Low-power-consumption intelligent collaborative management system and method of convergence gateway
By building a communication traffic prediction model and reinforcement learning algorithm, combined with dynamic task scheduling and power supply management, the high power consumption problem of the integrated gateway is solved, low-power intelligent collaborative management is achieved, and the energy efficiency and stability of the equipment are improved.
Patent Information
- Application Number
- CN202510858886.1
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-06-25
- Publication Date
- 2025-09-26
AI Technical Summary
The converged gateway has problems such as high power consumption of multi-protocol communication modules, unreasonable allocation of data processing tasks, and lack of intelligence in energy management, resulting in excessive power consumption, affecting device life and energy efficiency.
A combination of communication management module, task scheduling module and energy control module is adopted. By constructing a communication traffic prediction model, reinforcement learning algorithm and power consumption prediction model, the communication module working mode, task allocation and power supply parameters can be dynamically adjusted to achieve intelligent collaborative management.
Significantly reduce the power consumption of the converged gateway, improve energy utilization efficiency, extend equipment life, reduce energy costs, and ensure the real-time and high quality of network communication and audio and video decoding.
Smart Images
Figure CN120710885A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the technical field of converged gateways, and in particular to a low-power intelligent collaborative management system and method for converged gateways. Background Art
[0002] A converged gateway is a comprehensive network device or system that integrates multiple network protocol conversions, business function integration, data processing, and security control. Its core value lies in breaking down barriers between different networks, protocols, and services, enabling interconnection and business convergence in heterogeneous network environments. It is commonly used in scenarios requiring cross-domain collaboration, such as the Industrial Internet, the Internet of Things, and 5G communications.
[0003] With the rapid development of communication technology, converged gateways are becoming increasingly important as key devices that connect different networks and aggregate and process a variety of data. However, current converged gateways face serious power consumption issues in practical applications:
[0004] 1. High power consumption of multi-protocol communication modules: Converged gateways typically integrate multiple communication modules, such as video decoding, optical modem, Wi-Fi, and Bluetooth. These different protocol modules often operate independently, lacking effective collaborative management mechanisms. For example, when the Wi-Fi module is transmitting large amounts of data, even if the Bluetooth module is idle, it may still consume additional power because it fails to enter low-power mode in a timely manner.
[0005] 2. Irrational allocation of data processing tasks: Converged gateways need to process data from diverse sensor devices, resulting in a complex set of data processing tasks. Traditional task allocation methods, often based on simple priority or round-robin mechanisms, fail to fully consider the compatibility of task characteristics with hardware resources. For example, data analysis tasks that require minimal real-time performance but are computationally intensive can be assigned to high-performance but energy-intensive CPU cores, resulting in unnecessary power consumption.
[0006] 3. Lack of Intelligent Energy Management: Existing converged gateway energy management is mostly static, unable to intelligently adjust based on dynamic factors such as actual business load and environmental conditions. During periods of low business volume, the gateway maintains high operating power, failing to effectively reduce energy consumption. For example, during late night hours when device activity is low, gateway power consumption does not significantly decrease, resulting in significant energy waste. Summary of the Invention
[0007] To address the problems of high power consumption of multi-protocol communication modules, unreasonable allocation of data processing tasks, and lack of intelligence in energy management in converged gateways, the present invention provides a low-power intelligent collaborative management system and method for converged gateways to reduce the power consumption of converged gateways, improve their energy utilization efficiency in home multimedia scenarios, extend the service life of the equipment, and reduce energy costs.
[0008] In the first aspect, the present invention provides a low-power intelligent collaborative management system integrating a gateway, which solves the above-mentioned technical problems by adopting the following technical solutions:
[0009] A low-power intelligent collaborative management system integrating a gateway, comprising:
[0010] The communication management module is responsible for building and training a communication traffic prediction model to predict the traffic values of each communication module within a set time period in the future. It dynamically adjusts the working mode of the communication module based on the traffic value prediction results and feeds back the communication resource status to the task scheduling module and energy control module.
