Federal learning optimization system for low-power-consumption positioning of mobile terminal

Abstract

The invention relates to the technical field of power management, in particular to a federal learning optimization system for low-power-consumption positioning of a mobile terminal, which is used for acquiring multi-dimensional situation information of a wireless communication terminal, including real-time position coordinates, task progress, predetermined time planning and environment dynamic parameters of the wireless communication terminal. Based on federated learning, training information is obtained from a plurality of wireless communication terminals to construct an activity state prediction model. And generating an expected activity state of the terminal by using the activity state prediction model through the real-time position coordinate and the progress state of the wireless communication terminal. And comparing the expected active state with the current active state, and dynamically adjusting the power consumption configuration of the wireless communication terminal by applying a power consumption optimization decision model. Meanwhile, whether the actual moving track and state of the wireless communication terminal are abnormal or not is monitored; and if abnormity is detected, generating correction information, and adjusting the activity state prediction model and the power consumption optimization decision model to optimize the performance and the energy efficiency of the wireless communication terminal.

Description

Technical Field

[0001] The present invention relates to the field of power management technology, and in particular to a federated learning optimization system for low-power positioning on a mobile terminal. Background Art

[0002] Ultra-low power operation is a core technical challenge in Internet of Things (IoT) applications, particularly for mobile devices that require long-term deployment and struggle to frequently recharge or replace batteries. In complex environments like large industrial plants and construction sites, wireless communication terminals are often deployed on numerous mobile assets to provide status awareness for improved management efficiency and asset security. However, these scenarios place extremely stringent demands on the battery life of these wireless communication terminals, often requiring them to last for months or even years.

[0003] Obtaining location information is fundamental to many state-aware applications. However, the positioning process itself is often a major source of energy consumption for mobile devices. Existing positioning technologies, while striving for accuracy, often come at the cost of high power consumption, making it difficult to meet the ultra-long battery life requirements of these scenarios.

[0004] To reduce power consumption, existing technologies typically employ a periodic sleep / wake-up strategy, whereby the device remains in a low-power sleep state most of the time, waking up briefly only at preset intervals to perform sensing, positioning, and communication tasks. However, this fixed, non-intelligent periodic strategy is often suboptimal: when the asset's status is stable, frequent wake-ups still result in unnecessary energy waste; and when the asset's status requires close monitoring, fixed, long sleep intervals can lead to information acquisition delays.

[0005] Federated learning, a distributed machine learning technology, can train models using data from multiple devices while protecting data privacy. While federated learning can be used to optimize positioning accuracy or learn device behavior patterns, the federated learning process itself (local model training and communication for model updates) can also introduce significant energy consumption on mobile devices. Without optimization, frequent federated learning interactions can actually increase the energy burden on devices.

[0006] In recent years, accelerometers have been used to detect motion events and wake devices, or to monitor battery voltage to enter deep sleep. However, in the complex scenarios of industrial and construction site asset management, the parameters and events that need to be monitored are multi-dimensional. Furthermore, how to combine the distributed intelligence of federated learning to dynamically and collaboratively optimize trigger conditions and power mode transition strategies to achieve extremely low-power operation while also taking into account necessary information perception remains a problem that existing technologies have not effectively addressed.

[0007] To this end, a federated learning optimization system for low-power positioning on mobile terminals is proposed. Summary of the Invention

[0008] The present invention provides a federated learning optimization system for low-power positioning of mobile terminals. The system obtains multi-dimensional contextual information of wireless communication terminals, including the real-time location coordinates, task progress, scheduled time plan, and environmental dynamic parameters of the wireless communication terminals. Based on federated learning, training information is obtained from multiple wireless communication terminals to construct an activity state prediction model. The activity state prediction model is used to generate the expected activity state of the terminal based on the real-time location coordinates and progress status of the wireless communication terminal. The expected and current activity states are compared, and a power consumption optimization decision model is applied to dynamically adjust the power consumption configuration of the wireless communication terminal. At the same time, the actual operating trajectory and status of the wireless communication terminal are monitored for anomalies. If an anomaly is detected, correction information is generated to adjust the activity state prediction model and the power consumption optimization decision model to optimize the performance and energy efficiency of the wireless communication terminal.

[0009] To achieve the above object, the present invention provides the following technical solutions:

[0010] A federated learning optimization system for low-power positioning on mobile terminals, comprising:

[0011] A context information acquisition module is configured to acquire multi-dimensional context information related to the wireless communication terminal, wherein the multi-dimensional context information includes: the real-time location coordinates of the wireless communication terminal, the progress status of the tasks associated with the wireless communication terminal, the scheduled time plan, and the dynamic parameters of the environment in which the wireless communication terminal is located;

[0012] An activity state prediction module is configured to construct an activity state prediction model based on federated learning using training information obtained from multiple wireless communication terminals; and generate an expected activity state of the wireless communication terminal using the activity state prediction model based on the obtained real-time location coordinates and progress status;

[0013] a power consumption optimization control module for adjusting the power consumption operation configuration of the wireless communication terminal by applying a power consumption optimization decision model by comparing the expected activity state of the wireless communication with the current activity state;

[0014] The state verification and correction module is used to monitor whether there are any anomalies in the actual operation trajectory and state of the wireless communication terminal based on the scheduled time plan of the task, the dynamic parameters of the environment and the expected activity state. If an anomaly is detected, correction information is generated to adjust the activity state prediction model and the power consumption optimization decision model.

