Machine Learning Beamforming for Wireless Networks
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current wireless communication systems face inefficiencies in downlink beamforming, particularly in massive MIMO scenarios, due to limitations in PMI-based and eNB measurement-based methods, which require excessive reference signals and are not always accurate, especially for legacy devices and varying channel conditions.
Innovation Solution
A machine learning-based approach that utilizes a neural network to establish relationships between uplink channel information and downlink beamforming information, reducing the need for downlink reference signals and improving beamforming performance by leveraging sector-specific data logs and adaptive learning.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If PMI-based downlink beamforming methods are used, then beamforming performance can be improved, but excessive reference signals are required increasing overhead
Solution Approach 1:
The system performs preliminary actions by having the wireless device report downlink beamforming information in advance, and the network node stores this information for later use. This allows the system to retrieve previously determined beamforming information without requiring new reference signals, thereby reducing overhead while maintaining beamforming performance.
Solution Approach 2:
The system uses copying by storing previously reported downlink beamforming information and retrieved uplink channel information in data logs. These stored copies can be reused for beamforming decisions without requiring fresh measurements or reference signals, reducing the quantity of reference signals needed while preserving beamforming accuracy.
2Reliability
If downlink reference signals are transmitted frequently to maintain beamforming accuracy, then beamforming performance is improved, but radio spectrum efficiency deteriorates
Solution Approach 1:
The system implements feedback mechanisms where wireless devices report downlink beamforming information (such as PMI) to the network node. The network node stores this feedback information and uses it for subsequent beamforming operations, eliminating the need for frequent reference signal transmissions while maintaining beamforming accuracy through intelligent reuse of feedback data.
Solution Approach 2:
The system performs preliminary actions by collecting and storing beamforming information from device reports before it would be needed for actual beamforming operations. This advance preparation allows the system to retrieve stored information instead of transmitting new reference signals, improving radio spectrum efficiency while preserving beamforming accuracy.
3Ease of operation
If eNB measurement-based downlink beamforming is used, then beamforming can be provided without device reports, but accuracy deteriorates for legacy devices and varying channel conditions
Solution Approach 1:
The system merges two approaches by combining eNB measurement-based beamforming with device-reported beamforming information. The network node stores both types of information and can retrieve either or both depending on the situation, thereby maintaining operational simplicity while improving accuracy through the complementary use of multiple information sources.
Solution Approach 2:
The system applies parameter changes by adapting the beamforming approach based on device type and channel conditions. For legacy devices or specific scenarios, the system can rely more on stored eNB measurements, while for capable devices with stable channels, it uses device reports. This dynamic adjustment of information sources maintains accuracy across different device types and channel conditions.
4Reliability
If sector-specific data logs are maintained for machine learning, then beamforming performance is improved, but device complexity increases
Solution Approach 1:
The system applies universality by implementing a standardized data log structure that can serve multiple functions: storing downlink beamforming information, storing uplink channel information, supporting machine learning training, and enabling beamforming retrieval. This multi-functional approach consolidates what could be multiple separate systems into a single unified structure, reducing overall complexity while improving beamforming performance through comprehensive data availability.
Data Source
Figure 1
Figure 2
Figure 3
AI summary
A wireless communications system (10) and a method therein for transmission of a downlink signal in a wireless communications network (100) supporting beamforming. The system estimates uplink channel information for a radio link from a wireless device (120) to a radio network node (110). Further, the system obtains downlink beamforming information related to the estimated uplink channel information from a machine learning unit (300). The machine learning unit comprises relationships between beamforming information and uplink channel information determined based on stored pairs of reported downlink beamforming information and measured uplink channel information. The obtained downlink beamforming information is applicable for the radio cell sector. Furthermore, the system transmits, towards the wireless device (120), a beamformed downlink signal using the obtained downlink beamforming information.