[0011] The task scheduling module is responsible for analyzing the characteristics of tasks entering the converged gateway. Based on the communication resource status and power consumption constraints, it uses a reinforcement learning algorithm to generate a task allocation strategy among different processing units, thus achieving intelligent scheduling to reduce task processing power consumption and improve execution efficiency.
[0012] The energy control module is responsible for collecting environmental data and business load data in real time, and building a power consumption prediction model that integrates environmental factors, business load and system power consumption; it is responsible for dynamically adjusting the system power supply parameters based on the real-time collected data and the output results of the power consumption prediction model; it is responsible for triggering the task scheduling module to reallocate tasks when it detects that the system power consumption exceeds the threshold, and notifying the communication management module to adjust the working mode of the communication module to form a closed-loop energy control.
[0013] Optionally, the communication management module involved specifically includes:
[0014] A historical data collection unit is used to collect historical communication data of the converged gateway, including traffic data of different device types passing through each communication module in different time periods;
[0015] A model building and training unit is used to build a communication traffic prediction model based on a long short-term memory network. The communication traffic prediction model outputs the traffic prediction value of each communication module within a set time period in the future based on the data collected by the historical data collection unit;
[0016] The dynamic working mode switching unit is used to trigger the working mode switching mechanism when the traffic prediction value of a communication module output by the communication traffic prediction model is lower than the preset threshold within a set time length, and switch the communication module from the normal working mode to the low-power sniffing mode; it is also used to set the time in advance to wake up the dormant communication module and optimize the parameters when the traffic prediction value of a communication module output by the communication traffic prediction model reaches a peak within the future time t, and at the same time notify the task scheduling module to increase the task allocation priority of the link, so as to achieve coordinated optimization of communication and task processing.
[0017] Optionally, the task scheduling module may include:
[0018] The task monitoring unit is responsible for real-time monitoring of the task type when the task enters the fusion gateway;
[0019] The feature analysis unit is used to analyze tasks based on their types, obtain key feature information of the tasks, and quantify and encode them into vector forms suitable for algorithm processing, providing a data basis for subsequent task allocation decisions;
[0020] The reinforcement learning task allocation unit is used to build a task allocation environment based on the feature quantization results. By designing a reward function for task completion time, processing power consumption, and priority weight, the reinforcement learning algorithm is trained to generate the optimal strategy mapping. In combination with the communication resource status feedback from the communication management module and the power consumption constraints of the energy control module, the optimal task allocation action is automatically selected according to the current task and system status.
[0021] Preferably, the task allocation environment constructed by the reinforcement learning task allocation unit based on the task feature quantification results includes a task state space and a task action space, wherein: the task state space includes the current load of each processing unit, the remaining computing resources, and the task feature information in the task queue, and the task action space includes all possible choices for allocating tasks to different processing units.
[0022] Optionally, the energy control modules involved specifically include:
[0023] The environment and load sensing unit is used to collect real-time data on ambient temperature, light intensity, and humidity using sensor equipment, and to obtain real-time service load data using the system detection tools within the fusion gateway, and to share this data with the task scheduling module and the communication management module.
[0024] A model building unit is used to build a power consumption prediction model that integrates environmental factors, business load, and system power consumption based on historical data and physical relationships;
[0025] The energy control algorithm unit is used to dynamically adjust the power supply parameters of the integrated gateway using adaptive voltage and frequency scaling technology based on the real-time data of the environment and load sensing unit and the output results of the power consumption prediction model. It is also used to coordinate with the task scheduling stage to optimize the task allocation strategy when it detects that the system power consumption exceeds the threshold, and to coordinate with the communication management stage to adjust the working mode of the communication module.
[0026] In the second aspect, the present invention provides a low-power intelligent collaborative management method for a converged gateway, which solves the above-mentioned technical problems using the following technical solutions:
[0027] A low-power intelligent collaborative management method for a fusion gateway comprises the following steps:
[0028] S1, Communication Management Phase: By collecting and integrating historical communication data from the gateway, a communication traffic prediction model is constructed and trained to predict the traffic values of each communication module within a set time period in the future. The working mode of the communication module is dynamically adjusted based on the traffic value prediction results, and the communication resource status is fed back to the task scheduling phase and energy control phase.
