MUMBAI, India, Jan. 23 -- Intellectual Property India has published a patent application (202511126839 A) filed by Chandigarh University, Mohali, Punjab, on Dec. 15, 2025, for 'ai-powered load balancer system using telemetry asic for predictive cloud traffic.'
Inventor(s) include Shivansh Mishra; Vivek Kumar; Piyush Chaudhary; Parul Sood; and Dr. Parveen Bajaj.
The application for the patent was published on Jan. 23, under issue no. 04/2026.
According to the abstract released by the Intellectual Property India: "The present invention relates to an AI-powered load balancer system for predictive cloud traffic designed to achieves ultra-low-latency traffic management, robust adaptability to dynamic demand fluctuations, and energy-efficient operation. Within a single chassis, the appliance consolidates a Control CPU, an AI Accelerator (GPU/TPU/FPGA), a SmartNIC/DPU, and a specialized Telemetry ASIC Co-Processor. Real-time telemetry streams from servers and network infrastructure are captured by the SmartNIC and processed at line rate by the Telemetry ASIC Co-Processor, transforming raw performance metrics into structured features suitable for predictive analysis. The system utilizes a reinforcement-learning and LSTM-based algorithms, the AI Accelerator anticipates workload shifts and potential congestion before they occur. These predictive insights are then compiled by the Control CPU into optimized routing policies, which the SmartNIC/DPU enforces directly at the hardware level, enabling sub-millisecond packet distribution and ensuring seamless, intelligent traffic flow across the data center."
Disclaimer: Curated by HT Syndication.
