Estimating RSRP Values from CSI Data in Wireless Networks
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current RSRP reporting in wireless communication systems is inadequate due to irregular and infrequent reporting, averaging over extended periods, and inconsistencies among different UE equipment vendors, leading to reduced reliability and accuracy, which limits network operation and ML model performance.
Innovation Solution
A method using machine learning models, such as neural networks, to estimate RSRP values based on reported CSI values, including CQI, RI, and PMI, leveraging the strong correlation between CQI and RSRP to provide more frequent and accurate RSRP estimates, reducing signaling overhead while maintaining data quality.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Loss of energy
If RSRP reporting is performed irregularly and infrequently to reduce signaling overhead, then signaling overhead is reduced, but RSRP measurement precision and reliability deteriorate
Solution Approach 1:
The patent introduces CQI values as an intermediary parameter to estimate RSRP values. Instead of directly reporting RSRP, the system uses CQI (Channel Quality Indicator) which is already being reported, as a mediator to derive RSRP estimates through machine learning models. This approach leverages existing signaling to obtain additional measurement information without increasing overhead.
Solution Approach 2:
The patent creates estimated RSRP values as copies of actual RSRP measurements by using machine learning models trained on the relationship between CQI and RSRP. These estimated values serve as proxies for direct measurements, allowing the system to maintain measurement precision while reducing the need for frequent direct RSRP reporting.
2Productivity
If RSRP values are averaged over extended periods to reduce reporting frequency, then reporting frequency is reduced, but RSRP measurement precision deteriorates
Solution Approach 1:
The patent replaces the mechanical averaging process with a machine learning-based estimation system. Instead of simply averaging RSRP values over time, the system uses neural networks and other ML models that can capture complex temporal relationships and patterns in the data, providing more accurate estimates without requiring frequent direct measurements.
Solution Approach 2:
The patent changes the parameter being reported from direct RSRP measurements to CQI-based RSRP estimates. By transforming the measurement approach and using different parameters (CQI values) that correlate with RSRP, the system can maintain precision while changing the reporting characteristics.
3Productivity
If RSRP reporting is reduced to improve network efficiency, then network efficiency is improved, but RSRP reliability deteriorates
Solution Approach 1:
The patent implements a feedback mechanism where the machine learning model continuously learns from the relationship between CQI and RSRP values. The system uses historical data to train models that can reliably estimate current RSRP from current CQI values, creating a self-improving system that enhances reliability while reducing reporting requirements.
Solution Approach 2:
The patent creates reliable copies of RSRP information through estimated values derived from CQI. These estimated RSRP values maintain the reliability needed for network operations such as handover decisions while reducing the burden of direct RSRP measurement and reporting.
4Measurement precision
If machine learning models are used to estimate RSRP values, then RSRP measurement precision is improved, but device complexity increases
Solution Approach 1:
The patent uses CQI values as an intermediary that already exists in the system to drive the machine learning estimation process. By leveraging this existing parameter rather than requiring new sensors or measurement mechanisms, the patent reduces the complexity increase that would otherwise result from implementing ML-based RSRP estimation.
Data Source
Figure 1
Figure 2
Figure 3A
AI summary
According to an example embodiment, an apparatus includes at least one processor; and at least one memory including computer program code; at least one memory and the computer program code configured to, with the at least one processor, cause the apparatus at least to: receive a plurality of channel state information (CSI) values reported by the user device; determine a plurality of estimated reference signal received power (RSRP) values that have been estimated based at least on the plurality of CSI values; and output a sequence of RSRP values, including at least the estimated RSRP values.