[0029] S2, Task Scheduling: Analyze the characteristics of tasks entering the converged gateway. Combined with the communication resource status and power consumption constraints, a reinforcement learning algorithm is used to generate a task allocation strategy among different processing units. This enables intelligent scheduling to reduce task processing power consumption and improve execution efficiency.
[0030] S3, Energy Control Phase: Collect environmental data and business load data in real time, build a power consumption prediction model that integrates environmental factors, business load, and system power consumption; dynamically adjust system power supply parameters based on real-time collected data and the output of the power consumption prediction model; trigger the task scheduling phase to reallocate tasks when the system power consumption exceeds the threshold, and notify the communication management phase to adjust the working mode of the communication module to form a closed-loop energy control.
[0031] Optionally, step S1 specifically includes:
[0032] S1.1. Collect historical communication data of the converged gateway, including traffic data of different device types passing through each communication module in different time periods;
[0033] S1.2. Build a communication traffic prediction model based on the long short-term memory network. The communication traffic prediction model outputs the traffic prediction value of each communication module within a set time period in the future based on the collected historical communication data of the converged gateway;
[0034] S1.3. When the traffic prediction value of a communication module output by the communication traffic prediction model is lower than a preset threshold within a set period of time, a working mode switching mechanism is triggered to switch the communication module from the normal working mode to the low-power sniffing mode;
[0035] S1.4. When the traffic prediction value of a certain communication module output by the communication traffic prediction model reaches its peak value within the next t time, the time is set in advance to wake up the dormant communication module and optimize the parameters. At the same time, the task scheduling stage is notified to increase the priority of task allocation to this link, so as to achieve coordinated optimization of communication and task processing.
[0036] Further optionally, step S2 specifically includes the following operations:
[0037] S2.1, monitor the task type in real time when the task enters the fusion gateway;
[0038] S2.2. Analyze tasks based on their types, extract key feature information, and quantify and encode them into vector form suitable for algorithm processing, providing a data basis for subsequent task allocation decisions;
[0039] S2.3. Build a task allocation environment based on the feature quantization results. By designing a reward function based on task completion time, processing power consumption, and priority weight, train the reinforcement learning algorithm to generate the optimal strategy mapping. Combined with the communication resource status feedback from the communication management phase and the power consumption constraints from the energy control phase, the optimal task allocation action is automatically selected based on the current task and system status.
[0040] Preferably, the task allocation environment constructed based on the task feature quantification results includes a task state space and a task action space, wherein:
[0041] The task state space includes the current load of each processing unit, the remaining computing resources, and the task feature information in the task queue;
[0042] The task action space includes all possible choices for assigning tasks to different processing units.
[0043] Optionally, step S3 specifically includes:
[0044] S3.1. Use sensor equipment to collect real-time data on ambient temperature, light intensity, and humidity. Use system detection tools within the converged gateway to obtain real-time service load data. This data is synchronized with the task scheduling and communication management stages.
[0045] S3.2. Build a power consumption prediction model that integrates environmental factors, service load, and system power consumption based on historical data and physical relationships.
[0046] S3.3. Based on the real-time data of step S3.1 and the output results of the power consumption prediction model constructed in step S3.2, the power supply parameters of the integrated gateway are dynamically adjusted using adaptive voltage and frequency scaling technology; when the system power consumption is detected to exceed the threshold, the task allocation strategy is optimized in collaboration with the task scheduling stage, and the communication module working mode is adjusted in conjunction with the communication management stage.
[0047] The low-power intelligent collaborative management system and method of the present invention, which integrates gateways, has the following beneficial effects compared with the prior art:
[0048] 1. This invention reduces the power consumption of converged gateways through dynamic task perception and adaptive scheduling, intelligent collaboration and energy saving of communication modules, and dynamic energy management based on the environment and load. It has the advantages of significantly reducing power consumption, improving performance, and enhancing device stability and lifespan. It can be widely used in converged gateway products for broadcasting and television systems, and can also be extended to other product areas with low-power multi-network convergence scenarios.