[0015] Preferably, the real-time location coordinates are obtained through wireless network positioning, including latitude and longitude values; the progress status of the task is obtained through a task management system, and the progress status includes the completion percentage, the current task stage identifier and a Boolean value; the scheduled time plan of the task is obtained through a task scheduling system, including the planned start time, end time and key time nodes of the task; the environmental dynamic parameters are collected through sensors, including physical values ​​and environmental status classifications.

[0016] Preferably, the activity state prediction model includes:

[0017] The input layer is used to receive and process the normalized real-time position coordinates and the progress status;

[0018] a recurrent neural network layer for processing time series dependencies and learning the activity pattern of the wireless communication terminal based on historical location and task progress;

[0019] The output layer receives the output from the last hidden layer and generates a probability distribution and a specific category label of the expected activity state of the wireless communication terminal in the next time step through an activation function; the expected activity states include stationary, low-speed movement, high-speed movement, and staying at the task location.

[0020] Preferably, the power consumption optimization decision model includes a state deviation calculation unit, a trigger condition adjustment unit, a power mode conversion unit and a strategy collaborative optimization unit;

[0021] a state deviation calculation unit, configured to calculate a deviation between a state of the wireless communication terminal perceived in real time and a dynamically expected state, and generate a degree of deviation of an operating state of the wireless communication terminal;

[0022] A trigger condition adjustment unit dynamically adjusts the trigger conditions for positioning and context awareness operations based on the degree of deviation, specifically adjusting the data collection frequency of the wireless network and sensors, the network communication frequency, and the sensitivity threshold for state change detection;

[0023] a power mode conversion unit, which comprehensively analyzes the deviation degree and the adjusted trigger condition, generates and executes a power mode conversion instruction, and switches between multiple preset power modes;

[0024] The strategy collaborative optimization unit uses a federated learning framework to comprehensively analyze the deviation degree, trigger condition adjustment efficiency and power mode switching data of multiple wireless communication terminals, and optimizes the trigger condition adjustment strategy and power mode switching logic.

[0025] Preferably, the deviation degree generation process includes:

[0026] The latest positioning data is obtained through the real-time sensor of the wireless communication terminal to measure the current actual activity state; the expected activity state generated by the activity state prediction model is compared with the current actual activity state inferred in real time to generate a degree of deviation; the comparison method is: if the activity state is a predefined discrete category, a category matching comparison is performed; if the activity state is a continuous value, a difference measure between the two is calculated;

[0027] The degree of deviation is a Boolean value indicating the consistency between the two, a numerical value quantifying the degree of deviation, and a signal identifying the specific expected state and current state category.

[0028] Preferably, the specific process of determining whether there is an abnormality in the actual operation trajectory and state of the wireless communication terminal includes:

[0029] comparing the actual position coordinates of the wireless communication terminal acquired over a period of time with an expected trajectory deduced based on the expected activity state and the predetermined time plan of the task, and calculating a trajectory deviation;

[0030] Detecting the degree of deviation to determine whether there is a persistent or / and significant state mismatch; detecting whether the current location and activity state of the wireless communication terminal are consistent with the scheduled time plan of the task;

[0031] evaluating whether the activity state of the wireless communication terminal is logically inconsistent with the environmental dynamic parameter;

[0032] If any result exceeds the preset threshold and / or violates the preset logical rules, it is determined that an anomaly exists.

[0033] Preferably, the specific process of adjusting the activity state prediction model and the power consumption optimization decision model includes:

[0034] When an anomaly is detected, the state verification and correction module generates correction information, including the anomaly type, occurrence time, location, deviation degree, and relevant context information;

[0035] Adjusting the activity state prediction model: using the abnormal event data with the correct label as a new training sample for locally performing incremental training on the activity state prediction model on the wireless communication terminal;

[0036] Adjusting the power consumption optimization decision model: adjusting the parameters in the trigger condition adjustment unit and the switching logic of the power mode conversion unit according to the abnormality type in the correction information.

[0037] Compared with the prior art, the present invention has the following beneficial effects:

[0038] 1. The present invention integrates multi-dimensional contextual information such as the real-time positioning, task progress, and environmental dynamics of wireless communication terminals, constructs an activity state prediction model through federated learning, and uses a recurrent neural network to mine historical positioning data and task evolution patterns to predict states such as stillness, slow movement, high-speed sprinting, and task pauses, and formulates a low-power operation strategy. By intelligently adjusting the wireless network and sensor acquisition frequency, a balance is achieved between power consumption and positioning accuracy, effectively reducing energy consumption, extending device life, and adapting to changes in different scenarios. It excels in power consumption management and communication stability, enhancing the system's intelligence and adaptability.

