Low-power-consumption data transmission optimization system and method for smart watch

Abstract

The invention relates to the technical field of data transmission optimization, in particular to a low-power-consumption data transmission optimization system and method for a smart watch. According to the method, the data transmission strategy is dynamically adjusted by combining the user behavior state, the data content characteristics and the system operation parameters. The system firstly senses the wearing state, interaction frequency and equipment resource information of a user, evaluates the priority of to-be-transmitted data, predicts the short-time attention probability of the user based on the behavior trend, and calculates the cognitive perception index of the data. And further comprehensively considering energy consumption, cognitive risk and resource occupation through a scheduling cost function, selecting an optimal transmission mode, and executing instant uploading, delayed caching, compression processing or communication protocol switching operation at an equipment end. According to the scheme, the data redundancy transmission frequency can be effectively reduced, the timeliness of key data is ensured, the overall energy efficiency of the system is remarkably improved, and the method is suitable for the smart watches in various dynamic use scenes.

Description

Technical Field

[0001] The present invention relates to the technical field of data transmission optimization, and in particular to a low-power data transmission optimization system and method for smart watches. Background Art

[0002] The widespread adoption of wearable devices like smartwatches for health monitoring, activity tracking, and message synchronization has increased the demand for real-time data and communication energy efficiency. Given limited hardware resources, achieving low-power transmission of high-frequency data has become a key challenge in improving user experience and device battery life.

[0003] Existing technologies often use fixed-priority scheduling or transmission strategies based on a single energy consumption metric. These strategies lack modeling of user behavior and data focus, which can easily lead to critical data delays or excessive uploads of unnecessary data, resulting in redundant communications and wasted power. Furthermore, the lack of a unified multi-factor scheduling mechanism makes it difficult to dynamically adjust strategies based on real-time conditions, limiting the system's intelligence and adaptability in complex usage scenarios.

[0004] To this end, the present invention proposes a low-power data transmission optimization system and method for smart watches. Summary of the Invention

[0005] The present invention aims to provide a low-power data transmission optimization system and method for smart watches, realize intelligent scheduling and energy efficiency coordinated control of key data, and improve transmission efficiency and device endurance.

[0006] To achieve the above object, the present invention provides the following technical solutions: The present invention provides a low-power data transmission optimization system for smart watches, comprising: The state perception module collects information about the user's wearing status, interaction frequency, and activity intensity, and simultaneously obtains the smartwatch's battery level, communication protocol status, signal strength, and processor load, outputting user status tags and system resource parameters. The data evaluation module generates content priority tags based on the type of data to be transmitted, its historical usage frequency, and its timeliness sensitivity; The cognitive modeling module inputs user status tags and content priority tags into the prediction model, infers the user's short-term attention probability based on the sliding window behavior trend, and outputs the cognitive perception index; The scheduling decision module uses the cognitive perception index and system resource parameters as input to construct a scheduling cost function. It combines transmission energy consumption, cognitive risk, and resource occupancy factors to calculate and select the lowest-cost transmission strategy. The execution module is used to perform instant upload, delayed caching, lossy compression, data aggregation or communication protocol switching operations on the device side according to the selected strategy to achieve low-power data transmission scheduling.

[0007] As a preferred technical solution of the present invention, the step of outputting the user status label and the system resource parameter includes: The inertial measurement unit collects acceleration and gyroscope data and combines it with a photoelectric sensor to detect the user's wearing status; The interaction frequency is calculated based on the number of interaction events within a set time window, and the user activity intensity is evaluated using the change in cadence and acceleration amplitude. Integrate the wearing status, interaction frequency and activity intensity information to generate a user status label; By calling the system call interface of the operating system, the current battery power, communication protocol connection status, signal strength and processor utilization are periodically obtained, and a system resource parameter dataset is constructed.

[0008] As a preferred technical solution of the present invention, the step of generating a content priority label includes: Obtaining type information of the data to be transmitted, retrieving the frequency of occurrence of the data type in preset historical access records and its average user interaction response time, and calculating a historical usage frequency index; According to the set priority mapping matrix, the corresponding scoring weight template is matched according to the data type, and combined with the historical usage frequency and time sensitivity classification level, a comprehensive determination is made to generate a content priority label for subsequent scheduling strategy determination.

[0009] As a preferred technical solution of the present invention, the step of estimating the user's short-term attention probability includes: Constructing a behavior sequence based on a time window, which includes recent interaction frequency, activity intensity, and wearing duration; Building a behavior-data joint frequency matrix based on the behavior sequence and the content type distribution in the corresponding time period; The support vector regression model is used to fit the behavior-data joint frequency matrix and output the probability value of the current data being paid attention to by the user as the short-term attention probability.

