Title: Optimised floating point arithmetic-based QR decomposition for wireless communication systems
Authors: Alahari Radhika; Satya Prasad Kodati; Kishan Rao Kalitkar
Addresses: Department of Electronics and Communication Engineering, JNT University, Kakinada, Andhra Pradesh, India ' Department of Electronics and Communication Engineering, Vignan's Foundation for Science, Technology and Research (Deemed to be University), Guntur, Andhra Pradesh, India ' Department of Electronics and Communication Engineering, Srinidhi Institute of Science and Technology, Hyderabad, Telangana, India
Abstract: In wireless communication systems, matrix inversion is the most common operation to provide a solution to the system of linear equations. Also, OMP compressive sensing is widely preferred for signal reconstruction at the receiver side. In both these processes, the accuracy and achievable performance rate of fixed-point arithmetic restrict its applicability in many real-time applications. In this paper, floating-point enabled systolic array implementation of iterative QR decomposition (QRD) based on the modified Gram-Schmidt (MGS) algorithm is proposed for matrix inversion. Here, the computational efficiency is improved using floating-point optimisation techniques and parallel systolic implementation which results in robust numerical stability. By exploiting the potential metrics of the systolic array, the proposed QRD architecture offered maximum throughput rate and inherent errorless FPU computation which help in improving the accuracy in matrix inversion.
Keywords: orthogonal matching pursuit; QR decomposition; QRD; Gram-Schmidt; floating point unit; wireless communication; compressive sensing; matrix inversion; signal detection.
DOI: 10.1504/IJUWBCS.2021.119136
International Journal of Ultra Wideband Communications and Systems, 2021 Vol.4 No.3/4, pp.134 - 138
Received: 24 Jul 2020
Accepted: 10 Dec 2020
Published online: 24 Nov 2021 *