Energy-saving control method for equipment, storage medium, electronic device and program product

Abstract

The present application provides an energy-saving control method, storage medium, electronic device, and program product for a device. The method includes: a fiber-to-the-room (FTTR) master gateway constructing a device status database based on collected device data, wherein the FTTR master gateway includes an artificial intelligence (AI) model; the FTTR master gateway uses the AI ​​model to analyze the device data in the device status database to determine energy-saving control instructions, and then sends the energy-saving control instructions to the corresponding device. The present application solves the problem in the related art that the method of only issuing fixed energy-saving control instructions cannot accurately adapt to complex and changing network environments, resulting in low energy-saving efficiency.

Description

Technical Field

[0001] The embodiments of the present application relate to the field of communications, and specifically, to an energy-saving control method for a device, a storage medium, an electronic device, and a program product. Background Art

[0002] With the continuous development of optical networks, fiber to the room (FTTR) has been fully rolled out. FTTR technology uses optical fiber to connect FTTR slave units (SFUs) in different rooms or locations in homes, small and medium-sized enterprises, and other scenarios. This provides high-bandwidth, highly reliable connections between multiple SFUs. Main FTTR units (MFUs) can be connected to SFUs using a point-to-multipoint optical distribution network. The wired backhaul provided by FTTR technology allows MFUs and SFUs to exchange information via wired transmission, eliminating the need for frequent information exchange to consume large amounts of wireless time and frequency resources.

[0003] With the widespread adoption of FTTR technology, the energy consumption of network equipment has become increasingly prominent. Traditional energy-saving methods, often based on fixed rules or simple thresholds, cannot accurately adapt to complex and changing network environments and user needs. In FTTR systems, the varying operating states of devices such as master and slave gateways, as well as the diverse needs of various services, make achieving efficient energy conservation a pressing challenge. Summary of the Invention

[0004] The embodiments of the present application provide an energy-saving control method, storage medium, electronic device and program product for a device, so as to at least solve the problem in the related art that the device cannot accurately adapt to the complex and changing network environment due to the fact that only fixed energy-saving control instructions can be issued, resulting in low energy-saving efficiency.

[0005] According to one embodiment of the present application, a method for energy-saving control of a device is provided, including: a fiber-to-the-room FTTR main gateway constructs a device status database based on collected device data, wherein the FTTR main gateway includes an artificial intelligence (AI) model; the FTTR main gateway analyzes the device data in the device status database by using the AI ​​model to determine an energy-saving control instruction, and sends the energy-saving control instruction to the corresponding device.

[0006] According to another embodiment of the present application, a computer-readable storage medium is provided, in which a computer program is stored. The computer program is configured to execute the steps in the above method embodiment when running.

[0007] According to another embodiment of the present application, an electronic device is provided, including a memory and a processor, wherein the memory stores a computer program, and the processor is configured to run the computer program to execute the steps in the above method embodiment.

[0008] According to another embodiment of the present application, a computer program product is provided, including a computer program, which implements the steps in the above method embodiment when executed by a processor.

[0009] Through the above-mentioned embodiments of this application, the intelligent analysis of the AI ​​model can more accurately match the device status and reduce unnecessary energy consumption, thereby significantly improving energy utilization efficiency and reducing operating costs. Because the AI ​​model can automatically identify energy-saving opportunities based on the status of the device and generate refined energy-saving control instructions, it can achieve dynamic energy-saving management of the device, avoiding the limitations of traditional energy-saving methods based on fixed rules when responding to network changes. BRIEF DESCRIPTION OF THE DRAWINGS

[0010] Figure 1 It is a hardware structure block diagram of a computer terminal according to the energy-saving control method of the device of the embodiment of the present application;

[0011] Figure 2 is a flow chart of an energy-saving control method for a device according to an embodiment of the present application;

[0012] Figure 3 is a schematic diagram of an FTTR system scenario according to an embodiment of the present application;

[0013] Figure 4 is a structural block diagram of an energy-saving control device for a device according to an embodiment of the present application;

[0014] Figure 5 This is a schematic diagram of the AI ​​model architecture according to an embodiment of the present application;

[0015] Figure 6 Schematic diagram of an energy-saving state machine according to an embodiment of the present application. DETAILED DESCRIPTION

[0016] The embodiments of the present application will be described in detail below with reference to the accompanying drawings and in combination with the embodiments.

[0017] It should be noted that the terms "first", "second", etc. in the description and claims of this application and the above-mentioned drawings are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence.

[0018] The method embodiments provided in the embodiments of the present application can be executed in a mobile terminal, a computer terminal or a similar computing device. Taking running on a computer terminal as an example, Figure 1This is a hardware block diagram of the computer terminal running the method embodiment of this application. Figure 1 As shown, the computer terminal may include one or more ( Figure 1 Only one is shown) a processor 102 (the processor 102 may include but is not limited to a microprocessor MCU or a programmable logic device FPGA) and a memory 104 for storing data. The computer terminal may also include a transmission device 106 and an input / output device 108 for communication functions. It will be understood by those skilled in the art that Figure 1 The structure shown is only for illustration and does not limit the structure of the above-mentioned computer terminal. For example, the computer terminal may also include Figure 1 More or fewer components than shown, or with Figure 1 Different configurations shown.