[0010] As a preferred technical solution of the present invention, the calculation of the cognitive perception index includes: Set the historical latency tolerance distribution of the target data type as a baseline reference model; The short-term attention probability and content priority label are input into a bivariate exponential mapping function, and the output is a perception weight value used to measure the perceptibility of the current data delay. The degree of deviation between the current data to be transmitted and the historical tolerance threshold under the weight model is calculated to generate a normalized cognitive perception index.

[0011] As a preferred technical solution of the present invention, the step of constructing the scheduling cost function includes: The unit transmission energy consumption of target data, cognitive perception index and current processor utilization are used as transmission energy consumption factor, cognitive risk factor and resource occupancy factor respectively; Normalize each factor and map it to a unified evaluation interval; Based on the system's preset weight coefficients, different weights are assigned to the three types of factors to construct a scheduling cost function of weighted linear combination.

[0012] As a preferred technical solution of the present invention, the step of selecting the lowest cost transmission strategy includes: Enumerate the currently available transmission strategy options, including instant upload, delayed caching, lossy compression, data aggregation, and communication protocol switching; Calculate the corresponding scheduling cost function value for each strategy; Select the strategy with the lowest scheduling cost and satisfying the set constraints among all strategies, where the constraints include that the cognitive perception index must not be lower than the system safety perception threshold; Submit the selected strategy to the execution module for transmission operation control.

[0013] As a preferred technical solution of the present invention, the low-power data transmission scheduling specifically includes: Based on the scheduling type determined by the policy, perform immediate uploading or delayed caching; In case of delay, lossy compression or data aggregation processing is further performed based on the remaining resources; When it is detected that the energy consumption of the current protocol communication is too high, the communication protocol switching process is triggered.

[0014] The present invention also proposes a low-power data transmission optimization invention for smart watches, including: S100: Collect the user's wearing status, interaction frequency and activity intensity information, and simultaneously obtain the battery power, communication protocol status, signal strength and processor load of the smart watch, and output the user status label and system resource parameters; S200: Generate a content priority tag based on the type, historical usage frequency, and time sensitivity of the data to be transmitted; S300: Input the user status label and content priority label into the prediction model, estimate the user's short-term attention probability based on the sliding window behavior trend, and output the cognitive perception index; S400: Using the cognitive perception index and system resource parameters as input, a scheduling cost function is constructed. Combining transmission energy consumption, cognitive risk, and resource occupancy factors, the lowest-cost transmission strategy is calculated and selected. S500: Based on the selected strategy, instant uploading, delayed caching, lossy compression, data aggregation, or communication protocol switching operations are performed on the device side to achieve low-power data transmission scheduling.

[0015] The beneficial effects of the present invention are: 1. This invention introduces a "cognitive perception index" mechanism, which dynamically assesses the risk of user-perceived data delay by combining user status and content attributes through a support vector regression model. This index, no longer based solely on data type or preset priority, reflects the user's subjective attention to the data in the context of their current behavior. Combined with the historical distribution of delay tolerance, it dynamically determines whether to transmit data immediately or cache it later. This deeply aligns transmission behavior with the user's perceived experience, significantly improving the system's intelligent service response.

[0016] 2. This invention designs a scheduling cost function that integrates the cognitive perception index, system resource parameters, and transmission energy consumption estimates into a model, serving as a quantitative basis for scheduling strategy selection. This model does not rely on a fixed strategy table, but rather dynamically calculates the comprehensive cost of each strategy during actual operation. This model selects the optimal solution from a variety of scheduling methods, such as delay caching, lossy compression, and communication protocol switching, achieving a global balance between the real-time performance of critical data and overall device power consumption. This makes it particularly suitable for smartwatch applications in resource-constrained scenarios. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] The accompanying drawings are used to provide a further understanding of the present invention and constitute a part of the specification. Together with the embodiments of the present invention, they are used to explain the present invention and do not constitute a limitation of the present invention. In the accompanying drawings: Figure 1 This is a structural diagram of a low-power data transmission optimization system for smart watches according to the present invention; Figure 2 It is a flow chart of a low-power data transmission optimization method for smart watches of the present invention. DETAILED DESCRIPTION

[0018] The preferred embodiments of the present invention are described below with reference to the accompanying drawings. It should be understood that the preferred embodiments described herein are only used to illustrate and explain the present invention, and are not used to limit the present invention.

[0019] Example 1: Figure 1 As shown, the present invention provides a low-power data transmission optimization system for smart watches, including: The state perception module collects information about the user's wearing status, interaction frequency, and activity intensity, and simultaneously obtains the smartwatch's battery level, communication protocol status, signal strength, and processor load, outputting user status tags and system resource parameters. Furthermore, the step of outputting the user status tag and the system resource parameter includes: The inertial measurement unit collects acceleration and gyroscope data and combines it with a photoelectric sensor to detect the user's wearing status; The interaction frequency is calculated based on the number of interaction events within a set time window, and the user activity intensity is evaluated using the change in cadence and acceleration amplitude. Integrate the wearing status, interaction frequency and activity intensity information to generate a user status label; By calling the system call interface of the operating system, the current battery power, communication protocol connection status, signal strength and processor utilization are periodically obtained, and a system resource parameter dataset is constructed.