[0049] 2. The present invention significantly reduces the power consumption of the converged gateway through intelligent communication module management, innovative task scheduling mechanism and dynamic energy regulation, improves its energy utilization efficiency in home multimedia scenarios, extends the service life of the equipment and reduces energy costs.
[0050] 3. The present invention can significantly reduce the power consumption of the integrated gateway, effectively alleviate the heating problem caused by limited power supply, ensure its stable operation, improve the stability and reliability of the equipment, extend its service life, and reduce the frequency of user equipment replacement; while reducing power consumption, by building a prediction model for intelligent collaboration, it can ensure the real-time and high quality of network communication and audio and video decoding, provide users with a better user experience, and ensure high bandwidth and low latency fluency even under conditions of large traffic and limited hardware resources. BRIEF DESCRIPTION OF THE DRAWINGS
[0051] Attachment Figure 1 This is a block diagram of system module connections in accordance with the first embodiment of the present invention;
[0052] Attachment Figure 2 is a flow chart of the method of embodiment 2 of the present invention;
[0053] Attachment Figure 3 This is a flowchart of the task scheduling phase of the second embodiment of the present invention;
[0054] Attachment Figure 4 This is a flow chart of the communication management phase of the second embodiment of the present invention. DETAILED DESCRIPTION
[0055] In order to make the technical solution, the technical problems solved and the technical effects of the present invention more clear, the technical solution of the present invention is clearly and completely described below in conjunction with specific embodiments.
[0056] Example 1:
[0057] Combined with attachment Figure 1 This embodiment proposes a low-power intelligent collaborative management system integrating gateways, which includes:
[0058] The communication management module is responsible for building and training a communication traffic prediction model to predict the traffic values of each communication module within a set time period in the future. It dynamically adjusts the working mode of the communication module based on the traffic value prediction results and feeds back the communication resource status to the task scheduling module and energy control module.
[0059] The task scheduling module is responsible for analyzing the characteristics of tasks entering the converged gateway. Based on the communication resource status and power consumption constraints, it uses a reinforcement learning algorithm to generate a task allocation strategy among different processing units, thus achieving intelligent scheduling to reduce task processing power consumption and improve execution efficiency.
[0060] The energy control module is responsible for collecting environmental data and business load data in real time, and building a power consumption prediction model that integrates environmental factors, business load and system power consumption; it is responsible for dynamically adjusting the system power supply parameters based on the real-time collected data and the output results of the power consumption prediction model; it is responsible for triggering the task scheduling module to reallocate tasks when it detects that the system power consumption exceeds the threshold, and notifying the communication management module to adjust the working mode of the communication module to form a closed-loop energy control.
[0061] In this embodiment, refer to the attached Figure 4 The communication management modules involved specifically include:
[0062] A historical data collection unit is used to collect historical communication data of the fusion gateway, including traffic data of different device types (such as smart home devices, industrial sensors, and mobile terminals) through various communication modules (such as Wi-Fi modules, Bluetooth modules, and ZigBee modules) in different time periods (such as weekdays / weekends, daytime / nighttime);
[0063] The model building and training unit is used to build a communication traffic prediction model based on the long short-term memory network (LSTM). The communication traffic prediction model outputs the traffic prediction value of each communication module within a set time period in the future (such as the next 30 minutes) based on the data collected by the historical data collection unit;
[0064] The dynamic working mode switching unit is used to trigger the working mode switching mechanism when the traffic prediction value of a communication module (such as a Bluetooth module) output by the communication traffic prediction model is lower than a preset threshold (such as 20% of the peak traffic) within a set time length (such as within 15 consecutive minutes), and switch the communication module from the normal working mode to the low-power sniffing mode (such as extending the sleep period of the Bluetooth module from 10ms to 100ms); it is also used to wake up the dormant communication module in advance (such as 2 minutes in advance) and optimize parameters (such as increasing Wi-Fi transmission power and adjusting channel allocation) when the traffic prediction value of a communication module output by the communication traffic prediction model reaches a peak within the future time t (such as within the next 5 minutes). At the same time, it notifies the task scheduling module to increase the task allocation priority of the link, thereby achieving coordinated optimization of communication and task processing.
