Channel estimation with varying numbers of transmit layers

A method, comprising:

• transforming, by a system comprising at least one processor, a first array into second arrays, wherein the first array comprises a rough channel estimate over a first number of dimensions, wherein respective second arrays of the second arrays have a second number of dimensions, and wherein the second number is less than the first number;
• processing, by the system, the respective second arrays with a neural network that is configured to process arrays having the second number of dimensions, to produce third arrays;
• combining, by the system, the third arrays into a fourth array having the first number of dimensions, wherein the fourth array indicates a channel estimate;
• and
• using, by the system, the fourth array for broadband cellular communications with at least one user equipment.

A non-transitory computer-readable medium comprising instructions that, in response to execution, cause a system comprising at least one processor to perform operations, comprising:

• transforming a rough channel estimate array that comprises a first number of dimensions and a dimension that indicates a number of transmit layers into first arrays having a second number of dimensions, wherein respective first arrays of the first arrays omit an indication of the number of transmit layers;
• inputting the first arrays to a neural network that is configured to accept, as input, arrays having the second number of dimensions, the inputting resulting in an output of second arrays;
• transforming the second arrays into a third array that indicates a channel estimation and that has the first number of dimensions;
• and
• using the channel estimation as indicated by the third array for multiple-input multiple-output cellular communications with a user equipment.