[0020] Specifically, a photoplethysmography sensor on the back of the smartwatch collects light signals reflected from the skin to determine whether the watch is snugly attached to the user's wrist. When the amplitude of the reflected light signal exceeds a set threshold and remains stable for a period of time (e.g., more than one second), the system determines that the watch is being worn. If the signal remains unstable or below the set threshold for a continuous period of time, the watch is considered not to be worn.

[0021] The system uses a built-in inertial measurement unit to obtain raw motion data from the three-axis accelerometer and three-axis gyroscope. The system sets a fixed time window (e.g., 60 seconds) and processes the sensor data within this time window. The system calculates the user's cadence during this time period by statistically analyzing the amplitude and frequency of changes in the acceleration modulus. Combining the cadence and acceleration amplitude statistics, the system maps these to a preset activity intensity level (e.g., stillness, light activity, moderate activity, and vigorous activity). Simultaneously, the system monitors the number of user interaction events within this time window, including touchscreen clicks, swipe gestures, and button presses. The total count of all interaction events is divided by the length of the time window to obtain the interaction frequency per unit time.

[0022] Based on the three parameters (wearing status, activity intensity level, and interaction frequency), the system uses an internal mapping table to convert them into standardized user status labels. For example, if the wearing status is "wearing," the activity intensity level is "moderate activity," and the interaction frequency is "frequent," the generated user status label can be represented as {wearing, moderate activity, frequent interaction}, which is used for subsequent cognitive perception modeling and scheduling decisions.

[0023] The following device operating status parameters are collected periodically (e.g., every 30 seconds) through the system call interface provided by the operating system: Battery level: Calling the power management interface to read the current remaining battery percentage; Communication protocol connection status: Querying the currently active communication protocol type (such as Bluetooth / Wi-Fi) and its connection status; Signal strength: Reading the received signal strength indicator of the Bluetooth or Wi-Fi connection; Processor load: Using the CPU status query interface provided by the kernel to obtain the CPU utilization within the past time slice. These parameters are organized into a system resource parameter dataset to support subsequent scheduling policy decisions.

[0024] The system packages these generated user status tags and system resource parameters into structured data and transmits it to the scheduling decision module, which serves as the core input for determining data transmission strategies. By combining user perception-level behavioral characteristics with device operational resource status, the system significantly improves the subsequent scheduling mechanism's adaptability to user status and device load conditions, providing a more refined contextual basis for energy optimization.

[0025] The data evaluation module generates content priority tags based on the type of data to be transmitted, its historical usage frequency, and its timeliness sensitivity; Furthermore, the step of generating content priority tags includes: Obtaining type information of the data to be transmitted, retrieving the frequency of occurrence of the data type in preset historical access records and its average user interaction response time, and calculating a historical usage frequency index; According to the set priority mapping matrix, the corresponding scoring weight template is matched according to the data type, and combined with the historical usage frequency and time sensitivity classification level, a comprehensive determination is made to generate a content priority label for subsequent scheduling strategy determination.

[0026] Specifically, after the data to be transmitted is generated, the system first classifies its type using the data source identification module. This type is determined based on the data's generation module or its characteristic structure. These types include, but are not limited to, health monitoring data (such as heart rate, blood oxygen levels, and sleep), short message notifications, application logs, and device status logs. For example, if the system detects that the newly generated data is a heart rate measurement result, it will mark it as "health monitoring data."

[0027] The system then accesses a local or cloud-based historical access database to extract the frequency of occurrence of this type of data and user response behavior within a set period, recording the access frequency and average interaction delay. To enhance enforceability, the system categorizes access frequency according to pre-set grading standards (e.g., high frequency ≥ 10 times / day, medium frequency 5-9 times, low frequency <5 times). It also maps average response delay to a response level based on threshold intervals (e.g., <3 seconds, 3-10 seconds, 10 seconds).

[0028] Based on this, the system uses a priority scoring matrix, combining the type category, access level, and response level as inputs, and then looks up the table to obtain the corresponding frequency of use score. For example, if the heart rate data is "health monitoring", the access level is "high", and the response level is "short delay", the corresponding score is 8 points.

[0029] The system then calculates the difference between the data's generation time and the current time, and combines this with a preset lifespan limit. If the data falls within the validity period (e.g., 10 minutes), the timeliness is considered "strong," otherwise, "weak." The system then assigns different timeliness sensitivity scores based on this information (e.g., strong = 5, medium = 3, weak = 1).

[0030] Finally, based on the predefined multi-level scoring strategy, the system maps the type-based weight, usage frequency score, and timeliness sensitivity score into priority level factors, and classifies and synthesizes them through the scoring matrix structure to generate content priority labels for the data, such as "high priority", "medium priority", or "low priority", which serve as input for subsequent cognitive modeling and scheduling decisions.