[0019] The memory 104 can be used to store computer programs, for example, software programs and modules of application software, such as the computer program corresponding to the energy-saving control method in the embodiment of the present application. The processor 102 executes various functional applications and data processing by running the computer program stored in the memory 104, that is, implementing the above-mentioned method. The memory 104 may include a high-speed random access memory, and may also include a non-volatile memory, such as one or more magnetic storage devices, flash memory, or other non-volatile solid-state memory. In some instances, the memory 104 may further include a memory remotely located relative to the processor 102, and these remote memories can be connected to the computer terminal via a network. Examples of the above-mentioned network include, but are not limited to, the Internet, an intranet, a local area network, a mobile communication network, and combinations thereof.

[0020] Transmission device 106 is used to receive or transmit data via a network. A specific example of such a network may include a wireless network provided by a computer terminal's communications provider. In one embodiment, transmission device 106 includes a network interface controller (NIC), which can be connected to other network devices via a base station to enable communication with the Internet. In another embodiment, transmission device 106 may be a radio frequency (RF) module, which is used to communicate with the Internet wirelessly.

[0021] In this embodiment, a method for controlling energy saving running on the above-mentioned computer terminal is provided. Figure 2 is a flow chart of the energy-saving control method according to an embodiment of the present application, such as Figure 2 As shown, the process includes the following steps:

[0022] Step S202: The fiber-to-the-room FTTR master gateway builds a device status database based on the collected device data, wherein the FTTR master gateway includes an artificial intelligence (AI) model;

[0023] In an exemplary embodiment of the present application, the device data includes at least one of the following: Passive Optical Network (PON) port traffic data, Wireless Fidelity (WIFI) traffic data, Local Area Network (LAN) port traffic, device status data, power consumption information, and alarm information.

[0024] In one embodiment, device data is transmitted to the AI ​​model in the FTTR master gateway in real time or at a scheduled time via a specific transmission protocol. For example, the FTTR master gateway chip obtains real-time traffic data from each device port and sends it to the AI ​​model via wired or wireless means.

[0025] In an exemplary embodiment of the present application, the AI ​​model is an AI model that has undergone feature dimensionality reduction and model compression.

[0026] In one embodiment, feature dimensionality reduction (such as principal component analysis (PCA)) and model compression technology are used to reduce the AI ​​model's dependence on hardware resources, generating a lightweight AI model that supports efficient operation in the FTTR main gateway.

[0027] In an exemplary embodiment of the present application, the fiber-to-the-room FTTR main gateway constructs a device status database based on the collected device data, including: the FTTR main gateway uses data feature extraction and preprocessing algorithms to clean and normalize the collected device data to obtain preprocessed device data; the FTTR main gateway correlates and integrates the preprocessed device data based on the network behavior characteristics of the device and the correlation of network resources to construct a device status database.

[0028] In one embodiment, data such as PON / Wi-Fi / LAN traffic, device power consumption, and user connection time periods are integrated to create a real-time network status profile, which serves as a device status database. This integration of different device operating parameters provides a comprehensive data foundation for subsequent analysis and decision-making. For example, information such as device CPU usage, memory utilization, and the number of connected devices and traffic distribution across each Wi-Fi band is collected.

[0029] In an exemplary embodiment of the present application, before the fiber-to-the-room FTTR main gateway builds a device status database based on the collected device data, it also includes: the FTTR main gateway collects the device data uploaded by the terminal device connected to the FTTR main gateway according to a preset period, and the device data uploaded by the FTTR slave gateway device connected to the FTTR main gateway.

[0030] In one embodiment, the FTTR system scenario diagram is as follows: Figure 3 As shown in the figure, the FTTR master gateway is internally deployed with an AI model and an energy-saving data collection module. As the core of the entire FTTR system, the FTTR master gateway connects not only to the FTTR slave gateways but also to various terminal devices (such as personal computers, smartphones, and smart home appliances). The energy-saving data collection module collects device data uploaded by terminal devices and the FTTR slave gateways. The FTTR slave gateways are also internally deployed with a data collection module to collect device data from terminal devices connected to the FTTR slave gateways.

[0031] Step S204: The FTTR master gateway analyzes the device data in the device status database using an AI model to determine an energy-saving control instruction, and sends the energy-saving control instruction to the corresponding device;

[0032] In an exemplary embodiment of the present application, the FTTR main gateway determines energy-saving control instructions by analyzing the device data in the device status database using an AI model, including: the FTTR main gateway uses the AI ​​model to analyze the network status of the device in the device status database to determine whether the network status of the device meets the energy-saving conditions, wherein the AI ​​model includes at least one of the following: decision tree, naive Bayes and long short-term memory network LSTM; if the judgment result is yes, the FTTR main gateway determines the energy-saving strategy of the device according to the device type and service priority, and generates energy-saving control instructions according to the energy-saving strategy.

