Pilot Grouping for Massive MIMO Channel Estimation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current massive MIMO systems face challenges in achieving high channel estimation accuracy while reducing the resource requirement for training signals, particularly in downlink channel estimation for multiple user equipment (UEs) in frequency-division duplex (FDD) systems, where existing methods either consume excessive resources or result in estimation errors due to insufficient spatial correlation.
Innovation Solution
The method involves generating candidate pilots with indices, determining dominant spatial correlation matrices for each UE, computing average received signal strength, selecting subsets of pilots based on thresholds, grouping these subsets to form a union set, broadcasting indices, and transmitting the union set of pilots for channel estimation, allowing UEs to perform accurate channel estimation with reduced training signals.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If cell-specific reference signals (CRS) are used for downlink channel estimation, then channel estimation can be performed, but the number of training signals increases to the order of the number of antennas, consuming excessive resources
Solution Approach 1:
The patent segments the training signal resources by introducing a group-based pilot structure where users are divided into multiple groups, each using a common pilot sequence. This segmentation allows the system to reduce the total number of training signals from being proportional to the number of antennas to being proportional to the number of groups, thereby resolving the contradiction between measurement precision and quantity of training signals.
Solution Approach 2:
The patent applies universality by designing group-common pilot sequences that can be shared across multiple users within the same group. This multi-functional pilot structure allows a single pilot sequence to serve multiple users simultaneously, reducing the overall training signal overhead while maintaining channel estimation capability for all users in the group.
2Measurement precision
If dedicated pilots are assigned to each UE for channel estimation, then channel estimation can be performed for each user, but the number of training signals increases to the order of the number of UEs, consuming excessive resources
Solution Approach 1:
The patent merges the training signal resources by combining multiple users into groups that share common pilot sequences. Instead of assigning dedicated pilots to each UE, users within the same group share a common pilot, which merges the training signal requirements and reduces the total number of training signals from being proportional to the number of UEs to being proportional to the number of groups.
Solution Approach 2:
The patent applies universality by designing group-common pilot sequences that can be shared across multiple users within the same group. This multi-functional pilot structure allows a single pilot sequence to serve multiple users simultaneously, reducing the overall training signal overhead while maintaining channel estimation capability for all users in the group.
3Quantity of substance
If traditional MIMO CE method with scattered pilots on subset of antennas is used, then training overhead is reduced, but estimation error increases when spatial correlation is not high enough
Solution Approach 1:
The patent applies parameter changes by utilizing the spatial correlation parameter of the channel. Instead of treating all antenna elements independently, the system exploits the spatial correlation structure to infer channel information at unmeasured antenna positions based on measurements at a subset of antennas, thereby maintaining estimation accuracy while reducing training overhead.
Solution Approach 2:
The patent introduces spatial correlation as an intermediary that mediates between the limited pilot measurements and the full channel estimation. By using spatial correlation matrices as an intermediary, the system can accurately estimate channels at all antenna positions based on limited pilot measurements at a subset of antennas, resolving the contradiction between training overhead and estimation accuracy.
4Quantity of substance
If grouping users based on channel spatial correlation with predetermined group-specific correlation matrices is used, then training signal resources are reduced, but CE accuracy degrades because pilots for each group may not cover all spatial directions
Solution Approach 1:
The patent applies dynamics by making the group formation and pilot assignment adaptive rather than static. The system dynamically adjusts the grouping configuration and pilot selection based on the actual spatial correlation characteristics of users, ensuring that pilots are optimally distributed to cover the spatial directions required by each group, thereby maintaining accuracy while reducing training resources.
Data Source
AI summary
The presently claimed invention provides a method of channel estimation in a multi-user massive Multiple-Input Multiple-Output system. Candidate pilots are selected for each user equipment based on their spatial correlation matrices. Through determining the similarity of spatial correlation matrices among different user equipments, shared pilots among them can be found, and a base station can transmit a union set of pilots for channel estimation. The present invention is able to provide high channel estimation accuracy and reduce training signal resource.