[0031] By integrating the data's own attributes with the user's historical usage behavior, a refined distinction in the importance of the data to be transmitted before scheduling is achieved, thereby effectively improving the priority protection capability of key data in an environment with limited communication resources.

[0032] The cognitive modeling module inputs user status tags and content priority tags into the prediction model, infers the user's short-term attention probability based on the sliding window behavior trend, and outputs the cognitive perception index; Furthermore, the step of estimating the user's short-term attention probability includes: Constructing a behavior sequence based on a time window, which includes recent interaction frequency, activity intensity, and wearing duration; Building a behavior-data joint frequency matrix based on the behavior sequence and the content type distribution in the corresponding time period; The support vector regression model is used to fit the behavior-data joint frequency matrix and output the probability value of the current data being paid attention to by the user as the short-term attention probability.

[0033] Specifically, the system sets a time window and captures user status tags every fixed sampling period. These tags include: interaction frequency (the number of interaction events per unit time, such as touches and button presses); activity intensity (the level of movement amplitude calculated by the accelerometer); and wear duration (the duration of continuous wear detection). Multiple status tags within this time window are arranged chronologically to form a user behavior time series, which is used to characterize short-term behavioral trends.

[0034] The system calculates the distribution of all data types to be transmitted within the time window, including health monitoring data, notification message data, and device log data. The frequency of each type of data is normalized to the content distribution vector within the window (e.g., health monitoring data: 60%; message notification data: 30%; log data: 10%). The system combines user behavior sequences with content type distribution to construct a two-dimensional matrix. Rows represent combinations of behavioral characteristics (e.g., "high interaction + high activity" or "low interaction + inactivity"); columns represent content types (e.g., health, notifications, logs); and each cell represents the frequency of a certain type of content appearing under a given behavior state. This joint frequency matrix indicates which type of data a user is more likely to generate or pay attention to under a given behavior state, and serves as the input for the prediction model.

[0035] The system uses the above-mentioned joint frequency matrix as input features and whether the user actually clicked / viewed this type of data in history as a label. It uses a support vector regression model for training, establishes a nonlinear mapping relationship, and outputs the attention probability. After training is completed, the SVR model is deployed in this module, receiving the current behavior status and data features in real time, and outputting the corresponding short-term attention probability q.

[0036] This implementation significantly improves the system's ability to dynamically predict changing trends in user attention by constructing a correlation mapping between behavioral context and data content, thereby providing more timely decision support for subsequent cognitive perception delay judgment and scheduling path selection.

[0037] Furthermore, the calculation of the cognitive perception index includes: Set the historical latency tolerance distribution of the target data type as a baseline reference model; The short-term attention probability and content priority label are input into a bivariate exponential mapping function, and the output is a perception weight value used to measure the perceptibility of the current data delay. The degree of deviation between the current data to be transmitted and the historical tolerance threshold under the weight model is calculated to generate a normalized cognitive perception index.

[0038] Specifically, for each data type (such as heart rate data, short message notifications), the system calculates the proportion of users who actually access the data at different delay times based on historical interaction behavior records, and forms a delay-access probability distribution curve to characterize the delay tolerance characteristics of this type of data. Based on this, the maximum acceptable delay time for each type is extracted. As a benchmark reference, for example, for heart rate data, if 90% of accesses occur within 1 minute of generation, then 1 minute can be set as the "tolerance threshold" to build a tolerance threshold interval model for this data type.

[0039] The system calls the user's current short-term attention probability q output by the cognitive modeling module and reads the content priority level p of the data to be transmitted, where the priority level is expressed in discrete integer form, such as low = 1, medium = 2, and high = 3.

[0040] The system inputs p and q into the two-variable exponential mapping function: in, Indicates the user's current perception tolerance time threshold for the data in a specific behavior state. and The adjustment coefficient set for the system reflects the weight of priority and probability of attention on perceived sensitivity.

[0041] The system calculates the difference between the actual delay value calculated by the difference between the generation time of the current data and the current system time, and the perception tolerance time threshold obtained in the above steps. If the actual delay is less than the tolerance value, the impact is set to low; if the deviation exceeds the tolerance range, the impact is set to high.

[0042] Finally, the system outputs a cognitive perception index based on a normalized function of the delay deviation, with a value range of [0,1]. The higher the value, the more likely the user is to perceive the data delay and the more immediate the transmission strategy needs to be executed.

[0043] This method establishes a tolerance reference model at the user cognitive level and quantitatively integrates current behavior with content factors to form an operational cognitive perception index, effectively supporting the scheduling strategy to make priority judgments based on "user experience sensitivity", thereby suppressing the energy consumption of non-perceived data while ensuring key experience.