[0033] In one embodiment, the AI ​​model determines whether the device's network status meets energy-saving requirements. For example, if the average traffic volume is below a threshold and the device load is light, it triggers power reduction or shuts down non-essential modules. The model also differentiates execution strategies based on device type and service priority (e.g., retaining only critical service modules).

[0034] In one embodiment, when the transmit power of a device's Wi-Fi module needs to be reduced, the AI ​​model sends an instruction containing power adjustment parameters to the device. After receiving the instruction, the device adjusts the transmit power of the Wi-Fi module as required to achieve energy saving.

[0035] In an exemplary embodiment of the present application, after sending the energy-saving control instruction to the corresponding device, it also includes: updating the parameters of the AI ​​model according to the operating data of the device.

[0036] In one embodiment, device operating data is continuously collected to train and optimize the energy-saving AI model. New data is used to continuously update AI model parameters, improving the model's accuracy and adaptability. For example, if an energy-saving operation causes a network anomaly, the model will automatically adjust its strategy to prevent similar issues from recurring, forming a closed-loop optimization process of "perception-decision-execution-feedback."

[0037] In an exemplary embodiment of the present application, the energy-saving control instruction is sent to the corresponding device, including: classifying and encapsulating the energy-saving control instruction according to the device type and communication protocol, and sending it to the corresponding device through the Fiber Multimedia Control Interface (FMCI) protocol, the Wireless Multimedia Control Interface (WMCI) protocol, the Fiber Physical Layer Operations Administration and Maintenance (F-PLOAM) protocol or a custom control protocol.

[0038] In one embodiment, the AI ​​model generates energy-saving control instructions, categorizes and encapsulates them by device type and communication protocol, and sends them to the corresponding devices via FMCI, WMCI, F-PLOAM, or a custom control protocol. For example, when deciding to reduce the transmit power of a slave gateway's Wi-Fi module, a command containing power adjustment parameters is sent to the slave gateway. Upon receiving the command, the slave gateway adjusts the power accordingly, achieving energy savings.

[0039] In one embodiment, the FTTR master gateway's energy-saving agent works collaboratively with a complex cloud-based AI model to facilitate complex energy-saving decisions and scenario analysis. The agent initially processes local data and uploads it to the cloud. The cloud-based AI model leverages its powerful computing power and richer datasets for in-depth analysis. Based on deep learning algorithms such as convolutional neural networks (CNNs) and long-short-term memory (LSTMs), the cloud-based model accurately predicts and analyzes network traffic trends and user behavior patterns. Based on the analysis results, the cloud-based model generates detailed energy-saving recommendations and strategies, which the agent receives and executes, achieving more efficient energy-saving control.

[0040] Through the description of the above embodiments, those skilled in the art can clearly understand that the methods according to the above embodiments can be implemented by means of software plus the necessary general hardware platform. Of course, they can also be implemented by hardware, but in many cases the former is a better implementation method. Based on this understanding, the technical solution of the present application, or the part that contributes to the existing technology, can be embodied in the form of a software product. The computer software product is stored in a storage medium (such as ROM / RAM, magnetic disk, optical disk), and includes a number of instructions for enabling a terminal device (which can be a mobile phone, computer, server, or network device, etc.) to execute the methods described in each embodiment of the present application.

[0041] This embodiment also provides an energy-saving control device for a device, which is used to implement the above-mentioned embodiments and preferred embodiments. Details that have already been described will not be repeated. As used below, the term "module" may refer to a combination of software and / or hardware that implements a predetermined function. Although the devices described in the following embodiments are preferably implemented in software, implementation using hardware, or a combination of software and hardware, is also possible and contemplated.

[0042] Figure 4 is a structural block diagram of an energy-saving control device for a device according to an embodiment of the present application, such as Figure 4 As shown, the device includes: a building module 10 and a sending module 20.

[0043] A construction module 10 is used to construct a device status database based on the collected device data, wherein the construction module 10 includes an artificial intelligence AI model;

[0044] The sending module 20 is used to analyze the device data in the device status database by using the AI ​​model to determine the energy-saving control instruction, and send the energy-saving control instruction to the corresponding device.

[0045] It should be noted that the above modules can be implemented through software or hardware. For the latter, it can be implemented in the following ways, but not limited to: the above modules are all located in the same processor; or the above modules are located in different processors in any combination.

[0046] In order to facilitate the understanding of the technical solutions provided by the application embodiments, the embodiments are described below in conjunction with specific scenarios.

[0047] Scenario Example 1

[0048] The process of the device energy-saving control method according to another embodiment of the present application includes the following steps:

[0049] In step S501, data from the master and slave devices is collected in the FTTR system, including optical port traffic, Wi-Fi traffic, LAN port traffic, power consumption, and alarm information. This data is transmitted in real time or periodically to the AI ​​model in the master gateway via a specific transmission protocol. For example, the chips in the master and slave devices acquire real-time traffic data from each device port and transmit it to the AI ​​model via wired or wireless means.