[0039] 2. The present invention identifies data deviations and logical contradictions by integrating task scheduled time planning, environmental parameters and real-time terminal trajectories. The predicted state is compared with the actual state in real time, and through quantitative deviation calculation and category matching comparison, it is determined whether there are continuous anomalies or sudden imbalances during the terminal operation, which helps to quickly discover potential problems. When an operational anomaly is detected, the system generates detailed correction information, covering the anomaly type, occurrence time, location information and deviation value, and uses local data incremental training to accurately fine-tune the model, effectively realizing fault self-recovery. It not only improves the accuracy of anomaly identification, but also optimizes the data transmission and processing process, ensures the overall security and stability of the system, and enhances the collaborative decision-making capabilities between terminals.

[0040] 3. The present invention is built on the basis of the federated learning framework to realize multi-terminal data sharing and distributed model training, breaking through the bottleneck of traditional single-machine training and promoting the collaborative optimization of each node. Through cross-terminal information fusion, the activity state prediction and power consumption decision model can be updated in real time to form a dynamic feedback closed loop to ensure that each terminal can adaptively adjust the operating mode according to the actual environment. By making full use of the complementary advantages of data between terminals and adjusting the power consumption mode and trigger conditions in real time through the strategy collaborative optimization unit, not only the overall energy efficiency is improved, but also the positioning response speed and accuracy are significantly improved. Effectively reduce management costs and improve the overall operation efficiency and security level of the network. BRIEF DESCRIPTION OF THE DRAWINGS

[0041] Figure 1 A schematic diagram of the structure of a federated learning optimization system for low-power positioning on a mobile terminal provided by the present invention;

[0042] Figure 2 A schematic diagram of the power consumption optimization decision model structure provided by an embodiment of the present invention;

[0043] Figure 3 A schematic diagram of the low-power positioning process of a mobile terminal provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0044] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention.

[0045] Example 1

[0046] See also Figure 1 The present invention provides a federated learning optimization system for low-power positioning on a mobile terminal. The technical solution is as follows:

[0047] A federated learning optimization system for low-power positioning on mobile terminals, comprising:

[0048] A context information acquisition module is configured to acquire multi-dimensional context information related to the wireless communication terminal, wherein the multi-dimensional context information includes: the real-time location coordinates of the wireless communication terminal, the progress status of the tasks associated with the wireless communication terminal, the scheduled time plan, and the dynamic parameters of the environment in which the wireless communication terminal is located;

[0049] Furthermore, the real-time location coordinates are obtained through wireless network positioning, including longitude and latitude values; the progress status of the task is obtained through the task management system, and the progress status includes the completion percentage, the current task stage identifier and the Boolean value; the scheduled time plan of the task is obtained through the task scheduling system, including the planned start time, end time and key time nodes of the task; the environmental dynamic parameters are collected through sensors, including physical values ​​and environmental status classifications.

[0050] In this embodiment, by collecting multi-dimensional contextual information from wireless communication terminals, comprehensive, real-time monitoring of terminal status is achieved, providing solid data support for subsequent optimization decisions. Wireless network positioning technology is used to obtain real-time location coordinates, including latitude and longitude values, to ensure accurate tracking of terminal locations. At the same time, through the task management system and task scheduling system, real-time collection of progress status, completion percentage, task stage identifiers, and scheduled start, end, and key time nodes enables a comprehensive understanding of task execution. This improves data integrity and accuracy, and provides high-quality input data for subsequent modules such as activity state prediction, power consumption optimization, and abnormal state detection, thereby achieving intelligent, precise system operation and low power consumption goals. It also improves the overall system's response speed and adaptability, providing technical support for the efficient and stable operation of wireless communication terminals in various complex scenarios.

[0051] An activity state prediction module is configured to construct an activity state prediction model based on federated learning using training information obtained from multiple wireless communication terminals; and generate an expected activity state of the wireless communication terminal using the activity state prediction model based on the obtained real-time location coordinates and progress status;

[0052] Furthermore, the activity state prediction model includes:

[0053] The input layer is used to receive and process the normalized real-time position coordinates and the progress status;

[0054] a recurrent neural network layer for processing time series dependencies and learning the activity pattern of the wireless communication terminal based on historical location and task progress;

[0055] The output layer receives the output from the last hidden layer and generates a probability distribution and a specific category label of the expected activity state of the wireless communication terminal in the next time step through an activation function; the expected activity states include stationary, low-speed movement, high-speed movement, and staying at the task location.

[0056] In this embodiment, by introducing federated learning and deep recurrent neural networks, multi-terminal collaborative training is enabled to capture the historical activity patterns of wireless communication terminals, achieving accurate real-time status prediction. Using normalized real-time location coordinates and progress status as input, the input layer first performs effective data preprocessing. Then, with the help of a recurrent neural network layer, the dependencies between time series are deeply explored to accurately learn the movement patterns of terminals at different task stages. The output layer, based on an activation function, outputs the state probability distribution and specific category labels for the next time step, covering various activity states such as stationary, slow-moving, high-speed moving, and stopping at the task location. This not only improves prediction accuracy and response speed, but also provides reliable data support for subsequent power consumption optimization and intelligent scheduling. The activity state prediction model provided by this invention demonstrates improved effectiveness in terms of prediction accuracy, state transition sensitivity, and inter-device collaboration compared to rule-based prediction methods and single-device machine learning models. Rule-based prediction methods include decision tree models and linear regression models, as detailed in Table 1.