[0044] The scheduling decision module uses the cognitive perception index and system resource parameters as input to construct a scheduling cost function. It combines transmission energy consumption, cognitive risk, and resource occupancy factors to calculate and select the lowest-cost transmission strategy. Furthermore, the step of constructing the scheduling cost function includes: The unit transmission energy consumption of target data, cognitive perception index and current processor utilization are used as transmission energy consumption factor, cognitive risk factor and resource occupancy factor respectively; Normalize each factor and map it to a unified evaluation interval; Based on the system's preset weight coefficients, different weights are assigned to the three types of factors to construct a scheduling cost function of weighted linear combination.

[0045] Specifically, the system obtains the target data's unit transmission energy consumption from the state perception module. This energy consumption value is calculated based on the historical average transmission power consumption of similar data types under the current communication mode. The system records the current and time consumed during the actual transmission of the same type of data and calculates the energy consumption per byte, which is used as the transmission energy consumption factor E.

[0046] The system receives the cognitive perception index R output by the cognitive modeling module. This index characterizes whether the current data delay is likely to be perceived by the user. The higher the value, the more sensitive the user is to the delay. This value is regarded as a cognitive risk factor in the current assessment.

[0047] The system periodically obtains the processor's current CPU utilization, C, through the operating system interface. This metric reflects the device's processing load at the current moment. Higher processor utilization indicates fewer schedulable resources and higher marginal power consumption costs for corresponding transmission operations. Therefore, this utilization factor is used as a resource utilization factor.

[0048] After obtaining the three raw factor values, the system normalizes them. The transmission energy factor is linearly scaled between the minimum and maximum transmission energy consumption recorded within a certain time window. The cognitive perception index itself is a normalized value. Processor utilization is directly mapped to a normalized value in the range [0, 1] from 0% to 100%.

[0049] After normalization, the system uses the preset weight coefficient , which correspond to the transmission energy consumption factor E, the cognitive risk factor R, and the resource occupancy factor C, respectively. The sum of their weights is 1. The scheduling cost function is calculated by the following weighted linear combination: Among them, the weight coefficient is set by the product side or obtained through long-term usage behavior statistics: if the system prioritizes user experience, the weight of the cognitive risk factor is increased ; If the system works in power saving mode, the transmission energy consumption factor weight is increased ; If the device resources are tight, the weight of the resource occupancy factor will be increased accordingly .

[0050] The lower the cost value derived from the scheduling cost function, the lower the overall cost of the transmission strategy in the current situation. The system inputs all available transmission strategies (such as instant upload, delayed caching, lossy compression, data aggregation, and communication protocol switching) into the above function, calculates the corresponding cost values, and selects the strategy with the lowest cost as the final execution plan and submits it to the execution module.

[0051] This mechanism realizes intelligent scheduling path selection under multi-objective constraints by introducing a joint evaluation of three factors: cognitive risk, energy consumption and resource occupancy in the scheduling process, thereby effectively avoiding the policy deviation caused by "only considering power or priority" and improving the robustness and energy efficiency balance of data scheduling behavior.

[0052] Furthermore, the step of selecting the transmission strategy with the lowest cost includes: Enumerate the currently available transmission strategy options, including instant upload, delayed caching, lossy compression, data aggregation, and communication protocol switching; Calculate the corresponding scheduling cost function value for each strategy; Select the strategy with the lowest scheduling cost and satisfying the set constraints among all strategies, where the constraints include that the cognitive perception index must not be lower than the system safety perception threshold; Submit the selected strategy to the execution module for transmission operation control.

[0053] Specifically, the system pre-sets five candidate transmission strategies: instant upload, delayed caching, lossy compression, data aggregation, and communication protocol switching. Each strategy corresponds to different energy consumption levels, transmission delays, and processing resource utilization characteristics. Before each round of scheduling tasks, the system enumerates these five strategies as candidate options, forming a pool of strategies to be evaluated.

[0054] The system then inputs the characteristic parameters of each candidate strategy into the constructed scheduling cost function. Combining the target data's unit energy consumption, cognitive perception index, and system resource utilization, it calculates the scheduling cost corresponding to each of the five strategies. The cost calculation for each strategy is performed using a unified weighting coefficient to ensure consistent evaluation dimensions.

[0055] After obtaining the scheduling costs for all strategies, the system sorts them in ascending order by cost, prioritizing the lowest-cost strategy. At the same time, the system establishes a set of basic operational constraints to exclude strategies that are inappropriate in specific scenarios. The most important constraint is that the cognitive perception index corresponding to the selected strategy must not fall below the system's safety perception threshold, ensuring that the strategy does not cause unreasonable user-perceivable delays. This safety perception threshold is either a factory-set system setting or the result of long-term user feedback data training to ensure energy-saving scheduling within a controllable cognitive risk range.

