DeepSig, Inc.

Excited to see NSC member DeepSig, Inc.’s leadership as part of the OCUDU Ecosystem Foundation in developing an open source platform to support AI-native RAN. Read DeepSig’s latest blog for an update on their work: https://lnkd.in/e9YbJg_M

We at DeepSig, Inc. have just published the details behind the CUDA-accelerated OCUDU L1. It is in the open GitLab today, with the dApp/E3 framework close behind. → Several-fold to ~10x speedups on the heaviest shared-channel blocks; ~100x on fronthaul compression → BLER within 0.06 dB of the CPU implementation. Zero accuracy traded for throughput → One codebase from a laptop CPU to a GB10 or GH200 rack → A neural equalizer, AI scheduler, and spectrum/ISAC sensor running concurrently over the air under live traffic AI has always been kept out of the RAN's real-time core. This is what it looks like when that changes, using open, portable interfaces any vendor can build against. I am proud of what the team and the OCUDU community shipped here! The next AI-RAN Working Group call is August 5 if you want in. https://lnkd.in/egzjBWVf #AIRAN #OpenRAN #AI #6G

What does it take to make AI a truly native part of the RAN? In a new blog OCUCU EF member, DeepSig, Inc., explores how the OCUDU community is building an open platform where AI applications can run directly within the radio—not through proprietary integrations, but through shared, vendor-neutral interfaces. The post highlights: - The CUDA-accelerated OCUDU L1, now available in the open repositories - The forthcoming dApp/E3 framework for inline, real-time and observational AI applications - Open interfaces designed to support portable, interoperable AI-RAN solutions - Collaboration across the OCUDU Hardware Acceleration and AI-RAN Working Groups DeepSig is also bringing solutions including OmniPHY, OmniSIG and Axon onto these common interfaces, while contributing open reference designs that can help the broader ecosystem build, test and deploy AI-native RAN applications. This is the promise of OCUDU: creating an open, neutrally governed venue where organizations can collaborate on the infrastructure needed to move AI-RAN innovation from research to real-world deployment. Read the full DeepSig blog: https://lnkd.in/e9YbJg_M Interested in contributing? Join the OCUDU community and help build the open foundation for AI-native RAN.

AI-native RAN needs more than AI models—it requires an open platform to run and embed them efficiently. We are excited to contribute to that foundation with the release of the hardware-accelerated OCUDU L1, alongside the AI-RAN dApp and E3 platform of OCUDU (designed to run flexibly across both CPU and GPU). Together, they establish an open, interoperable foundation for an AI-Native RAN. Fostering this kind of vendor-neutral ecosystem is critical to advancing national wireless priorities, including those set forth in recent National Telecommunications and Information Administration (NTIA) requirements. Here is why this matters: Most AI in the RAN operates where latency is forgiving, like sensing or reporting, or slowly tuning policy. But to truly transform performance, AI-for-RAN applications like learned receivers and dynamic schedulers must sit directly inside the real-time processing loop. Through the collective efforts of the OCUDU working groups, we are helping close that gap in the open: - WG-0001 (HW-ACCEL) released a hardware-accelerated OCUDU L1, achieving significant latency and throughput performance gains across the PHY while maintaining algorithmic parity. - WG-0002 (AI-RAN) is rolling out the dApp/E3 framework, demonstrating real-time dApps including receiver, scheduler, and sensing running together over the air. DeepSig is grateful to help lay this groundwork alongside our peers — seeking a multi-vendor interoperable dApp ecosystem which brings all of the best AI-RAN ideas into reality and production. Moving forward, our own products (OmniSIG, OmniPHY, and Axon) will be available on these exact same interfaces as fully interoperable dApps. In the future these dApp APIs could be adopted by multiple RAN stacks, enabling dApp portability across an even broader vendor ecosystem. We invested in these core interfaces because the community needs them to make the most out of AI-RAN, and they are far more valuable shared openly with ecosystem consensus and interoperability to accelerate our whole community. The real promise of building in the open is what the industry builds next. We invite you to join the AI-RAN Working Group’s next call on 5 August. New participants, contributions, and perspectives are all welcome as we refine these interfaces together. Learn more: https://lnkd.in/e9YbJg_M #AIRAN #OCUDU #5G #6G #dApps