[0050] Specifically, for optical port traffic, optical power sensors and traffic monitoring chips are used to accurately measure traffic flow on the optical links between the master and slave gateways, and between the slave gateways and end devices, collecting data at regular intervals (e.g., every second). When collecting Wi-Fi traffic, the Wi-Fi chip's statistics function captures upload and download traffic data for different frequency bands (2.4G and 5G) and service set identifiers (SSIDs), as well as information such as the Media Access Control (MAC) address and signal strength of the connected devices. LAN port traffic collection utilizes a network interface controller to monitor the data transmission volume of devices connected to the LAN port (such as computers and smart appliances). For power consumption information collection, power consumption sensors are installed on the master and slave gateways, as well as key end devices (such as high-power smart TVs), to monitor device power consumption in real time and convert it into energy consumption data. Alarm information is generated by the device's built-in fault detection program. If a device anomaly (such as port failure or signal loss) occurs, an alarm is immediately generated and transmitted to the data collection module.

[0051] Specifically, the collected data is transmitted via Ethernet, Wi-Fi, and other network transmission methods, encapsulated in a custom data packet format, and transmitted to the AI ​​model in the main gateway. During the transmission process, checksum and error correction algorithms are used to ensure the accuracy and integrity of the data. After receiving the data, the AI ​​model first performs data cleaning to remove abnormal data caused by interference or equipment failure. For sudden changes in traffic data, a sliding window algorithm is used for smoothing; for power consumption data, a reasonable threshold range is set to eliminate abnormal values ​​that exceed the range. Data normalization is then performed to unify data of different types and magnitudes into the same numerical range to facilitate subsequent analysis and processing.

[0052] In step S502, after receiving data such as network / device status and power consumption, the AI ​​model constructs a real-time profile of the system's operating status. It integrates the operating parameters of different devices, providing a comprehensive data foundation for subsequent analysis and decision-making. For example, the AI ​​model collects information such as the central processing unit (CPU) utilization and memory usage of the master and slave devices, as well as the number of connected devices and traffic distribution on each Wi-Fi band.

[0053] Specifically, the AI ​​model continuously receives preprocessed data and builds a real-time database of network device status. It correlates and integrates data from different collection sources. For example, it associates Wi-Fi traffic data from a slave gateway with the MAC address, device type, and corresponding LAN port traffic data of the device connected to the slave gateway, forming a complete record of network usage. It also records device power consumption and alarm information to facilitate comprehensive analysis of network operation status and energy consumption.

[0054] In step S503, an AI model (such as a decision tree, Naive Bayesian, or LSTM) analyzes data in real time to determine energy-saving conditions. For example, if the average traffic volume falls below a threshold and the device load is light, power reduction or the shutdown of non-essential modules may be triggered. The model also differentiates execution strategies based on device type and service priority (for example, retaining only critical service modules).

[0055] Specifically, feature dimensionality reduction (PCA) and model compression techniques are used to reduce the AI ​​model's reliance on hardware resources, enabling it to run efficiently on FTTR mainframe equipment. For example, 5-dimensional data is reduced to 3 dimensions, reducing computational complexity. Minimum-maximum normalization is used to unify data levels, improving model inference speed (with latency under 150ms). This optimization ensures the feasibility of AI models on resource-constrained devices.

[0056] Specifically, the AI ​​model performs differentiated power-saving actions based on service type (such as video streaming and Voice over Internet Protocol (VoIP)) and device grouping (such as high-power groups). For example, chip CPU cores are shut down and chip frequency is reduced; Wi-Fi is turned off, Wi-Fi transmit power is reduced, and Multiple-Input Multiple-Output (MIMO) is downgraded; the PON energy-saving state machine enters doze mode and sleep mode; network ports reduce the negotiated rate, switch from Gigabit Ethernet (GE) to Fast Ethernet (FE), and shut down idle network ports; and the Universal Serial Bus (USB) enters low-power mode, idle for extended periods, and the Physical Layer Medium Access Control (PHY MAC) is put into sleep mode, suspending all functions.

[0057] Specifically, the AI ​​model is started, which is built based on the decision tree algorithm. First, the network traffic data is analyzed, and statistical characteristics such as the mean and variance of the traffic are calculated to determine the stability and change trend of the traffic. Combined with the device load (such as CPU usage and memory occupancy), if the traffic average is lower than a preset threshold for a continuous period of time (such as 10 minutes) and the device load is at a low level, the decision tree model determines that the current network state is suitable for energy-saving operations. The model further infers specific energy-saving measures based on the device type and service priority. For slave gateways with few connected devices and extremely low traffic, the decision tree model outputs the decision to shut down some of its Wi-Fi bands (such as retaining only the 2.4G band) or reduce the transmission power of the Wi-Fi module; for some LAN ports connected to non-critical terminal devices, the decision tree model decides to reduce the negotiation rate of the port to reduce energy consumption.

