Network Performance Prediction Using Historic Data
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current network performance measurements often lag behind real-time conditions due to delays in data collection and processing, leading to suboptimal control decisions that degrade network performance.
Innovation Solution
A method to predict current performance measurements using historic and recent data, which involves correlating cellular network cells with Wi-Fi cells and adjusting the RSSI-Admit threshold to control access selection, ensuring that user terminals switch between networks based on predicted throughput equality, thereby optimizing data throughput.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If current performance measurements are used for network control, then network performance can be optimized in real-time, but data collection and processing delays make current measurements unavailable
Solution Approach 1:
The system performs preliminary actions by collecting and storing historic performance measurement data in advance, so that when current measurements are unavailable due to delays, the system can still make informed control decisions using the pre-collected historic data combined with recent measurements
Solution Approach 2:
The invention introduces an intermediary approach by using a combination of historic and recent performance measurements as a mediator between the unavailable current measurement and the network control function, allowing control decisions to be made with sufficient accuracy despite measurement delays
2Ease of operation
If performance data that lags current time is used for network control, then data collection delays are accommodated, but network performance degrades
Solution Approach 1:
The system dynamically adjusts the weighting between historic and recent performance measurements based on current network conditions and data availability, allowing the control function to adapt to changing circumstances and maintain optimal performance despite lagging data
Solution Approach 2:
The invention changes the parameter of data freshness by combining historic and recent measurements in a weighted average, where the weights can be adjusted to balance between using older more reliable data and newer more current data, thereby optimizing both ease of operation and network performance
Data Source
Figure 1
Figure 2A~2B
Figure 3
AI summary
Prediction techniques are presented for predicting a current value of a performance measurement using historic performance measurement data and recent performance measurement data, which may not reflect the current conditions. A daily trend in a series of performance measurements is obtained. The daily trend comprises a set of data points at different times of day, wherein each data point represents an average value of the performance measurements at a corresponding time of day over a plurality of days. One or more recent performance measurements over a recent time interval are also obtained. A current value of the performance measurements is predicted, based on the daily trend and the recent performance measurements. The prediction is in particular useful when a current performance measurement may not be immediately available due to delays in data collection, data processing, and data transport. An embodiment uses the prediction for admission control in a wireless network.