[0057] Table 1 Performance comparison of activity state prediction model and existing algorithms

[0058]

[0059] Prediction accuracy is primarily measured by comparing model predictions with actual state labels on a test dataset. State transition detection sensitivity is measured by measuring the time it takes for the system to successfully detect a change in the actual terminal state. Dependence on historical data is assessed through model structure analysis and experimental verification. Multi-device collaboration is evaluated based on the implementation and test results of a federated learning framework. Adaptability to environmental changes is assessed through testing designed for environmental change scenarios.

[0060] a power consumption optimization control module for adjusting the power consumption operation configuration of the wireless communication terminal by applying a power consumption optimization decision model by comparing the expected activity state of the wireless communication with the current activity state;

[0061] Furthermore, the power consumption optimization decision model includes a state deviation calculation unit, a trigger condition adjustment unit, a power mode conversion unit and a strategy collaborative optimization unit, see Figure 2 ;

[0062] a state deviation calculation unit, configured to calculate a deviation between a state of the wireless communication terminal perceived in real time and a dynamically expected state, and generate a degree of deviation of an operating state of the wireless communication terminal;

[0063] A trigger condition adjustment unit dynamically adjusts the trigger conditions for positioning and context awareness operations based on the degree of deviation, specifically adjusting the data collection frequency of the wireless network and sensors, the network communication frequency, and the sensitivity threshold for state change detection;

[0064] a power mode switching unit, which comprehensively analyzes the deviation degree and the adjusted trigger condition, generates and executes a power mode switching instruction, and switches between multiple preset power modes (e.g., deep sleep mode, light sleep mode, low power awareness mode, and high-performance active mode);

[0065] The strategy collaborative optimization unit uses a federated learning framework to comprehensively analyze the deviation degree, trigger condition adjustment efficiency and power mode switching data of multiple wireless communication terminals, and optimizes the trigger condition adjustment strategy and power mode switching logic.

[0066] In this embodiment, intelligent energy management is achieved through a power consumption optimization decision model, using real-time comparisons between the expected and current activity states of wireless communication terminals. First, a state deviation calculation unit accurately measures the difference between the terminal's actual state and its dynamically expected state, providing precise data support for subsequent energy consumption control. Next, a trigger condition adjustment unit dynamically adjusts the data collection frequency, network communication frequency, and state change detection sensitivity of the wireless network and sensors based on the degree of deviation, enabling flexible response in different environments. Simultaneously, a power mode conversion unit comprehensively evaluates the degree of deviation and the adjusted trigger conditions to intelligently switch between multiple preset power modes, including deep sleep, light hibernation, low-power sensing, and high-performance active, ensuring that the terminal achieves optimal energy consumption in different scenarios. Finally, a policy collaborative optimization unit utilizes a federated learning framework, integrating operational data and adjustment feedback from multiple terminals to continuously optimize the trigger condition adjustment strategy and power mode switching logic, thereby achieving network-wide collaborative optimization. This not only effectively reduces device energy consumption and extends terminal battery life, but also significantly improves the system's adaptability and stability.

[0067] Furthermore, the deviation degree generation process includes:

[0068] The latest positioning data is obtained through the real-time sensor of the wireless communication terminal to measure the current actual activity state; the expected activity state generated by the activity state prediction model is compared with the current actual activity state inferred in real time to generate a degree of deviation; the comparison method is: if the activity state is a predefined discrete category, a category matching comparison is performed; if the activity state is a continuous value, a difference measure between the two is calculated;

[0069] The degree of deviation is a Boolean value indicating the consistency between the two, a numerical value quantifying the degree of deviation, and a signal identifying the specific expected state and current state category.

[0070] In this embodiment, the deviation degree generation process uses the real-time sensor data of the wireless communication terminal to accurately measure the current actual activity state and compare it with the expected state generated by the activity state prediction model, thereby presenting the difference between the two in multiple dimensions in the form of Boolean consistency, quantitative deviation, and specific state category signals. This comparison method, whether it is a matching comparison of discrete categories or a difference measurement of continuous values, can quickly capture subtle deviations between states, enabling the system to have higher accuracy and response speed in detecting anomalies, adjusting power consumption, and optimizing operating configurations. By identifying potential inconsistencies and abnormal situations in advance, this mechanism not only improves the reliability of intelligent control of wireless communication terminals, but also effectively reduces energy waste and the risk of system failure.

[0071] The state verification and correction module is used to monitor whether there are any anomalies in the actual operation trajectory and state of the wireless communication terminal based on the scheduled time plan of the task, the dynamic parameters of the environment and the expected activity state. If an anomaly is detected, correction information is generated to adjust the activity state prediction model and the power consumption optimization decision model.

[0072] Furthermore, the specific process of determining whether there is an abnormality in the actual operation trajectory and state of the wireless communication terminal includes:

[0073] comparing the actual position coordinates of the wireless communication terminal acquired over a period of time with an expected trajectory deduced based on the expected activity state and the predetermined time plan of the task, and calculating a trajectory deviation;

[0074] Detecting the degree of deviation to determine whether there is a persistent or / and significant state mismatch; detecting whether the current location and activity state of the wireless communication terminal are consistent with the scheduled time plan of the task;

[0075] evaluating whether the activity state of the wireless communication terminal is logically inconsistent with the environmental dynamic parameter;

[0076] If any result exceeds the preset threshold and / or violates the preset logical rules, it is determined that an anomaly exists.