[0058] In step S504, for complex energy-saving decisions and scenario analysis, the Energy Saving Agent collaborates with a complex cloud-based AI model. The Energy Saving Agent performs preliminary processing on local data and uploads it to the cloud. The cloud-based AI model leverages its powerful computing power and richer datasets for in-depth analysis. Based on deep learning algorithms such as convolutional neural networks (CNNs) and long short-term memory networks (LSTMs), the cloud-based model accurately predicts and analyzes network traffic trends and user behavior patterns. Based on the analysis results, the cloud-based model generates detailed energy-saving recommendations and strategies. The Energy Saving Agent receives and executes these recommendations, achieving more efficient energy-saving control.

[0059] In step S505, the AI ​​model generates an energy-saving decision, generates an energy-saving control instruction based on the decision, classifies and encapsulates it according to the device type and communication protocol, and sends it to the corresponding device via FMCI, WMCI, F-PLOAM, or a custom control protocol. For example, when deciding to reduce the transmit power of the WiFi module of a slave gateway, an instruction containing power adjustment parameters is sent to the slave gateway. After receiving the instruction, the slave gateway adjusts it as required to achieve energy saving. If it is detected that the traffic on the LAN port connected to a terminal device is extremely low and lasts for a long time, the AI ​​model sends an instruction to reduce the negotiation rate of the LAN port. After receiving the instruction, the device makes corresponding adjustments.

[0060] Step S506: Continuously collect operational data from the FTTR system to train and optimize the AI ​​model. New data is used to continuously update model parameters, improving its accuracy and adaptability. For example, if an energy-saving operation causes a network anomaly, the model will automatically adjust its strategy to prevent similar issues from recurring, forming a closed-loop optimization process of "perception-decision-execution-feedback."

[0061] Scenario Example 2

[0062] The data collection module in the FTTR master gateway collects data every minute on each slave gateway, including the number of connected Wi-Fi devices, device type (e.g., mobile phone, computer, smart appliance), Wi-Fi traffic, device operating mode (normal, energy-saving, sleep), and corresponding power consumption. (For example, at 10 p.m., a room has three mobile phones and one computer connected to the gateway, with Wi-Fi traffic of 20 Mbps, normal operating mode, and power consumption of 10 W.) This data is integrated and stored in the AI ​​model's database.

[0063] Extract features from the integrated data, encode device types, and calculate the hourly average number of Wi-Fi connected devices and average Wi-Fi traffic. After analysis, select the average number of Wi-Fi connected devices, device type codes, and average Wi-Fi traffic as key features. (For example, calculate that the average number of Wi-Fi connected devices for this slave gateway between 9:00 PM and 10:00 PM is 3.5, and the average Wi-Fi traffic is 22 Mbps.)

[0064] A Naive Bayesian model was selected to develop energy-saving strategies. Training was performed using historical data from the past month, with the average number of connected Wi-Fi devices, device type codes, and average Wi-Fi traffic as input features, and the corresponding device operating mode (normal, energy-saving, or sleep) as output. (For example, if the number of connected devices is small (e.g., less than two), the majority of devices are mobile phones, and Wi-Fi traffic is less than 15 Mbps, the corresponding operating mode is sleep.) The training algorithm calculates the probability distribution of each feature under different operating modes, constructing a Naive Bayesian model.

[0065] The AI ​​model acquires the latest network and device status data in real time. This real-time data is fed into a trained Naive Bayesian model for inference. (For example, at 2:00 AM, real-time data from a slave gateway shows that there is one Wi-Fi device connected (a mobile phone) and Wi-Fi traffic is 8 Mbps. The model calculates the probabilities of different operating modes based on the input features and, by comparing the probabilities, concludes that the slave gateway is most likely to enter sleep mode.)

[0066] According to the reasoning results of the Naive Bayes model, the preset energy-saving rules are matched and the specific energy-saving strategy is determined to switch the gateway to sleep mode.

[0067] The AI ​​model generates control instructions, which include information about switching to sleep mode and the device identifier of the target slave gateway. The Wi-Fi energy-saving decision is delivered to the SFU via the WMCI message "SFU Energy-Saving Decision Configuration," primarily including frequency band, number of streams, bandwidth, MCS, transmit power, and energy-saving period. Upon receiving the instruction, the slave gateway makes adjustments accordingly to achieve energy savings. The master gateway can also deliver energy-saving decisions to the SFU via other messages (such as F-PLOAM and FMCI messages). For example, an F-PLOAM message can be used to deliver a PON energy-saving state adjustment decision to the SFU. The SFU then shuts down unnecessary functional modules and enters sleep mode.

