Method for AI-based channel estimation with varying PRB set size
One example method includes receiving, by a pre-processing unit, a rough channel estimated array, splitting, by the pre-processing unit, the rough channel estimated array into a group of smaller arrays that each have a size that is smaller than a size of the rough channel estimated array, providing, by the pre-processing unit, the smaller arrays to a single NN (neural network), processing, by the NN, the smaller arrays to generate respective refined channel estimation outputs for each of the smaller arrays, and combining, by a post-processing unit, the smaller arrays having the respective refined channel estimation outputs to generate an output array with a size that is the same as the size of the rough channel estimated array.
1 . A method for channel estimation, comprising:
receiving, by a pre-processing unit, a rough channel estimated array;
splitting, by the pre-processing unit, the rough channel estimated array into a group of smaller arrays that each have a size that is smaller than a size of the rough channel estimated array;
providing, by the pre-processing unit, the smaller arrays to a single neural network (NN);
processing, by the NN, the smaller arrays to generate respective refined channel estimation outputs for each of the smaller arrays; and
combining, by a post-processing unit, the smaller arrays having the respective refined channel estimation outputs to generate an output array with a size that is the same as the size of the rough channel estimated array.
2 . The method as recited in claim 1 , wherein the rough channel estimated array was generated by a least squares (LS) channel estimation and interpolation process.
3 . The method as recited in claim 1 , wherein the rough channel estimated array comprises a physical resource block (PRB), and the output array comprises a PRB.
4 . The method as recited in claim 3 , wherein the respective sizes of the smaller arrays are independent of a size of the PRB, and the smaller arrays are then processed by the NN in a serial manner.
5 . The method as recited in claim 3 , wherein a size of the NN is independent of a size of the PRB of the rough channel estimated array.
6 . The method as recited in claim 1 , wherein the rough channel estimated array comprises a mini-slot of a physical resource block (PRB), and the output array comprises a mini-slot of a PRB.
7 . The method as recited in claim 6 , wherein the respective sizes of the smaller arrays are independent of a size of the mini-slot of the PRB of the rough channel estimated array, and the smaller arrays are then processed by the NN in a serial manner.
8 . The method as recited in claim 6 , wherein a size of the NN is independent of a size of the mini-slot of the PRB of the rough channel estimated array.
9 . The method as recited in claim 1 , wherein the smaller arrays produced by the splitting are overlapping arrays.
10 . The method as recited in claim 1 , wherein the NN is the only NN used to process the smaller arrays, and the NN is operable with rough channel estimated arrays of various different sizes.
11 . A non-transitory storage medium having stored therein instructions that are executable by one or more hardware processors to perform operations comprising:
receiving, by a pre-processing unit, a rough channel estimated array;
splitting, by the pre-processing unit, the rough channel estimated array into a group of smaller arrays that each have a size that is smaller than a size of the rough channel estimated array;
providing, by the pre-processing unit, the smaller arrays to a single neural network (NN);
processing, by the NN, the smaller arrays to generate respective refined channel estimation outputs for each of the smaller arrays; and
combining, by a post-processing unit, the smaller arrays having the respective refined channel estimation outputs to generate an output array with a size that is the same as the size of the rough channel estimated array.
12 . The non-transitory storage medium as recited in claim 11 , wherein the rough channel estimated array was generated by a least squares (LS) channel estimation and interpolation process.
13 . The non-transitory storage medium as recited in claim 11 , wherein the rough channel estimated array comprises a physical resource block (PRB), and the output array comprises a PRB.
14 . The non-transitory storage medium as recited in claim 13 , wherein the respective sizes of the smaller arrays are independent of a size of the PRB, and the smaller arrays are then processed by the NN in a serial manner.
15 . The non-transitory storage medium as recited in claim 13 , wherein a size of the NN is independent of a size of the PRB of the rough channel estimated array.
16 . The non-transitory storage medium as recited in claim 11 , wherein the rough channel estimated array comprises a mini-slot of a physical resource block (PRB), and the output array comprises a mini-slot of a PRB.
17 . The non-transitory storage medium as recited in claim 16 , wherein the respective sizes of the smaller arrays are independent of a size of the mini-slot of the PRB of the rough channel estimated array, and the smaller arrays are then processed by the NN in a serial manner.
18 . The non-transitory storage medium as recited in claim 16 , wherein a size of the NN is independent of a size of the mini-slot of the PRB of the rough channel estimated array.
19 . The non-transitory storage medium as recited in claim 11 , wherein the smaller arrays produced by the splitting are overlapping arrays.
20 . The non-transitory storage medium as recited in claim 11 , wherein the NN is the only NN used to process the smaller arrays, and the NN is operable with rough channel estimated arrays of various different sizes.