[0077] In this embodiment, by comparing the actual operating trajectory and status of the wireless communication terminal in real time, the system is effectively ensured to operate continuously and efficiently in a dynamic environment. The actual location data obtained by the terminal is compared with the expected trajectory deduced based on the expected activity state and the scheduled time plan of the task, the trajectory deviation is accurately calculated, and the degree of deviation is detected to promptly determine whether there is a persistent or significant state mismatch. At the same time, the module further detects whether the current activity state is consistent with the scheduled task time plan and the dynamic parameters of the environment, and evaluates the logical consistency between the operating state and the environment. If the detection result exceeds the preset threshold or violates the logical rules, correction information is quickly generated to dynamically adjust the activity state prediction model and the power consumption optimization decision model, thereby achieving adaptive optimization. Not only does it improve the system's sensitivity and response speed to abnormal situations, but it also significantly reduces energy waste and safety hazards caused by state deviations.

[0078] Furthermore, the specific process of adjusting the activity state prediction model and the power consumption optimization decision model includes:

[0079] When an anomaly is detected, the state verification and correction module generates correction information, including the anomaly type, occurrence time, location, deviation degree, and relevant context information;

[0080] Adjusting the activity state prediction model: using the abnormal event data with the correct label as a new training sample for locally performing incremental training on the activity state prediction model on the wireless communication terminal;

[0081] Adjusting the power consumption optimization decision model: adjusting the parameters in the trigger condition adjustment unit and the switching logic of the power mode conversion unit according to the abnormality type in the correction information.

[0082] In this embodiment, by introducing a dynamic adaptive mechanism, real-time updating and optimization of the activity state prediction model and the power consumption optimization decision model are achieved. When an anomaly is detected, the state verification and correction module can quickly generate correction information containing the anomaly type, occurrence time, location information, degree of deviation and related context information, providing an accurate basis for subsequent model adjustments. By using abnormal event data with correct labels for local incremental training or fine-tuning, the activity state prediction model can continuously adapt to environmental changes and significantly improve the prediction accuracy; at the same time, the trigger condition parameters and power mode conversion logic are dynamically adjusted according to the anomaly type, so that the power consumption optimization decision model can flexibly switch between multiple preset power modes, thereby effectively reducing energy consumption and system failure risks. It enhances the rapid response and self-healing capabilities of wireless communication terminals to abnormal states, and greatly improves the stability, energy efficiency and operational reliability of the system.

[0083] This invention integrates multi-dimensional contextual data collection, federated learning, and deep recurrent neural network technology to achieve real-time monitoring and accurate prediction of wireless terminal status. The system automatically calculates state deviations by comparing expected activity status with actual status, and dynamically adjusts data collection frequency, power mode, and trigger conditions to promptly identify and correct anomalies, thereby optimizing power consumption configuration. This mechanism effectively reduces energy consumption, extends battery life, and significantly improves system response speed and adaptability, ensuring stable and efficient operation of terminals in complex environments. For details, please refer to Figure 3 . For different activity scenarios, the system of the present invention shows significant power consumption advantages compared with the traditional periodic wake-up solution. As shown in Table 2, in the static state, the average power consumption of the system is only 15.2 mW, which is 80.6% lower than the 78.4 mW of the traditional solution, and the battery life is extended by 5.16 times. Even in high-speed mobile scenarios with large resource consumption, the system can still achieve a 9.5% power consumption reduction. In a typical mixed activity mode, the average power consumption of the system is 56.4 mW, which is 66.5% lower than the 168.3 mW of the traditional solution, and the battery life is extended by nearly 3 times. These data fully demonstrate the significant innovative effect of the present invention in power consumption optimization, and provide possibilities for the long-term deployment of mobile devices.

[0084] Table 2 Comparison of power consumption optimization effects

[0085]

[0086] Example 2

[0087] This embodiment introduces a complete implementation of a federated learning optimization system for low-power positioning on mobile terminals. By integrating multi-dimensional contextual information, federated learning, and deep neural network technology, the system achieves intelligent power management and precise positioning of wireless communication terminals. Specifically, it includes:

[0088] A context information acquisition module is configured to acquire multi-dimensional context information related to the wireless communication terminal, wherein the multi-dimensional context information includes: the real-time location coordinates of the wireless communication terminal, the progress status of the tasks associated with the wireless communication terminal, the scheduled time plan, and the dynamic parameters of the environment in which the wireless communication terminal is located;

[0089] Furthermore, the real-time location coordinates, including longitude and latitude values, are obtained through wireless network positioning. Specifically, these coordinates are obtained using technologies such as GPS, WiFi positioning, or cellular network triangulation, with an accuracy of ±5 meters. Coordinates are expressed in the WGS84 coordinate system and formatted as decimal degrees, such as 120.1234° longitude and 30.5678° latitude. The data update frequency is dynamically adjusted based on the terminal's activity state: 10 minutes per update when stationary, 1 minute per update when moving at low speeds, and 10 seconds per update when moving at high speeds.