[0068] Two hours after executing the sleep mode policy, the energy-saving effect is evaluated. (For example, by comparing power consumption data before and after execution, the slave gateway's power consumption was found to have dropped from 8W to 4W, indicating significant energy savings.) The evaluation results are fed back to the Naive Bayesian model. If the energy-saving effect is good and no network anomalies occur, the model maintains the current probability distribution parameters. If users report issues such as slow network connection recovery, data is collected again and the model's probability distribution is updated to improve its accuracy and adaptability.

[0069] Scenario Example 3

[0070] AI model training data sources include:

[0071] (1) Traffic data, including optical link traffic: PON port upstream and downstream traffic between MFU and SFU (granularity: 100ms level); Wi-Fi traffic: classified by SSID / frequency band (2.4G / 5G) / terminal type (mobile phone / PC / IoT device); LAN port traffic: classified by port number, device MAC address, and protocol type (HTTP / VoIP / video stream).

[0072] (2) Device status data, including hardware parameters: CPU / GPU utilization (sampling rate 10Hz), memory usage, and temperature sensor value; network parameters: SFU optical module optical power (accuracy ±0.1dBm), bit error rate (BER), and SFUPON / LAN uplink registration status.

[0073] (3) User behavior data, including terminal connection period characteristics (such as the sleep period from 0:00 to 6:00 on weekdays / weekends); service type labels (real-time interaction type / background transmission type / high-priority service); terminal movement trajectory (based on RSSI fingerprint positioning algorithm).

[0074] The AI ​​model training and optimization process includes a training dataset consisting of 24 / 7 historical data from 1,000 FTTR systems; a model architecture using a two-layer bidirectional LSTM (256 units per layer) with an attention mechanism; and a training strategy using a hybrid loss function (MAE × 0.6 + RMSE × 0.4). Example data is shown in Table 1.

[0075] Table 1 Sample data table

[0076]

[0077] The expected energy-saving effects are: the traffic prediction error rate is significantly reduced (a significant improvement compared to the traditional ARIMA model), and the model inference delay is controlled within the threshold that meets real-time decision-making needs.

[0078] Scenario Example 4

[0079] After receiving the device data, the AI ​​model first uses the data feature extraction and preprocessing algorithm. Taking the principal component analysis (PCA) algorithm as an example, for multi-dimensional data such as optical port traffic, WIFI traffic, LAN port traffic, power consumption information, etc., the PCA algorithm decomposes the eigenvalues ​​of the data covariance matrix to find the main components in the data and achieve data dimensionality reduction. In actual operation, assuming that network data of 10 different dimensions are collected, the PCA algorithm calculates the covariance matrix of these data, and then solves the eigenvalues ​​and eigenvectors. According to the size of the eigenvalues, the first few main eigenvectors are selected, and the original 10-dimensional data is projected into the low-dimensional space spanned by these eigenvectors, for example, reducing its dimensionality to 3 dimensions. This not only retains the main information of the data, but also reduces the data dimension and reduces the complexity of subsequent processing. At the same time, a normalization algorithm is used to process the data, such as minimum-maximum normalization. For traffic data, assuming that its value range is between 0-1000Mbps, the minimum-maximum normalization formula is used. , X is the data that needs to be normalized, X min is the minimum value in the data set used for normalization, X max is the maximum value in the data set, Xnorm For normalized data, the data is mapped to the range of 0-1, making data of different magnitudes comparable and improving the performance of subsequent machine learning algorithms.

[0080] Scenario Example 5

[0081] The AI ​​model uses a decision tree algorithm to determine whether the current network status meets energy-saving conditions. This algorithm builds a tree structure to make decisions. First, using network traffic data as an example, the mean traffic flow is selected as the root node of the decision tree. A mean traffic flow threshold, such as 50 Mbps, is set. If the current mean network traffic flow is greater than 50 Mbps, the decision tree branches down one branch to determine other conditions, such as device load. If it is less than 50 Mbps, the decision tree branches down another branch to determine other related conditions. At each node, the algorithm selects the optimal feature for splitting based on metrics such as information gain or the Gini coefficient. For example, when determining device load, CPU usage is used as the next criterion, with a CPU usage threshold of 50%. If both the mean traffic flow and CPU usage are less than 50 Mbps, the decision tree continues to determine other conditions until it reaches a leaf node. Each leaf node corresponds to a different decision outcome, such as meeting or not meeting the energy-saving condition. If the energy-saving condition is ultimately determined to be met, the decision tree model further infers specific energy-saving measures based on device type and service priority. For slave gateways with fewer connected devices and very low traffic, the decision tree model outputs a decision to shut down some of its Wi-Fi bands (such as retaining only the 2.4G band) or reduce the transmission power of the Wi-Fi module.

[0082] Scenario Example 6

[0083] In one embodiment, the AI ​​model architecture includes:

[0084] Decision tree (CART algorithm): The average traffic volume (threshold 50 Mbps), CPU usage (threshold 40%), and number of connected devices (threshold 3) are used as split nodes. Leaf nodes correspond to energy-saving actions (such as "power reduction" and "module shutdown").

[0085] K-means clustering: Groups devices by power consumption (high / medium / low) and service type, prioritizing deep sleep strategies for high-power groups.

