US 11,693,654 B2
Embedded GPU-based wideband parallel channelized receiving method
Guojun Li, Chongqing (CN); Feixiang Tian, Chongqing (CN); Changrong Ye, Chongqing (CN); Zunli Wang, Chongqing (CN); and Yiping Luo, Chongqing (CN)
Assigned to CHONGQING UNIVERSITY OF POSTS AND TELECOMMUNICATIONS, Chongqing (CN)
Appl. No. 17/919,984
Filed by CHONGQING UNIVERSITY OF POSTS AND TELECOMMUNICATIONS, Chongqing (CN)
PCT Filed Feb. 1, 2021, PCT No. PCT/CN2021/074646
§ 371(c)(1), (2) Date Oct. 19, 2022,
PCT Pub. No. WO2021/253840, PCT Pub. Date Dec. 23, 2021.
Claims priority of application No. 202010548596.4 (CN), filed on Jun. 16, 2020.
Prior Publication US 2023/0161586 A1, May 25, 2023
Int. Cl. H04B 1/06 (2006.01); G06F 8/76 (2018.01); G06F 17/16 (2006.01); H03H 17/02 (2006.01)
CPC G06F 8/76 (2013.01) [G06F 17/16 (2013.01); H03H 17/0266 (2013.01); H03H 17/0273 (2013.01); H04B 1/06 (2013.01)] 7 Claims
OG exemplary drawing
 
1. An embedded GPU-based wideband parallel channelized receiving method, comprising:
constructing an OpenCL platform, wherein the OpenCL platform comprises a host computer and a plurality of OpenCL devices, each of the plurality of OpenCL devices comprises a plurality of computing units, each of the plurality of computing units comprises a plurality of work groups and a plurality of processing elements, and each of the plurality of processing elements serves as a work item;
decimating a wideband signal read in the OpenCL platform at an interval indicated by the number of channels, wherein a two-dimensional matrix is formed by decimating one-dimensional data that is originally read, and the number of rows of the two-dimensional matrix is equal to the number of channels;
assigning data in each row to one of the plurality of work groups for processing, wherein input data in each of the plurality of work groups is multiplied by a factor (−1)m, and m indicates a serial number of a column where the data is located;
filtering data on each of the channels based on a coefficient of a polyphase filter on a branch;
multiplying the filtered data by a factor

OG Complex Work Unit Math
wherein p indicates a channel number of the branch and D indicates the number of channels; and
performing an FFT on the formed two-dimensional matrix by columns to obtain data outputted from each of the channels.