[0090] The progress status of the task is obtained through the task management system, and the progress status includes the completion percentage, the current task stage identifier and a Boolean value; specifically, it includes the completion percentage (expressed as a value between 0-100, accurate to one decimal place), the current task stage identifier (using an integer ID to identify the current task stage) and a Boolean value (indicating whether the task is in an active state).

[0091] The scheduled time for each task is obtained through the task scheduling system. This includes the task's planned start and end times, as well as key time points. These points are stored in a timestamp array format, with each time point accompanied by identification information. The system calculates the proximity D between the current time and the key time point using the formula: D = |Current time - Nearest key time point| / Total task duration. If D < 0.05, meaning the distance to the key time point is less than 5% of the total duration, the system increases the frequency of status monitoring.

[0092] Dynamic environmental parameters are collected through sensors, including physical values ​​such as temperature, humidity, light intensity, and noise level; as well as environmental status classifications such as indoor / outdoor, quiet / noisy, and bright / dark. The sensor sampling frequency is 5 minutes under normal conditions, increasing to 30 seconds when abrupt environmental changes are detected. Each physical parameter is normalized to the [0, 1] range after being normalized to the minimum and maximum values.

[0093] An activity state prediction module, configured to construct an activity state prediction model based on federated learning using training information obtained from multiple wireless communication terminals;

[0094] The federated learning framework includes local model training, parameter aggregation, and model update distribution. Local model training involves each terminal training a model based on local data and calculating parameter gradients. Parameter aggregation involves a central server collecting model parameter updates from each terminal and calculating a weighted average. Model update distribution involves distributing the updated global model parameters back to each terminal. Based on the acquired real-time location coordinates and progress status, the activity state prediction model is used to generate the expected activity state of the wireless communication terminal. The optimizer uses Adam with an initial learning rate of 0.001 and a decay rate of 0.95. The batch size is 32, the federated learning aggregation period is 48 hours, local training is 10 rounds per session, and the global model is updated once a week.

[0095] Furthermore, the activity state prediction model includes:

[0096] The input layer is used to receive and process the normalized real-time position coordinates and the progress status;

[0097] The recurrent neural network layer processes time series dependencies and learns the activity patterns of the wireless communication terminal based on its historical location and task progress. It employs an LSTM architecture, comprising components such as a forget gate, input gate, output gate, candidate memory cells, memory cell updates, and hidden state calculation. The network structure comprises two LSTM layers, each with 64 neurons. The time window length is 24, meaning that historical data from the previous 24 time points is considered. A dropout rate of 0.2 is used during training to prevent overfitting.

[0098] The output layer receives the output from the last hidden layer and, through an activation function, generates a probability distribution and specific category label for the wireless communication terminal's expected activity state at the next time step. The expected activity states include stationary, slow movement, high-speed movement, and dwelling at the task location. State determination criteria include: stationary (speed v < 0.5 m / min), slow movement (0.5 m / min ≤ v < 30 m / min), high-speed movement (v ≥ 30 m / min), and dwelling at the task location (distance d between the location coordinate and the task location < 50 meters and duration t > 3 minutes).

[0099] a power consumption optimization control module for adjusting the power consumption operation configuration of the wireless communication terminal by applying a power consumption optimization decision model by comparing the expected activity state of the wireless communication with the current activity state;

[0100] Furthermore, the power consumption optimization decision model includes a state deviation calculation unit, a trigger condition adjustment unit, a power mode conversion unit and a strategy collaborative optimization unit;

[0101] a state deviation calculation unit, configured to calculate a deviation between a state of the wireless communication terminal perceived in real time and a dynamically expected state, and generate a degree of deviation of an operating state of the wireless communication terminal;

[0102] The trigger condition adjustment unit dynamically adjusts the operation trigger conditions related to positioning and context awareness based on the degree of deviation, specifically adjusting the data collection frequency of the wireless network and sensors, the network communication frequency, and the sensitivity threshold of state change detection; when the degree of deviation exceeds the preset threshold for more than 3 cycles, an emergency state is triggered and all frequencies are increased to the highest level; when the system power is less than 20%, the collection frequency is reduced to the lowest level regardless of the degree of deviation.

[0103] The power mode conversion unit comprehensively analyzes the deviation degree and the adjusted trigger conditions, generates and executes the power mode conversion instruction, and switches between multiple preset power modes; the preset power modes include deep sleep mode (power consumption of about 10-20mW), light sleep mode (power consumption of about 50-80mW), low power awareness mode (power consumption of about 100-200mW) and high performance active mode (power consumption of about 300-500mW). The mode switching decision is based on the comprehensive score :

[0104] ;

[0105] in, 、 and are the weights of deviation degree, task urgency and battery power respectively, is the degree of deviation, For the task urgency, is the battery level;

[0106] The strategy collaborative optimization unit uses a federated learning framework to comprehensively analyze the deviation degree, trigger condition adjustment efficiency and power mode switching data of multiple wireless communication terminals, and optimizes the trigger condition adjustment strategy and power mode switching logic.

