Application Preloading Prediction Using KNN and Device Status
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current application preloading methods in terminal devices are inefficient as they often preload too many resources, wasting storage space and increasing power consumption, leading to reduced device fluency and accuracy in predicting the next application to be launched.
Innovation Solution
A method that acquires status feature information of a terminal device to predict the next application to be launched by comparing it with pre-collected samples, using the KNN algorithm to select the most frequently indicated applications based on storage space information, and preloading those applications to improve launching speed while conserving storage space.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Speed
If too many resources are preloaded, then application launching speed is improved, but storage space consumption increases and power consumption increases
Solution Approach 1:
The patent applies preliminary action by predicting which applications the user is likely to launch next and preloading their resources in advance. The system analyzes historical usage patterns, current context (time, location, device state), and user behavior to proactively prepare application resources before they are actually needed, thus improving launching speed without requiring all applications to be preloaded simultaneously.
Solution Approach 2:
The patent changes the parameter of preloading quantity from a fixed or maximum value to a dynamically adjusted value based on multiple factors including available storage space, battery level, current application usage, and predicted user behavior. This allows the system to optimize the balance between launching speed and resource consumption by adjusting how many applications are preloaded under different conditions.
2Speed
If too many resources are preloaded, then application launching speed is improved, but power consumption increases
Solution Approach 1:
The system performs preliminary action by predicting future application needs and preloading resources only for those specific applications, rather than continuously maintaining all applications in a ready state. This reduces overall power consumption by activating preloading only when prediction confidence is high and resources are actually needed soon.
Solution Approach 2:
The patent adjusts the preloading parameter dynamically based on battery level and power management policies. When battery level is low or power saving mode is active, the system reduces or suspends preloading operations. When battery level is high and charging is available, the system can perform more aggressive preloading, thus adapting power consumption to actual energy availability.
3Quantity of substance
If precise prediction of next application is implemented, then storage space is conserved, but system complexity increases
Solution Approach 1:
The patent employs a multi-functional prediction system that simultaneously analyzes multiple data sources (historical usage patterns, current context, device state, user behavior) using a unified machine learning model. This single system performs multiple functions including pattern recognition, context analysis, and prediction, rather than requiring separate systems for each function, thus managing complexity while achieving accurate predictions.
Solution Approach 2:
The system uses machine learning models that are trained on historical data and then copied/deployed to the device for real-time prediction. The complex training and analysis are performed offline to create simplified prediction models that can run efficiently on the device, copying only the essential prediction logic rather than the entire analysis system.
Data Source
Figure 1
Figure 2
Figure 3
AI summary
A method for preloading an application, a storage medium, and a terminal device are provided. The method includes the following. First status feature information of a terminal device is acquired in response to an application-preloading-prediction event being detected to be triggered. The first status feature information is compared with a plurality of pre-collected samples of a sample set. The plurality of pre-collected samples include status second feature information of the terminal device in a preset sampling period, and each sample of the plurality of pre-collected samples corresponds to a sample tag indicating a next application to be launched. A target application to be launched is predicted according to a comparison result. The target application is preloaded. By adopting the above technical solutions of embodiments of the disclosure, the application to be launched can be predicted accurately, thus improving launching speed of the application under the precondition of saving storage space.