Abstract

Open RAN and 5G are two key technologies designed to qualitatively improve network infrastructure and provide greater flex-ibility and efficiency to mobile operators and users. 5G creates new capabilities for high-speed Internet, Internet of Things, telemedi-cine and many other applications, while Open RAN enables open and standardized network architectures, which reduces cost and risk for operators and promotes innovations. Given the growing number of users and data volumes, the purely software implementa-tion of certain functions of the 5G protocol, and especially computationally complex ones, requires significant computer resources and energy.These, for example, arelow-density parity-check (LDPC)coding,FFTandiFFT algorithms on physical (PHY)layer, and NEA and NIA security algorithms on Packet Data Convergence Protocol (PDCP)layer. Therefore, one of the activity areas in the development of means for 5G systems isthe hardware acceleration of such functions execution, which provides the possibility of processing large volumes of data in real time and with high efficiency. The high-performance hardware basis for implementing these functions today is field-programmable gate array(FPGA)integrated circuits.Along with this, the efficiency of the 5G protocol stack functions hardware acceleration depends significantly on the size of the data packets transmitted to the hardware accelerator. As ex-perience shows, for certain types of architecture of computer systems with accelerators, the acceleration value can take even a nega-tive value. This necessitates the search for alternative architectural solutions for the implementation of such systems.In this article the approaches for hardware acceleration using reconfigurable FPGA-based computing components are explored, their comparative analysis is performed, and architectural alternatives are evaluated for the implementation of a computing platform to perform the functions ofthe 5G protocol stack with hardware acceleration of PHY and medium access control(MAC)layers functions.

Full Text
Published version (Free)

Talk to us

Join us for a 30 min session where you can share your feedback and ask us any queries you have

Schedule a call