For 5G/B5G, a software-defined RAN requires high reliability, high performanceflexibility that software-defined physical layer is key to ensure the performancequality of multi-service connections. With radio function splitdistributed CPUGPU cooperation, we can ensure software-defined RAN working properly for service-aware networking where our  trial shows stable DLUL over 

1. Distributed CPUGPGPU Cooperation with Radio Function Split 
2. UL  800 Mbps via NIVIDA GPU
3. Software-Defined Physical Layer on Shared-Memory Disaggregated Infrastructure

With radio function splitdistributed computation, software-defined RAN which shares computing resource with edge clouddata center, is well-fitted Taiwan’s echo system to make dynamic management, latency-aware service, load balancefail recovery well for industry 4.0