
TeraSignal to Showcase CMIS-Based Link Training for Linear Optics and Copper Interconnects at OFC 2026
TeraSignal, a recognized leader in intelligent interconnect technology, today announced that it will present a groundbreaking demonstration of CMIS-based link training for linear optics and copper interconnects at the Optical Fiber Communication Conference (OFC) 2026. This demonstration underscores a major advancement in deploying next-generation linear interfaces designed to meet the growing demands of high-performance computing, AI data centers, and advanced optical networking. By embedding TeraSignal’s TSLink™ link training algorithm within the Common Management Interface Specification (CMIS) framework, the company is enabling automatic configuration and optimization of each optical and copper link, maximizing signal-to-noise ratio (SNR) and link margin in real-world conditions.
As modern data centers and AI infrastructure evolve, the need for highly reliable, low-latency connectivity is greater than ever. Linear interconnect architectures—including Linear Pluggable Optics (LPO), Near-Packaged Optics (NPO), Co-Packaged Optics (CPO), and advanced Active Copper Cables (ACC)—have become a key part of this transformation. However, operating these systems without digital signal processing (DSP) requires precision link configuration, dynamic equalization, and continuous tuning of operating parameters. TeraSignal’s CMIS-enabled TSLink technology meets this need by allowing both host ASICs and pluggable modules to dynamically adjust link parameters, ensuring optimal performance under any channel conditions.
The demonstration at OFC 2026 will allow attendees to witness TSLink-enabled CMIS link training in real-time. By combining hardware and software intelligence, the platform automatically configures link parameters between host ASICs and QSFP/OSFP modules, optimizing each connection for the highest possible SNR. This ensures robust link margin and stability across both optical and copper interconnects, simplifying deployment for manufacturers and operators while improving reliability and efficiency across linear interconnect systems.
“Next-generation linear interconnects require adaptive, standards-aligned configuration to achieve the performance and reliability modern networks demand,” said Dr. Armond Hairapetian, CEO of TeraSignal. “By integrating TSLink link training into the CMIS framework and supporting emerging standards such as IEEE’s Inline Link Training / Automatic Path Startup (ILT/APSU), we enable automatic optimization that both simplifies deployment and ensures peak performance for optical and copper links alike. This demonstration at OFC 2026 highlights the practical benefits of intelligent link training and how it can accelerate the adoption of high-speed interconnects in AI and high-performance computing environments.”
TeraSignal’s TSLink technology supports the CMIS-LT protocol, as defined by the Optical Internetworking Forum (OIF), and the company actively contributes to OIF technical working groups to advance interoperable link training standards. With a focus on future-proofing, TeraSignal plans to support all emerging link training protocols, including IEEE’s ILT/APSU specification, providing customers with standards-aligned solutions that can evolve alongside the linear interconnect ecosystem. This proactive approach ensures that as the industry continues to transition to linear architectures, operators and manufacturers can rely on interoperable, high-performance solutions.
The OFC 2026 demonstration will include:
- Live CMIS-based link training: Attendees will see real-time link configuration between host ASICs and pluggable QSFP/OSFP modules.
- TSLink automatic optimization: Each link dynamically adjusts parameters to maximize SNR and link margin without requiring DSP intervention.
- Protocol compliance: The demonstration adheres to OIF CMIS-LT standards and is ready for IEEE ILT/APSU, ensuring interoperability and future readiness.
- Performance across mediums: Both optical and copper interconnects benefit from improved stability, reliability, and operational predictability.
- Operational insights: Integrated diagnostics and analytics provide deep visibility into link behavior, allowing engineers to monitor, troubleshoot, and optimize connectivity across complex deployments.
This demonstration is especially critical for AI data centers, high-performance computing clusters, and advanced networking environments where minimizing latency and maximizing link reliability directly impact computational efficiency. By emulating real-world operational conditions, TeraSignal allows operators and manufacturers to validate system performance prior to large-scale deployment, reducing risk and enabling faster, more confident rollouts.
The transition to linear interconnects presents unique challenges. Unlike traditional network designs, linear optical and copper links rely on precise configuration to maintain integrity across multiple ports and channels. Factors such as crosstalk, signal attenuation, and microbursts can significantly impact performance if not managed correctly. TSLink addresses these challenges by continuously monitoring link performance and dynamically adjusting parameters, ensuring optimal operation for each connection.
TeraSignal has long been at the forefront of high-speed interconnect innovation. Its portfolio of intelligent optical and copper interconnect solutions focuses on increasing power efficiency, reducing latency, lowering bit-error rates, and providing advanced diagnostics. By integrating adaptive link training with CMIS, TeraSignal enables seamless interoperability between host ASICs and pluggable modules, ensuring consistent performance and simplified deployment across both legacy and emerging data center architectures.
“OFC 2026 provides a perfect stage to showcase how intelligent link training can transform the deployment and management of linear interconnects,” said Dr. Hairapetian. “We are excited to show how TSLink-enabled CMIS link training not only improves performance but also simplifies operations for network designers, manufacturers, and operators in demanding AI and HPC environments.
Source link: https://www.businesswire.com