[0065] In this embodiment, refer to the attached Figure 3 , the task scheduling modules involved specifically include:
[0066] The task monitoring unit is responsible for real-time monitoring of task types (such as data processing, image recognition, real-time communication, etc.) when the task enters the fusion gateway;
[0067] The feature analysis unit is used to analyze tasks based on their types, obtain key feature information of the tasks (such as data transmission volume, computational complexity, real-time requirements, memory requirements, etc.), and quantify and encode them into vector form suitable for algorithm processing, providing a data basis for subsequent task allocation decisions;
[0068] The reinforcement learning task allocation unit is used to build a task allocation environment based on the feature quantization results. By designing a reward function for task completion time, processing power consumption, and priority weight, the reinforcement learning algorithm is trained to generate the optimal strategy mapping. In combination with the communication resource status feedback from the communication management module and the power consumption constraints of the energy control module, the optimal task allocation action is automatically selected according to the current task and system status.
[0069] Specifically, the task allocation environment constructed by the reinforcement learning task allocation unit based on the task feature quantification results includes a task state space and a task action space, where the task state space includes the current load of each processing unit, the remaining computing resources, and the task feature information in the task queue, and the task action space includes all possible choices for allocating tasks to different processing units.
[0070] In this embodiment, the energy control module involved specifically includes:
[0071] The environment and load sensing unit uses sensor devices to collect real-time data on ambient temperature, light intensity, and humidity. It also uses system detection tools within the converged gateway to obtain real-time service load data such as CPU utilization, memory occupancy, and task queue length, and shares this data with the task scheduling module and communication management module.
[0072] The model building unit is used to build a power consumption prediction model based on historical data and physical relationships, integrating environmental factors (e.g., power consumption increases by 5%-10% for every 10°C temperature increase), business load (e.g., power consumption increases by 15%-20% for every 20% CPU increase), and system power consumption.
[0073] The energy control algorithm unit is used to dynamically adjust the power supply parameters of the converged gateway (such as reducing the 3.3V voltage to 2.8V and the 1GHz frequency to 800MHz) using adaptive voltage and frequency scaling (AVFS) technology based on the real-time data of the environment and load sensing unit and the output results of the power consumption prediction model. It is also used to coordinate with the task scheduling stage to optimize the task allocation strategy when it detects that the system power consumption exceeds the threshold, and to coordinate with the communication management stage to adjust the working mode of the communication module.
[0074] Example 2:
[0075] Combined with attachment Figure 2-4 This embodiment proposes a low-power intelligent collaborative management method for a converged gateway, comprising the following steps:
[0076] S1. Communication management stage: By collecting and integrating historical communication data of the gateway, a communication traffic prediction model is constructed and trained to predict the traffic value of each communication module within a set time period in the future; the working mode of the communication module is dynamically adjusted according to the traffic value prediction results, and the communication resource status is fed back to the task scheduling stage and energy regulation stage.
[0077] The specific operations performed during the communication management phase include:
[0078] S1.1. Collect historical communication data from the fusion gateway, including traffic data from different device types (e.g., smart home devices, industrial sensors, mobile terminals) through various communication modules (e.g., Wi-Fi, Bluetooth, ZigBee) during different time periods (e.g., weekdays / weekends, daytime / nighttime).
[0079] S1.2. Build a communication traffic prediction model based on a long short-term memory (LSTM) network. The communication traffic prediction model outputs traffic prediction values for each communication module within a set time period (e.g., within the next 30 minutes) based on the collected historical communication data of the converged gateway.
