AI-Native applied to vRAN baseband provides better power efficiency, better spectral efficiency, improved resilience, reduced component costs, and numerous additional benefits.
The most immediate impact from AI-Native baseband algorithm design will be realized from transparent insertion of new AI software modules into existing 5G systems. Especially within the context of OpenRAN and vRAN, drop-in software functionality replacements offer the ability to leverage existing virtualized infrastructure while delivering improved performance and capability. DeepSig’s patented ML vRAN product, OmniPHY® 5G, provides our first AI-Native software module that operates in the 5G O-DU to improve MIMO Uplink performance which improves link margin and reduces power consumption. This can be coupled with a RIC xAPP to monitor and continuously improve performance over time.
OmniPHY® 5G will further continue to improve mMIMO performance in the coming months and will significantly enhance the resource scheduling coupled with L1 information.
Below we show an example of OmniPHY® 5G running in our 5G-AI Lab over the air using a commercial handset and providing enhanced SINR and link capacity.
AI & ML are being further explored within the 3GPP Rel 18 - 5G Advanced standardization process, as well as the Next-G and other 6G-centric initiatives looking at key candidate technologies for the next generation of wireless. AI-Native design of core communications system functions is continuing to expand across the RAN stack in many of these candidate designs as the industry understands how to best leverage and standardize AI/ML in new functions. As the pioneers of the channel-autoencoder, the first AI-Native communications system approach, we are continually pushing the boundaries of where AI can rethink core wireless functionalities and will be sharing key technologies over the coming years that we believe will help form a foundation for an intelligent and efficient next-generation air interface.
AI-Native design of communications systems outside of the RAN standards process, is one key area where we are focused on building out and maturing these candidate technologies. One of the key area for this is in resilient emergency communications systems, where physical layer links must operate in the face of a wide range of impairments and interference by adapting and responding intelligently to their environment.