A milestone for the future of Wi-Fi, and a proud moment for our team. 👏 This month, the IEEE 802.11 Working Group voted to approve the formation of the WLAN Intelligent Networking Study Group (WIN SG), the first formal step toward the next generation of Wi-Fi. The motion passed with overwhelming support, following more than 20 hours of technical presentations from over 20 companies/entities over seven months. This incubation period was guided by DeepSig’s VP of Technical Standards, Jim Lansford, who is the Chair of the IEEE 802.11 Wireless Next Generation Standing Committee. What makes this generation different is that AI is at the center of it from the start. The study group will investigate technology to improve throughput, latency predictability, resilience, and adaptation to emerging application requirements, including AI traffic and services. In addition to his role as Chair of the WNG committee, Dr. Lansford contributed to this effort by presenting how the AI-native techniques we have developed for wireless communications can be applied to Wi-Fi to deliver similar gains in resilience and robustness. The study group is expected to conclude its work next year, with specification development in a Task Group to follow and an eventual Wi-Fi Alliance certification anticipated around 2031. It is early, but the direction is clear: the next generation of Wi-Fi will be designed around AI and machine learning, and that foundational technology is exactly what DeepSig has been building for a decade. #WirelessInnovation #AIinWireless #ML IEEE802

OmniSIG 4.0 is built for missions that change in real time. Our team spent the day out in the field testing the latest version. With swappable AI models, operators can quickly switch between detecting drones, aircraft, and boats without swapping out the system. At the same time, OmniSIG provides the information needed to understand what's happening across the spectrum, including signal classification, timestamp, center frequency, upper and lower frequency bounds, and direction of arrival. All of it happens in one piece of software. Learn more https://lnkd.in/eh_y8dT2 #RFML #SignalIntelligence #SpectrumSensing #SpectrumAwareness #AI #RFSensing

DeepSig, Inc., the acknowledged pioneer in the realm of AI-based wireless communications technology, has now released its latest OmniSIG version - #OmniSIG 4.0. The latest version of the software is a game changer not only because it is the most radical change in terms of user interface design that has been done since the first market release but also because it offers tremendous enhancements in the area of AI performance, making spectrum situational awareness both faster and easier to achieve than ever before. #communications #DeepSig #News #OmniSIG4.0 #Technology

OmniSIG 4.0 represents a significant advancement in AI-powered spectrum awareness. Rather than relying on handcrafted signal processing techniques, the platform applies deep learning optimized for the radio-frequency domain to detect and classify signals across a wide range of operating conditions. The new release also adds support for eight additional hardware receivers, expanding deployment options from laboratory environments to edge devices, field operations, and autonomous systems. Learn more: https://lnkd.in/e7mc6taP #RFML #SpectrumIntelligence #SpectrumAwareness #AI

OmniSIG 4.0 is here. 🚀 Today, we announced a major update to our flagship RF signal detection and analysis platform, including its first full interface redesign since the original launch. "Modern spectrum operations require both speed and clarity," said James Shea, DeepSig CEO. "OmniSIG 4.0 combines advances in AI with a completely redesigned user experience to help operators detect, classify, and understand signals faster." Available now to existing customers. Read the full announcement: https://lnkd.in/e7mc6taP Learn more about our solutions: https://lnkd.in/e6KykZEB #AI #SpectrumAwareness #RF #SignalDetection #RFML

Today marks 250 years of the American story. It's a moment to reflect on the values that continue to move our country forward: curiosity, resilience, innovation, and a willingness to tackle hard problems together. At DeepSig, we're proud to be part of a community of builders, thinkers, and innovators working to shape what comes next. Progress has always been driven by those who imagine a better future, and then get to work making it real. As we celebrate America's 250th birthday, we're grateful for the generations who have contributed to that legacy and for the opportunities ahead. Happy Independence Day. 🇺🇸