[0080] S1.3. When the traffic prediction value of a communication module (e.g., a Bluetooth module) output by the communication traffic prediction model is lower than a preset threshold (e.g., 20% of the peak traffic) for a set period of time (e.g., 15 consecutive minutes), the working mode switching mechanism is triggered, switching the communication module from the normal working mode to the low-power sniffing mode (e.g., extending the sleep period of the Bluetooth module from 10ms to 100ms);
[0081] S1.4. When the traffic prediction value of a communication module output by the communication traffic prediction model reaches its peak within the next time t (e.g., within the next 5 minutes), the dormant communication module is awakened at a set time in advance (e.g., 2 minutes in advance) and its parameters are optimized (e.g., increasing Wi-Fi transmission power, adjusting channel allocation). At the same time, the task scheduling stage is notified to increase the priority of task allocation for this link, thereby achieving coordinated optimization of communication and task processing.
[0082] S2, Task Scheduling Stage: Perform feature analysis on tasks entering the converged gateway, combine the communication resource status and power consumption constraints, and generate a task allocation strategy among different processing units through a reinforcement learning algorithm to achieve intelligent scheduling to reduce task processing power consumption and improve execution efficiency.
[0083] The task scheduling phase specifically includes the following operations:
[0084] S2.1. Real-time monitoring of task types (e.g., data processing, image recognition, real-time communication, etc.) when tasks enter the fusion gateway.
[0085] S2.2. Analyze tasks based on their types, extract key task characteristics (such as data transfer volume, computational complexity, real-time requirements, memory requirements, etc.), and quantify and encode them into vector form suitable for algorithm processing, providing a data basis for subsequent task allocation decisions;
[0086] S2.3. Build a task allocation environment based on the feature quantization results. By designing a reward function based on task completion time, processing power consumption, and priority weight, train the reinforcement learning algorithm to generate the optimal strategy mapping. Combined with the communication resource status feedback from the communication management phase and the power consumption constraints from the energy control phase, the optimal task allocation action is automatically selected based on the current task and system status.
[0087] It should be noted that the task allocation environment constructed based on the quantified results of task features includes the task state space and the task action space, where:
[0088] The task state space includes the current load of each processing unit, the remaining computing resources, and the task feature information in the task queue;
[0089] The task action space includes all possible choices for assigning tasks to different processing units.
[0090] S3, Energy Control Phase: Collect environmental data and business load data in real time, build a power consumption prediction model that integrates environmental factors, business load, and system power consumption; dynamically adjust system power supply parameters based on real-time collected data and the output of the power consumption prediction model; trigger the task scheduling phase to reallocate tasks when the system power consumption exceeds the threshold, and notify the communication management phase to adjust the working mode of the communication module to form a closed-loop energy control.
[0091] The specific operations performed during the energy regulation phase are as follows:
[0092] S3.1. Use sensor devices to collect real-time data on ambient temperature, light intensity, and humidity. Use system monitoring tools within the converged gateway to obtain real-time data on business load, such as CPU utilization, memory usage, and task queue length. This data is then synchronized with the task scheduling and communication management stages.
[0093] S3.2. Based on historical data and physical relationships, build a power consumption prediction model that integrates environmental factors (e.g., power consumption increases by 5%-10% for every 10°C temperature increase), business load (e.g., power consumption increases by 15%-20% for every 20% CPU increase), and system power consumption.
[0094] S3.3. Based on the real-time data of step S3.1 and the output results of the power consumption prediction model constructed in step S3.2, the adaptive voltage frequency scaling (AVFS) technology is used to dynamically adjust the power supply parameters of the converged gateway (such as reducing the 3.3V voltage to 2.8V and the 1GHz frequency to 800MHz); when the system power consumption is monitored to exceed the threshold, the task allocation strategy is optimized in coordination with the task scheduling stage, and the communication module working mode is adjusted in conjunction with the communication management stage.
[0095] In summary, the low-power intelligent collaborative management system and method of the integrated gateway of the present invention can significantly reduce the power consumption of the integrated gateway, improve its energy utilization efficiency in home multimedia scenarios, extend the service life of the equipment, and reduce energy costs.
[0096] The above specific examples are used to illustrate the principles and implementation methods of the present invention in detail. These examples are only used to help understand the core technical content of the present invention. Based on the above specific embodiments of the present invention, any improvements and modifications made by those skilled in the art without departing from the principles of the present invention should fall within the scope of patent protection of the present invention.