[0086] Real-time decision logic: In low-traffic scenarios, if the average traffic flow is <30 Mbps for 10 consecutive minutes and the CPU is <30%, the 5G band radio module is turned off, retaining the 2.4G band (power consumption is reduced from 12W to 6W). Device grouping strategy: For slave gateways clustered as "High Power Consumption - Video Streaming", the 4K video caching function is disabled during off-peak hours (power consumption is reduced by 25%).

[0087] Figure 5 This is a schematic diagram of the AI ​​model architecture according to an embodiment of the present application. Figure 5 The AI ​​model in the paper includes input parameters, prediction system, decision optimization and execution energy saving.

[0088] The input parameters include network topology parameters, traffic-related parameters, energy-saving state machine parameters, and service characteristic parameters.

[0089] The prediction system includes traffic prediction, business demand prediction, equipment load prediction, and user behavior prediction.

[0090] Decision optimization includes energy-saving strategy decisions, resource allocation decisions, state transition decisions, and parameter optimization decisions.

[0091] Energy saving includes equipment status adjustment, power consumption module management, resource dynamic configuration, status monitoring and feedback.

[0092] Scenario Example 7

[0093] In an FTTR system supporting AI models, the AI ​​model continuously analyzes real-time system data. If it detects that the number of Wi-Fi devices connected to a slave gateway remains low (less than two) for a period of time (e.g., 10 minutes), the average Wi-Fi traffic is below 15 Mbps, and the device load is low (e.g., CPU utilization less than 30% and memory usage less than 40%), the model determines that the current network state is suitable for energy-saving operation. For this slave gateway, the AI ​​model decides to disable its 5G frequency band, retaining only the 2.4 GHz band, and reduces the Wi-Fi module transmit power, for example, from 15 dBm to 10 dBm. If it detects a sudden increase in network traffic—for example, at 2 a.m., a slave gateway's Wi-Fi traffic rapidly increases from 10 Mbps to 50 Mbps within 5 minutes and continues to rise—the AI ​​model, based on learned traffic trends and device load, predicts that traffic growth is likely to continue. It immediately restores the slave gateway's 5G frequency band and appropriately increases the transmit power, for example, to 15 dBm, to ensure network performance. At the same time, the AI ​​model will make decisions based on business priorities. For high-priority businesses (such as VoIP calls), their network resources will be prioritized even during low-traffic periods, and energy-saving operations that may affect business quality will not be easily performed.

[0094] Traditional FTTR system energy-saving solutions without AI models typically employ fixed energy-saving rules. For example, a low-traffic period is set between midnight and 6 a.m. daily. During this period, the system uniformly reduces the transmit power of the Wi-Fi modules of all slave units (SFUs) to a fixed value, such as from 20dBm to 10dBm, and simultaneously shuts down some non-critical LAN ports. This approach fails to account for actual network traffic fluctuations and device load. For example, suppose a family member is participating in an online video conference, resulting in a brief spike in network traffic at 2 a.m. However, the system continues to operate according to the preset rules, without adjusting its energy-saving strategy based on traffic fluctuations. This can lead to degraded network performance and lag in the video conference. Furthermore, energy consumption is not flexibly adjusted based on actual conditions, resulting in unnecessary energy waste. Even if traffic increases, the device will not return to normal power until the preset low-traffic period ends.

[0095] Compared to traditional fixed-rule energy-saving methods, AI models can monitor changes in network status in real time and, when energy-saving opportunities are identified or network performance may be impacted, make immediate decisions and adjust the energy-saving status of the equipment. In the above example, when network traffic changes at 2 a.m., the AI ​​model can respond within seconds, while traditional methods would not adjust until the preset time has expired. This real-time nature enables the FTTR system to maximize energy-saving optimization while ensuring network performance, avoiding unnecessary energy consumption and improving energy efficiency.

[0096] Scenario Example 8

[0097] During the operation of the FTTR system, the FTTR main gateway adjusts the energy-saving status parameters in real time based on the device data through the AI ​​model, and makes the most suitable energy-saving decision for the current working scenario according to the specific working scenario. Figure 6 is a schematic diagram of an energy-saving state machine according to an embodiment of the present application, Figure 6 Different conditions are used to guide the state machine to switch between "normal mode," "energy-saving mode," and "deep sleep." These conditions include, but are not limited to, a traffic surge ratio exceeding A%, a low device load duration of B seconds, whether the device is currently at night, and whether a wake-up command has been received, where both A and B are greater than zero. Table 2 shows the initial energy-saving state parameters, including power consumption thresholds and response times.

[0098] Table 2 Initial energy-saving state parameters

[0099]

[0100] This embodiment integrates an AI model into the FTTR master gateway to achieve real-time, intelligent energy-saving control of the FTTR system, reducing system energy consumption while ensuring network performance and improving user experience. By deeply integrating AI technology with FTTR scenarios, through localized real-time decision-making, multi-dimensional data fusion, dynamic priority perception, and closed-loop optimization, it addresses the poor adaptability and low energy efficiency of traditional energy-saving solutions, providing an intelligent, high-precision energy-saving method for FTTR systems.