[0056] Under the premise of meeting the above constraints, the system ultimately selects the strategy with the lowest scheduling cost as the execution plan for this transmission task. This strategy is submitted to the execution module through the internal interface, which controls the smartwatch to complete the corresponding upload, caching, compression, aggregation, or protocol switching operations.

[0057] This step introduces a unified scheduling cost function, incorporating multiple system constraints and user perception factors into the calculation, ensuring that the selected transmission strategy takes into account the timeliness and perceived experience of key data while optimizing energy consumption, effectively improving the global optimization level and practicality of the scheduling strategy.

[0058] The execution module is used to perform instant upload, delayed caching, lossy compression, data aggregation or communication protocol switching operations on the device side according to the selected strategy to achieve low-power data transmission scheduling.

[0059] Furthermore, the low-power data transmission scheduling specifically includes: Based on the scheduling type determined by the policy, perform immediate uploading or delayed caching; In case of delay, lossy compression or data aggregation processing is further performed based on the remaining resources; When it is detected that the energy consumption of the current protocol communication is too high, the communication protocol switching process is triggered.

[0060] Specifically, the execution module receives a transmission control instruction from the scheduling decision module. This instruction clearly identifies the scheduling type of the data to be processed (for example, immediate upload, delayed buffering, compressed transmission, or aggregated transmission). The system initiates the corresponding sub-process based on this instruction.

[0061] If the instruction is "immediate upload", the system will establish a data channel through the currently selected communication protocol (such as Wi-Fi or Bluetooth low energy protocol) and prioritize the transmission of the target data block. During the upload process, it will monitor data integrity and the number of failed retransmissions to ensure the timeliness of critical data.

[0062] If the instruction is "delayed caching," the system stores the pending data in the device's local cache queue and records the entry timestamp. A periodic scheduler is also set to dynamically adjust the dequeue time based on the remaining system battery or changes in user behavior. This strategy is primarily used for data that is not time-sensitive, mitigating the increased power consumption caused by frequent communication module activation.

[0063] During the caching process, if the device's remaining battery life falls below a preset threshold or the communication channel is in poor condition, the system activates the compression processing submodule, invoking a lossy compression algorithm for image, audio, or log data to reduce the data size to a certain percentage of its original size, thereby reducing energy consumption and bandwidth usage during transmission. This process ensures that the compression ratio does not exceed the set distortion tolerance threshold, thereby avoiding affecting subsequent data analysis or presentation.

[0064] If there are multiple low-priority data items of the same source type in the current cache queue (such as sensor records generated consecutively within a short period of time), the system performs an aggregation operation, concatenating the multiple data items into a single composite data packet. This composite packet is uniformly compressed and transmitted, reducing the protocol overhead per data item.

[0065] If the system status parameters indicate that the current communication protocol is experiencing high power consumption or low efficiency (e.g., increased Wi-Fi module power consumption or high packet loss), the system will use pre-defined network evaluation logic to determine an alternative protocol (e.g., switching to Bluetooth LE) and execute the communication module switch command to reestablish the connection path and resume data transmission. This process prioritizes maintaining communication stability to avoid connection interruptions caused by frequent switching.

[0066] Through the above specific steps, the execution module can dynamically adapt the data transmission behavior according to the selected strategy, thereby effectively reducing communication power consumption in the resource-constrained environment of the smart watch while maintaining the transmission experience of key user data.

[0067] Example 2: A user went for a morning run wearing a smartwatch equipped with the system described in the present invention. The device runs an Android embedded system and supports both Bluetooth Low Energy (BLE) and WiFi communication protocols. It integrates a photoelectric sensor, an IMU module (including an accelerometer and gyroscope), a heart rate sensor, and a NAND local cache module.

[0068] The system activates the state perception module. The IMU module collects user acceleration and angular velocity data at a 200ms interval. Combined with the reflection intensity of the photosensor over a 5-second period, it determines whether the user is in a "continuous wear + moderate to high-intensity exercise" state. If the number of touch interactions within the 5-second period is zero, the system determines that the user has not actively engaged and marks it as "no interaction." The operating system interface provides resource parameters such as battery power (27%), processor load (45%), normal BLE connection status, and weak WiFi signal. After comprehensive analysis, the user status label "wearing and exercising" is generated and the current system resource parameters are output.

[0069] The system identifies the data currently to be transmitted as heart rate monitoring data (1Hz), motion trajectory data (refreshed every 10 seconds), and step count logs (written every 1 minute). Based on historical access frequency and user response latency, the system assigns these three data types priority labels of High, Medium, and Low, respectively, which serve as the basis for subsequent scheduling.

[0070] The system constructs a 30-second behavioral sequence, recording it as continuous high-intensity exercise with no user interaction. Based on a pre-trained support vector regression model, after inputting the behavioral sequence and content distribution data, the system predicts the probability of the current user paying attention to each data item: 0.93 for heart rate data, 0.68 for trajectory data, and 0.12 for step count logs.