Claims
1. A low-power intelligent collaborative management system integrating gateways, characterized in that: It includes: The communication management module is responsible for building and training a communication traffic prediction model to predict the traffic values of each communication module within a set time period in the future. It dynamically adjusts the working mode of the communication module based on the traffic value prediction results and feeds back the communication resource status to the task scheduling module and energy control module. The task scheduling module is responsible for analyzing the characteristics of tasks entering the converged gateway. Based on the communication resource status and power consumption constraints, it uses a reinforcement learning algorithm to generate a task allocation strategy among different processing units, thus achieving intelligent scheduling to reduce task processing power consumption and improve execution efficiency. The energy control module is responsible for collecting environmental data and business load data in real time, and building a power consumption prediction model that integrates environmental factors, business load and system power consumption; it is responsible for dynamically adjusting the system power supply parameters based on the real-time collected data and the output results of the power consumption prediction model; it is responsible for triggering the task scheduling module to reallocate tasks when it detects that the system power consumption exceeds the threshold, and notifying the communication management module to adjust the working mode of the communication module to form a closed-loop energy control.
2. The low-power intelligent collaborative management system of the integrated gateway according to claim 1 is characterized in that: The communication management module specifically includes: A historical data collection unit is used to collect historical communication data of the converged gateway, including traffic data of different device types passing through each communication module in different time periods; A model building and training unit is used to build a communication traffic prediction model based on a long short-term memory network. The communication traffic prediction model outputs the traffic prediction value of each communication module within a set time period in the future based on the data collected by the historical data collection unit; The dynamic working mode switching unit is used to trigger the working mode switching mechanism when the traffic prediction value of a communication module output by the communication traffic prediction model is lower than the preset threshold within a set time length, and switch the communication module from the normal working mode to the low-power sniffing mode; it is also used to set the time in advance to wake up the dormant communication module and optimize the parameters when the traffic prediction value of a communication module output by the communication traffic prediction model reaches a peak within the future time t, and at the same time notify the task scheduling module to increase the task allocation priority of the link, so as to achieve coordinated optimization of communication and task processing.
3. The low-power intelligent collaborative management system of the integrated gateway according to claim 2 is characterized in that: The task scheduling module specifically includes: The task monitoring unit is responsible for real-time monitoring of the task type when the task enters the fusion gateway; The feature analysis unit is used to analyze tasks based on their types, obtain key feature information of the tasks, and quantify and encode them into vector forms suitable for algorithm processing, providing a data basis for subsequent task allocation decisions; The reinforcement learning task allocation unit is used to build a task allocation environment based on the feature quantization results. By designing a reward function for task completion time, processing power consumption, and priority weight, the reinforcement learning algorithm is trained to generate the optimal strategy mapping. In combination with the communication resource status feedback from the communication management module and the power consumption constraints of the energy control module, the optimal task allocation action is automatically selected according to the current task and system status.
4. The low-power intelligent collaborative management system of the integrated gateway according to claim 3 is characterized in that: The task allocation environment constructed by the reinforcement learning task allocation unit based on the task feature quantization results includes a task state space and a task action space. The task state space includes the current load of each processing unit, the remaining computing resources, and the task feature information in the task queue, while the task action space includes all possible options for allocating tasks to different processing units.
5. The low-power intelligent collaborative management system of the integrated gateway according to claim 1, characterized in that: The energy control module specifically includes: The environment and load sensing unit is used to collect real-time data on ambient temperature, light intensity, and humidity using sensor equipment, and to obtain real-time service load data using the system detection tools within the fusion gateway, and to share this data with the task scheduling module and the communication management module. A model building unit is used to build a power consumption prediction model that integrates environmental factors, business load, and system power consumption based on historical data and physical relationships; The energy control algorithm unit is used to dynamically adjust the power supply parameters of the integrated gateway using adaptive voltage and frequency scaling technology based on the real-time data of the environment and load sensing unit and the output results of the power consumption prediction model. It is also used to coordinate with the task scheduling stage to optimize the task allocation strategy when it detects that the system power consumption exceeds the threshold, and to coordinate with the communication management stage to adjust the working mode of the communication module.
