Keynote II
Title: Building AI-RAN and 6G on Open Source: From dApps and E3 to Real-Time Inline AI in the PHY
Speaker: Dr. Tim O'SheaCTO, DeepSig Inc.Member of Governing Board, OCUDU Ecosystem Foundation
Abstract:
Software and AI now move faster than the standards processes that traditionally defined how radio access networks are built. This keynote gives an update on building AI-RAN and the road to 6G on open source. We describe OCUDU's open source dApp and E3 architecture for AI-RAN applications, and our extension of dApps into real-time inline functions that host AI modules in the PHY pipeline with low latency and high efficiency across hardware platforms and vendors, including recent work in OCUDU's AI-RAN and Hardware-Acceleration working groups and the new CUDA-accelerated OCUDU capability.
We then show how DeepSig's AI-driven spectrum sensing, AI-driven receivers and AI-native air interface are prototyped, deployed and commercialized on OCUDU and open source cores such as free5GC, and close with the shifting landscape of industry consensus, where open source increasingly complements 3GPP and the O-RAN Alliance as software and AI accelerate.
Biography:
Dr. Tim O'Shea is the CTO at DeepSig Inc, focused building and deploying leading AI-RAN & RF-ML solutions, including the world's first AI-Native OpenRAN in a macro-network with Viettel and industry leading AI-Driven Spectrum Sensing solution. He is actively involved in driving AI-RAN on the OCUDU Ecosystem Board & TSC, the AI-RAN Alliance Board & TSC, 3GPP 6G and AI-Native Air Interface standardization, and with IEEE's Machine Learning for Communications initiatives. He is an IEEE Fellow and FCC TAC member and was previously Research Assistant at Virginia Tech, and at Hawkeye 360, Federated Wireless, GNU Radio, Cisco Systems, and the Department of Defense.