Real-Time Media Flow Prediction for Adaptive Wireless FEC
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Wireless communication networks face challenges in predicting future network conditions for real-time media streaming, leading to suboptimal selection of transmission rates and error correction rates, especially in mobile environments where link quality changes rapidly, resulting in delayed adaptation and inaccurate packet loss prediction.
Innovation Solution
A method and system that collect radio network statistics from multiple UEs in a cell to predict future throughput and packet loss for a specific UE, providing this information to the sender for selecting suitable transmission rates and error correction, using machine-learning algorithms to build intelligence from historical data for accurate predictions.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If receiver-based feedback methods are used to estimate packet loss and adapt FEC, then adaptive error correction can be achieved, but the response is delayed because the sender must wait for feedback from the receiver
Solution Approach 1:
The base station performs preliminary actions by collecting radio network statistics and training machine learning models in advance, before real-time media transmission begins. This allows the prediction model to be ready immediately when transmission starts, eliminating the need to wait for receiver feedback during the initial adaptation phase.
Solution Approach 2:
A machine learning prediction model acts as an intermediary between the base station's radio network statistics and the sender's transmission parameter selection. The model predicts future throughput and packet loss without requiring direct feedback loops between receiver and sender, thereby reducing adaptation delay.
2Ease of manufacture
If static FEC with fixed redundancy rate is used, then implementation is simple, but bandwidth efficiency is poor when error rate varies
Solution Approach 1:
The system implements dynamic FEC by using machine learning models to predict future network conditions and adjust the redundancy rate accordingly. Instead of a fixed redundancy rate, the system adapts the FEC parameters in real-time based on predicted throughput and packet loss, optimizing bandwidth efficiency while maintaining implementation feasibility through automated model-based decisions.
3Measurement precision
If receiver-based packet loss prediction is used, then end-to-end loss estimation is achieved, but prediction accuracy is reduced compared to network-based prediction
Solution Approach 1:
The machine learning prediction model serves as an intermediary that processes comprehensive radio network statistics (including RSRP, RSRQ, cell load, number of users) at the base station level. This network-based approach provides more accurate predictions than receiver-based methods because the base station has access to broader network context information that individual receivers cannot observe.
Data Source
Figure 1~2
Figure 3
Figure 4~5
AI summary
Disclosed is a method performed by a system of a wireless communication network (100) for determining transmission conditions for a real-time media flow to be transmitted wirelessly to a first User Equipment, UE, (121) residing in a first cell (115) served by a first base station (110) of the network. The method comprises obtaining radio network statistics of the first cell (115), the radio network statistics comprising radio network data for individual of a plurality of UEs (122) in the first cell, UE throughput for individual of the plurality of UEs, and UE packet loss or packet delay for individual of the plurality of UEs, and obtaining, at a second time point occurring later in time than the obtaining of the radio network statistics, radio network data for the first UE (121). The method further comprises determining a predicted first UE throughput and a predicted first UE packet loss or packt delay based on the obtained radio network statistics of the first cell (115) and on the obtained radio network data for the first UE (121), and triggering sending of information of the predicted first UE throughput and of the predicted first UE packet loss or excessive packet delay to a sender of the real-time media flow to the first UE (121). A Forward Error Correction, FEC, optimization module (406) takes the predicted UE packet loss rate, the predicted UE throughput, and adapts the amount of FEC redundancy packets according to the throughput and packet loss rate. Further, the adapted amount of FEC redundancy packets are then used by the sender of the video when transmitting the video.