[0101] An embodiment of the present application further provides a computer-readable storage medium, in which a computer program is stored. The computer program is configured to execute the steps of any of the above method embodiments when run.

[0102] In an exemplary embodiment, the computer-readable storage medium may include, but is not limited to, various media that can store computer programs, such as a USB flash drive, a read-only memory (ROM), a random access memory (RAM), a mobile hard disk, a magnetic disk, or an optical disk.

[0103] An embodiment of the present application further provides an electronic device, comprising a memory and a processor, wherein the memory stores a computer program, and the processor is configured to run the computer program to execute the steps in any one of the above method embodiments.

[0104] In an exemplary embodiment, the electronic device may further include a transmission device and an input / output device, wherein the transmission device is connected to the processor, and the input / output device is connected to the processor.

[0105] For specific examples in this embodiment, reference may be made to the examples described in the above embodiments and exemplary implementation modes, and this embodiment will not be described in detail here.

[0106] Obviously, those skilled in the art should understand that the modules or steps of the present application described above can be implemented using a general-purpose computing device, they can be concentrated on a single computing device, or distributed across a network composed of multiple computing devices, they can be implemented using program code executable by the computing device, and thus, they can be stored in a storage device and executed by the computing device, and in some cases, the steps shown or described can be performed in a different order than herein, or they can be fabricated into separate integrated circuit modules, or multiple modules or steps can be fabricated into a single integrated circuit module for implementation. Thus, the present application is not limited to any specific combination of hardware and software.

[0107] The above description is merely a preferred embodiment of the present application and is not intended to limit the present application. Various modifications and variations are possible for those skilled in the art. Any modifications, equivalent substitutions, improvements, etc. made within the principles of the present application shall be included within the scope of protection of the present application.

Claims

1. A method for energy-saving control of a device, characterized in that: include: A fiber-to-the-room FTTR master gateway builds a device status database based on the collected device data, wherein the FTTR master gateway includes an artificial intelligence (AI) model; The FTTR master gateway analyzes the device data in the device status database by using the AI ​​model to determine an energy-saving control instruction, and sends the energy-saving control instruction to the corresponding device; Among them, the FTTR main gateway uses the AI ​​model to analyze the device data in the device status database to determine the energy-saving control instruction, including: the FTTR main gateway uses the AI ​​model to analyze the network status of the device in the device status database to determine whether the network status of the device meets the energy-saving conditions; if the judgment result is yes, the FTTR main gateway determines the energy-saving strategy of the device according to the device type and service priority, and generates the energy-saving control instruction according to the energy-saving strategy.

2. The method according to claim 1, characterized in that in, The device data includes at least one of the following: passive optical network (PON) port traffic data, wireless fidelity (WIFI) traffic data, local area network (LAN) port traffic data, device status data, power consumption information, and alarm information.

3. The method according to claim 1, characterized in that in, The AI ​​model is an AI model that has undergone feature dimensionality reduction and model compression.

4. The method according to claim 1, wherein The fiber-to-the-room FTTR master gateway builds a device status database based on the collected device data, including: The FTTR master gateway uses a data feature extraction and preprocessing algorithm to perform data cleaning and data normalization on the collected device data to obtain preprocessed device data; The FTTR main gateway correlates and integrates the pre-processed device data according to the network behavior characteristics of the device and the correlation of network resources to build the device status database.

5. The method according to claim 1, characterized in that in, The AI ​​model includes at least one of the following: decision tree, naive Bayes and long short-term memory network LSTM.

6. The method according to claim 1, characterized in that Before the fiber-to-the-room FTTR master gateway builds a device status database based on the collected device data, the method further includes: The FTTR master gateway collects device data uploaded by terminal devices connected to the FTTR master gateway and device data uploaded by FTTR slave gateway devices connected to the FTTR master gateway according to a preset period.

7. The method according to claim 1, characterized in that After sending the energy-saving control instruction to the corresponding device, it also includes: updating the parameters of the AI ​​model according to the operating data of the device.

8. The method according to claim 1, characterized in that The sending the energy-saving control instruction to the corresponding device includes: The energy-saving control instructions are classified and encapsulated according to the device type and communication protocol, and sent to the corresponding device through the fiber multimedia control interface FMCI protocol, the wireless multimedia control interface WMCI protocol, the fiber physical layer operation management and maintenance F-PLOAM protocol or a custom control protocol.

9. A computer-readable storage medium, characterized in that The computer-readable storage medium stores a computer program, wherein when the computer program is executed by a processor, the steps of the method described in any one of claims 1 to 8 are implemented.

10. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein: When the processor executes the computer program, the steps of the method according to any one of claims 1 to 8 are implemented.

11. A computer program product comprising a computer program, characterized in that When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 8 are implemented.

Images