6. A low-power intelligent collaborative management method for a fusion gateway, characterized in that: It includes the following steps: S1, Communication Management Phase: By collecting and integrating historical communication data from the gateway, a communication traffic prediction model is constructed and trained to predict the traffic values of each communication module within a set time period in the future. The working mode of the communication module is dynamically adjusted based on the traffic value prediction results, and the communication resource status is fed back to the task scheduling phase and energy control phase. S2, Task Scheduling: Analyze the characteristics of tasks entering the converged gateway. Combined with the communication resource status and power consumption constraints, a reinforcement learning algorithm is used to generate a task allocation strategy among different processing units. This enables intelligent scheduling to reduce task processing power consumption and improve execution efficiency. S3, Energy Control Phase: Collect environmental data and business load data in real time, build a power consumption prediction model that integrates environmental factors, business load, and system power consumption; dynamically adjust system power supply parameters based on real-time collected data and the output of the power consumption prediction model; trigger the task scheduling phase to reallocate tasks when the system power consumption exceeds the threshold, and notify the communication management phase to adjust the working mode of the communication module to form a closed-loop energy control.
7. The low-power intelligent collaborative management method of a converged gateway according to claim 6, characterized in that: The step S1 specifically includes: S1.
1. Collect historical communication data of the converged gateway, including traffic data of different device types passing through each communication module in different time periods; S1.
2. Build a communication traffic prediction model based on the long short-term memory network. The communication traffic prediction model outputs the traffic prediction value of each communication module within a set time period in the future based on the collected historical communication data of the converged gateway; S1.
3. When the traffic prediction value of a communication module output by the communication traffic prediction model is lower than a preset threshold within a set period of time, a working mode switching mechanism is triggered to switch the communication module from the normal working mode to the low-power sniffing mode; S1.
4. When the traffic prediction value of a certain communication module output by the communication traffic prediction model reaches its peak value within the next t time, the time is set in advance to wake up the dormant communication module and optimize the parameters. At the same time, the task scheduling stage is notified to increase the priority of task allocation to this link, so as to achieve coordinated optimization of communication and task processing.
8. The low-power intelligent collaborative management method of a converged gateway according to claim 7, characterized in that: The step S2 specifically includes the following operations: S2.1, monitor the task type in real time when the task enters the fusion gateway; S2.
2. Analyze tasks based on their types, extract key feature information, and quantify and encode them into vector form suitable for algorithm processing, providing a data basis for subsequent task allocation decisions; S2.
3. Build a task allocation environment based on the feature quantization results. By designing a reward function based on task completion time, processing power consumption, and priority weight, train the reinforcement learning algorithm to generate the optimal strategy mapping. Combined with the communication resource status feedback from the communication management phase and the power consumption constraints from the energy control phase, the optimal task allocation action is automatically selected based on the current task and system status.
9. The low-power intelligent collaborative management method of a converged gateway according to claim 8, characterized in that: The task allocation environment constructed based on the quantified results of task features includes task state space and task action space, where: The task state space includes the current load of each processing unit, the remaining computing resources, and the task feature information in the task queue; The task action space includes all possible choices for assigning tasks to different processing units.
10. The low-power intelligent collaborative management method of a converged gateway according to claim 6, characterized in that: The step S3 specifically includes: S3.
1. Use sensor equipment to collect real-time data on ambient temperature, light intensity, and humidity. Use system detection tools within the converged gateway to obtain real-time service load data. This data is synchronized with the task scheduling and communication management stages. S3.
2. Build a power consumption prediction model that integrates environmental factors, service load, and system power consumption based on historical data and physical relationships. S3.
3. Based on the real-time data of step S3.1 and the output results of the power consumption prediction model constructed in step S3.2, the power supply parameters of the integrated gateway are dynamically adjusted using adaptive voltage and frequency scaling technology; when the system power consumption is detected to exceed the threshold, the task allocation strategy is optimized in collaboration with the task scheduling stage, and the communication module working mode is adjusted in conjunction with the communication management stage.