US 12,279,208 B2
Machine learning (ML)-based dynamic demodulator parameter selection
Jacob Pick, Beit Zayit (IL); Assaf Touboul, Netanya (IL); Shay Landis, Hod Hasharon (IL); Alexei Yurievitch Gorokhov, San Diego, CA (US); Hari Sankar, San Diego, CA (US); and Peer Berger, Hod Hasharon (IL)
Assigned to QUALCOMM Incorporated, San Diego, CA (US)
Filed by QUALCOMM Incorporated, San Diego, CA (US)
Filed on Jun. 23, 2022, as Appl. No. 17/848,295.
Prior Publication US 2023/0422175 A1, Dec. 28, 2023
Int. Cl. H04M 3/00 (2024.01); G06N 3/02 (2006.01); H04L 12/66 (2006.01); H04M 5/00 (2006.01); H04W 52/02 (2009.01)
CPC H04W 52/0261 (2013.01) [G06N 3/02 (2013.01)] 28 Claims
OG exemplary drawing
 
1. A method of wireless communication by a receiving device, comprising:
predicting, with an artificial neural network, at each data block of a set of data blocks, a least complex set of demodulator parameters that will achieve a goal, based on features of a data block expected to be received; and
dynamically selecting the least complex set of demodulator parameters, from a plurality of sets of demodulator parameters, based on the features of the data block expected to be received, the selecting occurring to prevent degradation of demodulation performance for each data block with the selected set of demodulator parameters for the data block, with respect to a more complex set of demodulator parameters, the plurality of sets of demodulation parameters defining at least one of a search space, a processing sequence for a quantity of layers, or a combination of the search space and processing sequence.