[0107] Furthermore, the deviation degree generation process includes:

[0108] The latest positioning data is obtained through the real-time sensor of the wireless communication terminal to measure the current actual activity state; the expected activity state generated by the activity state prediction model is compared with the current actual activity state inferred in real time to generate a degree of deviation; the comparison method is: if the activity state is a predefined discrete category, a category matching comparison is performed; if the activity state is a continuous value, a difference measure between the two is calculated;

[0109] The degree of deviation is a Boolean value of the consistency between the two, a numerical value that quantifies the degree of deviation, and a signal that identifies the specific expected state and current state category. It also performs a deviation persistence assessment, distinguishes short-term deviation, medium-term deviation and long-term deviation, and calculates a weighted deviation.

[0110] The state verification and correction module is used to monitor whether there are any anomalies in the actual operation trajectory and state of the wireless communication terminal based on the scheduled time plan of the task, the dynamic parameters of the environment and the expected activity state. If an anomaly is detected, correction information is generated to adjust the activity state prediction model and the power consumption optimization decision model.

[0111] Furthermore, the specific process of determining whether there is an abnormality in the actual operation trajectory and state of the wireless communication terminal includes:

[0112] The actual position coordinates obtained by the wireless communication terminal within a period of time are compared with the expected trajectory deduced based on the expected activity state and the scheduled time plan of the task, and the trajectory deviation is calculated; the trajectory deviation calculation collects the actual position coordinates within a period of time, compares them with the expected trajectory deduced based on the expected activity state and the scheduled time plan of the task, and calculates the point-to-point distance deviation, average trajectory deviation, maximum trajectory deviation and standard deviation.

[0113] The degree of deviation is tested to determine whether there is a persistent or significant state mismatch; when the calculated continuous state mismatch period is greater than 20 minutes or the state mismatch ratio is greater than 0.4, it is determined to be a persistent or significant mismatch. The current location and activity status of the wireless communication terminal are tested to see if they are consistent with the scheduled time plan of the task; the task time plan consistency test extracts the key time nodes of the current task, checks the consistency between the actual state and the expected state near each key time point, and calculates the task time plan consistency score.

[0114] Environmental logic consistency assessment extracts the current dynamic parameters of the environment, defines a set of logical rules linking activity states and environmental parameters, evaluates rule violations, and calculates an environmental consistency score. Comprehensive anomaly determination sets trajectory deviation thresholds, state mismatch thresholds, mission time plan consistency thresholds, and environmental consistency thresholds, integrating determination criteria and quantifying anomaly severity.

[0115] If any result exceeds the preset threshold and / or violates the preset logical rules, it is determined that an anomaly exists.

[0116] Furthermore, the specific process of adjusting the activity state prediction model and the power consumption optimization decision model includes:

[0117] When an anomaly is detected, the state verification and correction module generates correction information including the anomaly type, occurrence time, location, degree of deviation, and relevant context information;

[0118] Adjust the activity status prediction model:

[0119] Using abnormal event data with correct labels (actual status) as new training samples for locally performing incremental training on the activity status prediction model on the wireless communication terminal;

[0120] Through the federated learning mechanism, the abnormal information or the corrected local model parameter update is contributed to the global model aggregation process to improve the robustness and accuracy of the overall model;

[0121] Adjust the power consumption optimization decision model:

[0122] Adjusting parameters in the power consumption policy trigger condition adjustment unit (e.g., deviation threshold, adjustment amplitude) or the switching logic of the power mode intelligent switching unit (e.g., increasing the state verification frequency, adopting a more conservative power consumption mode) based on the abnormal pattern reflected by the correction information (e.g., frequent errors in the prediction model leading to failure of the power consumption policy);

[0123] The effect of the adjustment strategy is fed back to the federated collaborative optimization unit to optimize the global power consumption optimization decision strategy under the federated learning framework.

[0124] Table 3 Comparison of overall system performance and resource usage

[0125]

[0126] The proposed system achieves an excellent balance between performance and resource utilization. As shown in Table 3, the system's average positioning accuracy reaches ±8.6 meters, significantly exceeding the ±18.4 meters of the traditional low-power system A. Simultaneously, the average battery life reaches 84 days, 2.6 times that of the traditional system and 16.8 times that of the high-precision positioning system B (5 days). While the proposed system's CPU utilization (4.8%) and memory usage (8.4MB) are slightly higher than those of the traditional low-power system, they are significantly lower than the resource consumption of the high-precision positioning system. In particular, regarding data storage requirements, the proposed system requires only 12.6MB per month, while the high-precision system requires 268.4MB. Furthermore, the proposed system employs an adaptive network communication frequency strategy, averaging 5.2 communications per hour, dynamically adjusting according to actual conditions, avoiding the limitations of fixed frequency strategies. Furthermore, the system's response latency is only 740ms, significantly superior to traditional low-power systems. The traditional low-power system A utilizes a low-power ESP32C3 microcontroller, while the high-precision positioning system includes Bluetooth-assisted positioning, UWB, and an inertial measurement unit.

[0127] While embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions, and variations may be made to these embodiments without departing from the principles and spirit of the invention, and that the scope of the invention is defined by the appended claims and their equivalents.