[0071] The system inputs the aforementioned attention probabilities and content priority labels into a two-variable perception function model, generating corresponding perception weights of 0.93 for heart rate, 0.47 for trajectory, and 0.05 for log. Combined with the preset historical delay tolerance distribution (heart rate ≤ 0.5s, trajectory ≤ 5s, log ≤ 30s), the system evaluates the deviation of the current scheduling policy's delay settings and ultimately calculates the cognitive perception index for each type of data: approximately 0.9 for heart rate, approximately 0.5 for trajectory, and approximately 0.1 for log.

[0072] The scheduling module constructs a cost function based on transmission energy consumption, cognitive perception index, and resource utilization, assesses the cost of each data type, and generates a scheduling strategy. The strategy is as follows: Heart rate data: Due to its high interest and low tolerance, it is uploaded immediately via BLE; trajectory data: A 3-second buffer period is set, and uploads are concentrated to reduce redundant transmissions; step logs: Lossy compression is performed, and batch uploads are made after the WiFi signal is restored.

[0073] Based on this, the execution module starts the BLE channel to transmit heart rate data, enables the local cache to temporarily store trajectory data, and hangs the compressed log data into the non-urgent task queue, waiting for transmission when the low-power channel is activated.

[0074] To facilitate evaluation of the energy efficiency improvements of the present invention in actual use, this embodiment uses a traditional scheduled upload mechanism as a baseline for comparison. This traditional system behavior includes: All collected data is uploaded uniformly at fixed intervals (e.g., every 10 seconds), without distinguishing between data types or urgency; It does not consider the user's current behavior or interaction status, nor does it model user attention or cognitive delay; It uses the strongest signal protocol available for data transmission, without adaptive selection based on battery power, signal quality, or processor status; It also lacks a buffer delay mechanism, requiring real-time transmission of low-priority data, resulting in a high overall system communication load. This traditional mechanism, widely used in the data synchronization modules of existing commercial smartwatches, ensures uniform data upload, but suffers from significant power consumption redundancy and critical data transmission delays in complex environments.

[0075] Based on the above comparison mechanism, after a 5-minute morning jog test, under the premise of the same data integrity rate: communication energy consumption decreased by an average of 28.4%, mainly due to the cache delay and compression processing of non-critical data; the packet loss rate of critical data (such as heart rate) dropped to 0.4%, far better than the 2.1% packet loss rate under the traditional solution; the peak power consumption of WiFi transmission decreased by about 35%, attributed to the protocol switching mechanism reducing the frequency of high-power transmission.

[0076] Example 3: A low-power data transmission optimization method for smart watches, such as Figure 2 As shown, the following steps are included: S100: Collect the user's wearing status, interaction frequency and activity intensity information, and simultaneously obtain the battery power, communication protocol status, signal strength and processor load of the smart watch, and output the user status label and system resource parameters; S200: Generate a content priority tag based on the type, historical usage frequency, and time sensitivity of the data to be transmitted; S300: Input the user status label and content priority label into the prediction model, estimate the user's short-term attention probability based on the sliding window behavior trend, and output the cognitive perception index; S400: Using the cognitive perception index and system resource parameters as input, a scheduling cost function is constructed. Combining transmission energy consumption, cognitive risk, and resource occupancy factors, the lowest-cost transmission strategy is calculated and selected. S500: Based on the selected strategy, instant uploading, delayed caching, lossy compression, data aggregation, or communication protocol switching operations are performed on the device side to achieve low-power data transmission scheduling.

[0077] In summary, the present invention includes five core steps: user state perception, data content evaluation, cognitive modeling, policy decision-making and transmission execution. Specifically, the system first collects user behavior status and device resource information to generate status labels and resource parameters; then generates content priority labels based on data type, frequency of use and timeliness. Subsequently, the behavior and content labels are combined to predict the probability of short-term user attention, forming a cognitive perception index to measure the perceived risk of delayed transmission. The system constructs a scheduling cost function based on the perception index and resource parameters, selects the lowest cost strategy, and the execution module implements upload, caching, compression or communication switching operations to achieve a balance between energy consumption optimization and key data protection.

[0078] Finally, it should be noted that the above descriptions are merely preferred embodiments of the present invention and are not intended to limit the present invention. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art will be able to modify the technical solutions described in the aforementioned embodiments or substitute equivalents for some of the technical features. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of the present invention shall be included within the scope of protection of the present invention.

Claims

1. A low-power data transmission optimization system for smart watches, characterized in that: include: The state perception module collects information about the user's wearing status, interaction frequency, and activity intensity, and simultaneously obtains the smartwatch's battery level, communication protocol status, signal strength, and processor load, outputting user status tags and system resource parameters. The data evaluation module generates content priority tags based on the type of data to be transmitted, its historical usage frequency, and its timeliness sensitivity; The cognitive modeling module inputs user status tags and content priority tags into the prediction model, infers the user's short-term attention probability based on the sliding window behavior trend, and outputs the cognitive perception index; The scheduling decision module uses the cognitive perception index and system resource parameters as input to construct a scheduling cost function. It combines transmission energy consumption, cognitive risk, and resource occupancy factors to calculate and select the lowest-cost transmission strategy. The execution module is used to perform instant upload, delayed caching, lossy compression, data aggregation or communication protocol switching operations on the device side according to the selected strategy to achieve low-power data transmission scheduling.

