Article: Comparative study of kernel algorithms on SIMD vector processor for 5G massive MIMO Journal: International Journal of Advanced Intelligence Paradigms (IJAIP) 2024 Vol.29 No.2/3 pp.111 - 121 Abstract: Currently world is moving towards achieving Gigabit data rates via 5G mobile revolution. Massive multi-in-multi-out (MIMO) is one of the key enabler and recently lot of interest is evinced in this area. The efficiency of the algorithms to estimate and detect the channel plays a very crucial role for the success of Massive MIMO. The existing algorithms of LTE-A for this purpose are not efficient in terms of power consumption and lower latency, which is one of the foremost necessity of 5G communications. The biggest hurdle to achieve the ultra-low latency in 5G massive MIMO is - a very huge number of computations required for the matrix inversion while performing channel estimation and detection. In this paper, a comparative study has been done for two parallel processing schemes: Gauss-Jordan elimination and LU decomposition kernel algorithms on a single instruction multiple data (SIMD) stream vector processor for the realisation of matrix inversion with optimum latency, which is the pre-requisite for the 5G channel estimation and detection. In this paper both matrix inversion algorithms Gauss Jordan and LU decomposition are analysed and LU decomposition provides the required level of reduction of computational operations, which translates low latency and less battery power consumption. Inderscience Publishers - linking academia, business and industry through research

Title: Comparative study of kernel algorithms on SIMD vector processor for 5G massive MIMO

Authors: Ravi Sekhar Yarrabothu; Pitchaiah Telagathoti

Addresses: Department of Electronics and Communication Engineering, VFSTR (Deemed to be University), Vadlamudi, Guntur – 522213, India ' Department of Electronics and Communication Engineering, VFSTR (Deemed to be University), Vadlamudi, Guntur – 522213, India

Abstract: Currently world is moving towards achieving Gigabit data rates via 5G mobile revolution. Massive multi-in-multi-out (MIMO) is one of the key enabler and recently lot of interest is evinced in this area. The efficiency of the algorithms to estimate and detect the channel plays a very crucial role for the success of Massive MIMO. The existing algorithms of LTE-A for this purpose are not efficient in terms of power consumption and lower latency, which is one of the foremost necessity of 5G communications. The biggest hurdle to achieve the ultra-low latency in 5G massive MIMO is - a very huge number of computations required for the matrix inversion while performing channel estimation and detection. In this paper, a comparative study has been done for two parallel processing schemes: Gauss-Jordan elimination and LU decomposition kernel algorithms on a single instruction multiple data (SIMD) stream vector processor for the realisation of matrix inversion with optimum latency, which is the pre-requisite for the 5G channel estimation and detection. In this paper both matrix inversion algorithms Gauss Jordan and LU decomposition are analysed and LU decomposition provides the required level of reduction of computational operations, which translates low latency and less battery power consumption.

Keywords: massive MIMO; SIMD; single instruction multiple data; 5G; DMRS; demodulation reference signal; SRS; sounding reference signal; long term evolution - advanced.

DOI: 10.1504/IJAIP.2024.142662

International Journal of Advanced Intelligence Paradigms, 2024 Vol.29 No.2/3, pp.111 - 121

Received: 11 Apr 2019
Accepted: 22 May 2019
Published online: 15 Nov 2024 *

Full-text access for editors Full-text access for subscribers Purchase this article Comment on this article

Title: Comparative study of kernel algorithms on SIMD vector processor for 5G massive MIMO

Keep up-to-date