm through the application programming interface, forming a global state map that spans the hardware resource layer, the kernel operating system layer, and the business logic layer. Step 2: Conduct in-depth analysis of computational communication conflicts. Input the collected multi-dimensional resource data into the correlation analysis logic. By establishing a kernel context switching time consumption model, quantitatively calculate the degree of pressure on the communication message processing flow caused by the interrupt storm triggered by the computation task, and identify the soft blocking phenomenon caused by the kernel processing logic being locked by the floating-point operation task. Step 3: Implement deterministic dynamic intervention for communication. When the jitter trend of communication response latency exceeds the preset safety benchmark, dynamically remap the priority weight of communication tasks in the kernel scheduler, promote them from ordinary user-mode tasks to near real-time task sequences, and establish a green channel for control instructions to be sent. Utilize the priority arbitration mechanism of the direct memory access controller to enable critical messages to bypass the ordinary buffer queue and directly interact with the physical communication network card. Step 4: Perform cross-domain resource fusion scheduling. Based on the kernel processing bottleneck analysis results, perform time-domain peak-shifting scheduling on the computing task flow and communication data flow inside the terminal. During the critical communication window period, limit the memory bandwidth occupation of high-volume computing tasks or perform kernel process migration to free up resource space for the communication protocol stack. During the idle period of communication services, restore the full performance operation mode of computing tasks. Step 5: Complete closed-loop performance evaluation and strategy optimization. After the scheduling cycle ends, measure the end-to-end deterministic indicators of communication services and calculate the attenuation of the impact of adjustment operations on the execution efficiency of edge computing tasks. Fine-tune the trigger threshold of perception parameters through multi-criteria decision logic.

8. A communication data management method based on an intelligent converged terminal according to claim 7, characterized in that, In step two, the specific method for determining the kernel processing bottleneck is as follows: Establish a nonlinear mapping function between system bus utilization and message queuing delay; The system identifies the bus request frequency in the system bus utilization. When the bus request frequency exceeds a preset percentage of the physical limit, the heavy load protection logic of the kernel-mode protocol stack is triggered before an actual timeout error occurs.

9. A communication data management method based on an intelligent converged terminal according to claim 7, characterized in that, In step three, the specific strategy for implementing deterministic dynamic intervention in communication is as follows: A traffic shaping and priority redistribution mechanism based on the token bucket algorithm is introduced; For the various communication protocols accessed by intelligent converged terminals, data packets of different protocols are classified and labeled to ensure that power dispatching instructions and distributed feeder automation signaling have the right to obtain tokens first in a resource competition environment.

10. A communication data management method based on an intelligent converged terminal according to claim 7, characterized in that, In step four, the process of performing cross-domain resource fusion scheduling is achieved through deep interaction with the operating system kernel, specifically including: The micro-suspending technique is adopted, which utilizes the kernel preemption mechanism to insert communication task detection logic into the clock interrupt handling function. If a high-priority communication message exists, the time slice of the current computing process is suspended. During the intervals between execution units, a search is conducted to determine if there are any high-priority communication tasks to be processed. If so, the communication messages are processed preferentially during the intervals, thereby resolving resource conflicts between computational and communication tasks at the micro level.

Images