DM-RS Overhead Reduction via AI CSI Inference
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current communication technologies face challenges in reducing DM-RS overhead while maintaining effective data transmission, especially in scenarios where CSI is stable or infrequently changing, such as in AI/ML-enabled networks, FWA, or low mobility scenarios, leading to interference issues and increased complexity in data mapping.
Innovation Solution
The method involves determining whether to transmit no or less DM-RS on specific symbols and using PT-RS in those slots, along with indications in DCI for terminal devices to demodulate data based on trained CSI, thereby reducing DM-RS overhead and avoiding interference.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Quantity of substance
If DM-RS transmission density is reduced to lower overhead, then resource efficiency is improved, but demodulation reliability deteriorates due to insufficient channel state information
Solution Approach 1:
The patent applies preliminary action by having the terminal device infer CSI in advance using AI/ML algorithms before actual data transmission. This pre-inferred CSI is then used for demodulation, eliminating the need for real-time DM-RS transmission while maintaining demodulation reliability. The terminal device performs channel estimation proactively rather than reactively.
Solution Approach 2:
The patent implements self-service by enabling the terminal device to autonomously infer its own CSI using AI/ML models without relying on network-provided DM-RS signals. The device uses previously received reference signals and stored channel characteristics to self-determine current channel state, reducing dependency on continuous network signaling.
2Quantity of substance
If no or less DM-RS transmission is used, then resource overhead is reduced, but data mapping complexity increases due to need for alternative demodulation approaches
Solution Approach 1:
The patent replaces the traditional mechanical DM-RS transmission system with an AI/ML-based virtual CSI inference system. Instead of using physical reference signals to convey channel information, the system uses intelligent algorithms to predict and infer channel state, substituting a complex signaling mechanism with a cognitive processing approach.
Solution Approach 2:
The patent changes the fundamental parameter of CSI acquisition from signal-based measurement to algorithm-based inference. By transitioning from measuring channel characteristics through DM-RS signals to predicting them through AI/ML models, the system alters how channel state is obtained, enabling reduced DM-RS overhead without proportionally increasing mapping complexity.
3Productivity
If DM-RS transmission is reduced in stable CSI scenarios, then resource efficiency is improved, but interference management becomes more challenging between co-scheduled UEs
Solution Approach 1:
The patent implements feedback mechanisms where terminal devices with AI/ML capability report their CSI inference accuracy and channel stability characteristics to the network. This feedback enables the network to identify suitable candidates for reduced DM-RS transmission and adjust scheduling decisions accordingly, managing interference through informed resource allocation rather than conservative universal DM-RS transmission.
Solution Approach 2:
The patent applies local quality by differentiating DM-RS transmission requirements across different terminal devices and channel conditions. Instead of uniform DM-RS transmission to all UEs, the system selectively applies reduced DM-RS overhead only to devices with stable CSI characteristics and AI/ML capability, while maintaining normal DM-RS transmission for others, thereby managing interference locally rather than globally.
Data Source
AI summary
Embodiments of the present disclosure relate to methods, devices and computer readable media of communication. A transmitting device determines whether no DM-RS is to be transmitted on a first symbol configured for DM-RS transmission and if no DM-RS is to be transmitted on the first symbol, transmits a signal to a receiving device in at least part of REs on the first symbol, the signal comprising at least one of data or a PT-RS. The receiving device demodulates the signal from the at least part of REs on the first symbol. In this way, DM-RS overhead may be reduced and the interference to co-scheduled UE may be avoided.