GPU Accelerated High Capacity, AI-Ready 5G/6G Reference Design and Verification Methodology

The demands from wireless connectivity continue to grow, placing greater emphasis in designing efficient wireless networks and even greater emphasis on End-to-End (E2E) system verification. Next generation Telecom networks need to enable the proliferation of Artificial Intelligence (AI) in wireless communications, in addition to expanding traditional road map items (such as throughput, latency, user capacity, etc.). We provide a 5G network architecture reference design and present a methodology to evaluate and validate the 5G Stand Alone (SA) network. The methodology consists of three levels. First is the O-RAN fronthaul (FH) connection (called eCPRI), providing the ability to validate FH protocol functionality, server utilization, cell throughput, user capacity, etc. The second level is called RF Cabled and includes the impact of an O-RAN Radio Unit (O-RU) in an RF-controlled environment to evaluate performance. Lastly, over-the-air (OTA) is presented to introduce commercially available user equipment and mobility, as well as to experience real-world testing. We show how each level is used to include application-level testing to confirm the end-user experience.