analiza aiqbench
This research demonstrates a closed-loop agentic workflow where an LLM agent directly implements and optimizes physical-layer algorithms for 6G baseband silicon, bridging the gap between theoretical discovery and hardware constraints. By validating against real chip metrics rather than GPU simulations, the system discovers efficient kernel structures that blind searches miss, proving that AI agents can navigate the specific cost structures of embedded processors. This marks a significant shift toward autonomous hardware-aware design flows essential for accelerating 6G development.
Toward Agentic AI for the 6G Physical Layer: From Algorithm Discovery to Verified Baseband Silicon
arXiv:2610.27797v1 Announce Type: new Abstract: 6G will be designed faster than it can be built: AI-driven search already proposes physical-layer algorithms at machine speed, while carrying a single one onto embedded baseband silicon remains specialist work. That silicon imposes hard memory and numeric limits, but also exposes cost structures a GPU lacks, and a design flow that never runs on the target sees neither. We present an agentic workflow in which an LLM coding agent implements algorithms on a commercial baseband SoC, optimizes them down to agent-written vector assembly, and validates them on silicon. The agent is guided by a knowledge base of measured platform facts, checked at every step against evaluators it is instructed never to edit, and metered per phase. We showcase the workflow on an AI-discovered orthogonal time frequency space (OTFS) detector portfolio on the open 6G4All platform. It yields twelve kernels that run cycle-exactly on silicon, their optimized versions at 18k-405k cycles (30-660 $\mu$s) per frame. It also yields a new detector shaped by the chip's measured costs that needs 13.9% fewer cycles than the port of the discovered detector while matching its error rate, with a measured edge our evaluation cannot statistically resolve, a structure that a hardware-blind search objective had no reason to find. What is not specific to OTFS is the workflow: an LLM agent exploring an implementation space under real hardware limits, with the cost of every choice measured rather than assumed.
Abstract:6G will be designed faster than it can be built: AI-driven search already proposes physical-layer algorithms at machine speed, while carrying a single one onto embedded baseband silicon remains specialist work. That silicon imposes hard memory and numeric limits, but also exposes cost structures a GPU lacks, and a design flow that never runs on the target sees neither. We present an agentic workflow in which an LLM coding agent implements algorithms on a commercial baseband SoC, optimizes them down to agent-written vector assembly, and validates them on silicon. The agent is guided by a knowledge base of measured platform facts, checked at every step against evaluators it is instructed never to edit, and metered per phase. We showcase the workflow on an AI-discovered orthogonal time frequency space (OTFS) detector portfolio on the open 6G4All platform. It yields twelve kernels that run cycle-exactly on silicon, their optimized versions at 18k-405k cycles (30-660 $\mu$s) per frame. It also yields a new detector shaped by the chip's measured costs that needs 13.9% fewer cycles than the port of the discovered detector while matching its error rate, with a measured edge our evaluation cannot statistically resolve, a structure that a hardware-blind search objective had no reason to find. What is not specific to OTFS is the workflow: an LLM agent exploring an implementation space under real hardware limits, with the cost of every choice measured rather than assumed.
7 pages, 3 figures, 5 tables. Accepted at the 2026 IEEE Globecom Workshops (GC Wkshps), 2nd Workshop on Agentic AI Empowered Wireless Communications and Networking, Macau, China, December 2026. (c) 2026 IEEE
Artificial Intelligence (cs.AI); Networking and Internet Architecture (cs.NI); Signal Processing (eess.SP)
arXiv:2610.27797 [cs.AI]
(or arXiv:2610.27797v1 [cs.AI] for this version)
https://doi.org/10.48550/arXiv.2610.27797 arXiv-issued DOI via DataCite (pending registration)
Text integral sub licența deschisă a publicației · 293 cuvinte
Subiecte conexe
ERBench: A Benchmark and Testsuite for Equation Discovery Algorithms
arXiv · 9 iun. 2026
Extending Desbordante with Probabilistic Functional Dependency Discovery Support
arXiv · 28 iul. 2026
Cross-validating causal discovery via Leave-One-Variable-Out
arXiv · 25 aug. 2026
MLEvolve: A Self-Evolving Framework for Automated Machine Learning Algorithm Discovery
arXiv · 6 iun. 2026