ML Compression Gain Prediction for 5G Uplink Buffer Status
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
In 5G NR wireless communications, the existing buffer status report mechanism does not account for the time delay incurred during dynamic scheduling, leading to excessive uplink resource allocation and wastage of network resources, as well as increased power consumption in user equipment (UE) due to inefficient compression gain prediction.
Innovation Solution
A machine learning model, specifically a neural processing engine, is used to predict the compression gain of uplink data during the time delay, allowing for a more accurate buffer status report generation that considers the amount of data that can be compressed, thereby optimizing uplink bandwidth usage and reducing unnecessary resource allocation.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If the existing buffer status report mechanism is used without compression gain prediction, then the network resource allocation is simpler, but it leads to excessive uplink resource allocation and wastage of network resources
Solution Approach 1:
The patent applies preliminary action by predicting compression gain before the actual data transmission occurs. The machine learning model estimates how much data will be compressed during the time delay period, allowing the buffer status report to be generated with accurate expectations of future compression effects. This enables the network to allocate uplink resources more efficiently by knowing in advance how much data will actually need to be transmitted after compression.
2Loss of energy
If compression gain is predicted accurately considering time delay, then uplink bandwidth usage is optimized, but the prediction process increases computational complexity
Solution Approach 1:
The patent applies parameter changes by using a machine learning model that takes multiple input parameters into account, including the original amount of uplink data, current compression buffer occupancy, CPU utilization, and MAC padding data. The model processes these parameters through learned transformations to output the predicted compression gain. This approach optimizes UE power consumption by accurately predicting how much data will be compressed, thereby reducing unnecessary resource allocation and transmission time.
3Productivity
If the buffer status report includes predicted compression gain, then network resource wastage is reduced, but the report generation time increases
Solution Approach 1:
The patent applies mechanics substitution by replacing traditional mechanical buffer status reporting with a machine learning-based prediction system. Instead of simply reporting current buffer contents, the system uses a neural network model to predict future compression outcomes based on historical data and current conditions. This substitution enables accurate network resource allocation decisions while the model processes predictions efficiently, minimizing the time added to report generation.
Data Source
AI summary
A method of wireless communication by a user equipment (UE) includes determining an original amount of uplink data stored in a non-compression buffer, the original amount of data designated to be transmitted to a network. The method also includes predicting, with a machine learning model, an amount of compression gain for the original amount of uplink data. The compression gain is obtained by compressing the original amount of uplink data during a time delay. The method further includes generating a buffer status report based on the original amount of uplink data and the predicted amount of compression gain. The method transmits the buffer status report to the network.