Method and device for learning-based channel estimation with data pre-processing
AssigneeSamsung Electronics Co., Ltd.
InventorsFederico Penna, Hyuk Joon Kwon
Life of this patent
Estimating the term…
Abstract
Methods and apparatuses are provided in which frequency domain (FD) interpolation is performed on wideband reference signal (RS) resource elements (REs) to generate channel estimates on corresponding RS symbols. Denoising may be performed on the channel estimates in a time domain (TD) or a delay-Doppler domain (DD). Learning-based channel estimation (LCE) is applied on the channel estimates using a neural network (NN) to generate refined channel estimates on the corresponding RS symbols. Time domain (TD) interpolation is performed on the refined channel estimates to produce a channel estimate across symbols.
Claims20
A method comprising:
- performing, by a user equipment (UE), frequency domain (FD) interpolation on wideband reference signal (RS) resource elements (REs) to generate channel estimates on corresponding RS symbols;
- applying, by the UE, learning-based channel estimation (LCE) on the channel estimates using a neural network (NN) to generate refined channel estimates on the corresponding RS symbols;
- and
- performing, by the UE, time domain (TD) interpolation on the refined channel estimates to produce a channel estimate across symbols.
A method comprising:
- generating, by a user equipment (UE), channel estimates on reference signal (RS) symbols in a frequency domain (FD);
- converting, by the UE, the channel estimates to a time domain (TD) or a delay-Doppler domain (DD) to generate converted channel estimates;
- performing, by the UE, denoising on the converted channel estimates;
- applying, by the UE, learning-based channel estimation (LCE) on the denoised channel estimates to generate refined channel estimates on the RS symbols;
- converting the refined channel estimates to the FD;
- and
- performing, by the UE, TD interpolation on the refined channel estimates to produce a channel estimate across symbols.
A user equipment (UE) comprising:
- a processor;
- and
- a non-transitory computer readable storage medium storing instructions that, when executed, cause the processor to:
- perform frequency domain (FD) interpolation on wideband reference signal (RS) resource elements (REs) to generate channel estimates on corresponding RS symbols;
- apply learning-based channel estimation (LCE) on the channel estimates using a neural network (NN) to generate refined channel estimates on the corresponding RS symbols;
- and
- performing time domain (TD) interpolation on the refined channel estimates to produce a channel estimate across symbols.
Dependencies are inferred from the claim text.