Müller R (2021)
Publication Type: Conference contribution
Publication year: 2021
Publisher: IEEE Computer Society
Book Volume: 2021-July
Pages Range: 186-190
Conference Proceedings Title: IEEE Workshop on Statistical Signal Processing Proceedings
Event location: Virtual, Rio de Janeiro, BRA
ISBN: 9781728157672
DOI: 10.1109/SSP49050.2021.9513777
We utilize the new concept of linear computation coding to reduce the computational effort in digital beamforming. For that purpose the beamforming matrix is decomposed into the product of two or more matrices which are only composed of zeros and signed powers of two. As shown in [1], this allows for a multifold reduction of the required number of additions (multiplications are treated as multiple additions and bit-shifts) per entry of the beamforming matrix.We also provide an approximate analysis of the performance of linear computation coding for matrices with independent identically distributed matrices.For a typical implementation in beamforming, the proposed algorithm reduces the number of required addition units several times. To achieve the accuracy of 4-bit signed integer arithmetic for 1k-vectors, no multipliers and only half an adder per matrix entry is needed.
APA:
Müller, R. (2021). Energy-Efficient Digital Beamforming by Means of Linear Computation Coding. In IEEE Workshop on Statistical Signal Processing Proceedings (pp. 186-190). Virtual, Rio de Janeiro, BRA: IEEE Computer Society.
MLA:
Müller, Ralf. "Energy-Efficient Digital Beamforming by Means of Linear Computation Coding." Proceedings of the 21st IEEE Statistical Signal Processing Workshop, SSP 2021, Virtual, Rio de Janeiro, BRA IEEE Computer Society, 2021. 186-190.
BibTeX: Download