2. A low-power data transmission optimization system for smart watches according to claim 1, characterized in that: The step of outputting the user status label and the system resource parameter comprises: The inertial measurement unit collects acceleration and gyroscope data and combines it with a photoelectric sensor to detect the user's wearing status; The interaction frequency is calculated based on the number of interaction events within a set time window, and the user activity intensity is evaluated using the change in cadence and acceleration amplitude. Integrate the wearing status, interaction frequency and activity intensity information to generate a user status label; By calling the system call interface of the operating system, the current battery power, communication protocol connection status, signal strength and processor utilization are periodically obtained, and a system resource parameter dataset is constructed.

3. The low-power data transmission optimization system for smart watches according to claim 1, characterized in that: The step of generating a content priority label includes: Obtaining type information of the data to be transmitted, retrieving the frequency of occurrence of the data type in preset historical access records and its average user interaction response time, and calculating a historical usage frequency index; According to the set priority mapping matrix, the corresponding scoring weight template is matched according to the data type, and combined with the historical usage frequency and time sensitivity classification level, a comprehensive determination is made to generate a content priority label for subsequent scheduling strategy determination.

4. The low-power data transmission optimization system for smart watches according to claim 1, characterized in that: The step of estimating the user's short-term attention probability includes: Constructing a behavior sequence based on a time window, which includes recent interaction frequency, activity intensity, and wearing duration; Building a behavior-data joint frequency matrix based on the behavior sequence and the content type distribution in the corresponding time period; The support vector regression model is used to fit the behavior-data joint frequency matrix and output the probability value of the current data being paid attention to by the user as the short-term attention probability.

5. The low-power data transmission optimization system for smart watches according to claim 1, characterized in that: The calculation of the cognitive perception index includes: Set the historical latency tolerance distribution of the target data type as a baseline reference model; The short-term attention probability and content priority label are input into a bivariate exponential mapping function, and the output is a perception weight value used to measure the perceptibility of the current data delay. The degree of deviation between the current data to be transmitted and the historical tolerance threshold under the weight model is calculated to generate a normalized cognitive perception index.

6. The low-power data transmission optimization system for smart watches according to claim 1, characterized in that: The steps of constructing the scheduling cost function include: The unit transmission energy consumption of target data, cognitive perception index and current processor utilization are used as transmission energy consumption factor, cognitive risk factor and resource occupancy factor respectively; Normalize each factor and map it to a unified evaluation interval; Based on the system's preset weight coefficients, different weights are assigned to the three types of factors to construct a scheduling cost function of weighted linear combination.

7. The low-power data transmission optimization system for smart watches according to claim 1, characterized in that: The step of selecting the transmission strategy with the lowest cost includes: Enumerate the currently available transmission strategy options, including instant upload, delayed caching, lossy compression, data aggregation, and communication protocol switching; Calculate the corresponding scheduling cost function value for each strategy; Select the strategy with the lowest scheduling cost and satisfying the set constraints among all strategies, where the constraints include that the cognitive perception index must not be lower than the system safety perception threshold; Submit the selected strategy to the execution module for transmission operation control.

8. The low-power data transmission optimization system for smart watches according to claim 1, characterized in that: The low-power data transmission scheduling specifically includes: Based on the scheduling type determined by the policy, perform immediate uploading or delayed caching; In case of delay, lossy compression or data aggregation processing is further performed based on the remaining resources; When it is detected that the energy consumption of the current protocol communication is too high, the communication protocol switching process is triggered.

9. A low-power data transmission optimization invention for smart watches, characterized in that: include: S100: Collect the user's wearing status, interaction frequency and activity intensity information, and simultaneously obtain the battery power, communication protocol status, signal strength and processor load of the smart watch, and output the user status label and system resource parameters; S2 00: Generates content priority tags based on the type of data to be transmitted, historical usage frequency, and time sensitivity; S300: Input the user status label and content priority label into the prediction model, estimate the user's short-term attention probability based on the sliding window behavior trend, and output the cognitive perception index; S400: Using the cognitive perception index and system resource parameters as input, a scheduling cost function is constructed. Combining transmission energy consumption, cognitive risk, and resource occupancy factors, the lowest-cost transmission strategy is calculated and selected. S500: Based on the selected strategy, instant uploading, delayed caching, lossy compression, data aggregation, or communication protocol switching operations are performed on the device side to achieve low-power data transmission scheduling.

Images