Claims

1. A federated learning optimization system for low-power positioning on mobile terminals, characterized by: include: A context information acquisition module is configured to acquire multi-dimensional context information related to the wireless communication terminal, wherein the multi-dimensional context information includes: the real-time location coordinates of the wireless communication terminal, the progress status of the tasks associated with the wireless communication terminal, the scheduled time plan, and the dynamic parameters of the environment in which the wireless communication terminal is located; An activity state prediction module is configured to construct an activity state prediction model based on federated learning using training information obtained from multiple wireless communication terminals; and generate an expected activity state of the wireless communication terminal using the activity state prediction model based on the obtained real-time location coordinates and progress status; a power consumption optimization control module, configured to adjust the power consumption operation configuration of the wireless communication terminal by applying a power consumption optimization decision model by comparing the expected activity state of the wireless communication with the current activity state; The state verification and correction module is used to monitor whether there are any anomalies in the actual operation trajectory and state of the wireless communication terminal based on the scheduled time plan of the task, the dynamic parameters of the environment and the expected activity state. If an anomaly is detected, correction information is generated to adjust the activity state prediction model and the power consumption optimization decision model.

2. The federated learning optimization system for low-power mobile terminal positioning according to claim 1, characterized in that: The real-time location coordinates are obtained through wireless network positioning, including latitude and longitude values; the progress status of the task is obtained through the task management system, and the progress status includes the completion percentage, the current task stage identifier and a Boolean value; the scheduled time plan of the task is obtained through the task scheduling system, including the planned start time, end time and key time nodes of the task; the environmental dynamic parameters are collected through sensors, including physical values ​​and environmental status classifications.

3. The federated learning optimization system for low-power mobile terminal positioning according to claim 1, characterized in that: The activity state prediction model includes: The input layer is used to receive and process the normalized real-time position coordinates and the progress status; a recurrent neural network layer for processing time series dependencies and learning the activity pattern of the wireless communication terminal based on historical location and task progress; The output layer receives the output from the last hidden layer and generates a probability distribution and a specific category label of the expected activity state of the wireless communication terminal in the next time step through an activation function; the expected activity states include stationary, low-speed movement, high-speed movement, and staying at the task location.

4. The federated learning optimization system for low-power mobile terminal positioning according to claim 1, characterized in that: The power consumption optimization decision model includes a state deviation calculation unit, a trigger condition adjustment unit, a power mode conversion unit and a strategy collaborative optimization unit; a state deviation calculation unit, configured to calculate a deviation between a state of the wireless communication terminal perceived in real time and a dynamically expected state, and generate a degree of deviation of an operating state of the wireless communication terminal; A trigger condition adjustment unit dynamically adjusts the trigger conditions for positioning and context awareness operations based on the degree of deviation, specifically adjusting the data collection frequency of the wireless network and sensors, the network communication frequency, and the sensitivity threshold for state change detection; a power mode conversion unit, which comprehensively analyzes the deviation degree and the adjusted trigger condition, generates and executes a power mode conversion instruction, and switches between multiple preset power modes; The strategy collaborative optimization unit uses a federated learning framework to comprehensively analyze the deviation degree, trigger condition adjustment efficiency and power mode switching data of multiple wireless communication terminals, and optimizes the trigger condition adjustment strategy and power mode switching logic.

5. The federated learning optimization system for low-power mobile terminal positioning according to claim 4, characterized in that: The deviation degree generation process includes: The latest positioning data is obtained through the real-time sensor of the wireless communication terminal to measure the current actual activity state; the expected activity state generated by the activity state prediction model is compared with the current actual activity state inferred in real time to generate a degree of deviation; the comparison method is: if the activity state is a predefined discrete category, a category matching comparison is performed; if the activity state is a continuous value, a difference measure between the two is calculated; The degree of deviation is a Boolean value indicating the consistency between the two, a numerical value quantifying the degree of deviation, and a signal identifying the specific expected state and current state category.

6. The federated learning optimization system for low-power mobile terminal positioning according to claim 1, characterized in that: The specific process of determining whether the actual operating trajectory and status of the wireless communication terminal are abnormal includes: comparing the actual position coordinates of the wireless communication terminal acquired over a period of time with an expected trajectory deduced based on the expected activity state and the predetermined time plan of the task, and calculating a trajectory deviation; Detecting the degree of deviation to determine whether there is a persistent or / and significant state mismatch; detecting whether the current position and state of the wireless communication terminal are consistent with the scheduled time plan of the task; evaluating whether the activity state of the wireless communication terminal is logically inconsistent with the environmental dynamic parameter; If any result exceeds the preset threshold and / or violates the preset logical rules, it is determined that an anomaly exists.

7. The federated learning optimization system for low-power mobile terminal positioning according to claim 1, characterized in that: The specific process of adjusting the activity state prediction model and the power consumption optimization decision model includes: When an anomaly is detected, the state verification and correction module generates correction information, including the anomaly type, occurrence time, location, deviation degree, and relevant context information; Adjusting the activity state prediction model: using the abnormal event data with the correct label as a new training sample for locally performing incremental training on the activity state prediction model on the wireless communication terminal; Adjusting the power consumption optimization decision model: adjusting the parameters in the trigger condition adjustment unit and the switching logic of the power mode conversion unit according to the abnormality type in the correction information.

Images