# AI Networking

AI networking integrates AI and machine learning with networking systems to improve network performance, security, automation, and support for AI workloads at scale.

This is one page of public article previews, not the complete archive. Follow Next page to continue. Summaries are not the original full articles.

## What's next for Azure networking?

DevFeed: [What's next for Azure networking?](<https://devfeed.tech/articles/what-s-next-for-azure-networking-146430.md>)

Original publisher: [Read original article](<https://www.networkworld.com/article/4251523/whats-next-for-azure-networking.html>)

Author: Sean Michael Kerner

Published: 2026-11-20T16:48:53Z

Content type: news

Language: en

Sources: [Network World](<https://devfeed.tech/sources/network-world.md>)

Topics: [AI Networking](<https://devfeed.tech/topics/ai-networking.md>), [Cloud Networking](<https://devfeed.tech/topics/cloud-networking.md>), [Azure Kubernetes Service](<https://devfeed.tech/topics/azure-kubernetes-service.md>), [Azure](<https://devfeed.tech/topics/azure.md>)

Tags: [agent](<https://devfeed.tech/tags/agent.md>), [agentic](<https://devfeed.tech/tags/agentic.md>), [agentic-ai](<https://devfeed.tech/tags/agentic-ai.md>), [agents](<https://devfeed.tech/tags/agents.md>), [ai](<https://devfeed.tech/tags/ai.md>), [ai-agents](<https://devfeed.tech/tags/ai-agents.md>), [ai-workloads](<https://devfeed.tech/tags/ai-workloads.md>), [aks](<https://devfeed.tech/tags/aks.md>), [amazon-web-services-aws](<https://devfeed.tech/tags/amazon-web-services-aws.md>), [apis](<https://devfeed.tech/tags/apis.md>), [at-t](<https://devfeed.tech/tags/at-t.md>), [azure](<https://devfeed.tech/tags/azure.md>), [cloud](<https://devfeed.tech/tags/cloud.md>), [cloud-computing](<https://devfeed.tech/tags/cloud-computing.md>), [cloud-computing-networking](<https://devfeed.tech/tags/cloud-computing-networking.md>), [connectivity](<https://devfeed.tech/tags/connectivity.md>), [data-center](<https://devfeed.tech/tags/data-center.md>), [data-centers](<https://devfeed.tech/tags/data-centers.md>), [edition-us](<https://devfeed.tech/tags/edition-us.md>), [enterprises](<https://devfeed.tech/tags/enterprises.md>), [exploit](<https://devfeed.tech/tags/exploit.md>), [fabric](<https://devfeed.tech/tags/fabric.md>), [language-en](<https://devfeed.tech/tags/language-en.md>), [networking](<https://devfeed.tech/tags/networking.md>), [storytype-news](<https://devfeed.tech/tags/storytype-news.md>)

### AI overview

AI workloads increase east-west traffic and shift demand across locations, so Microsoft is flattening its data center fabric, extending connectivity, and adding network controls. Its two-tier fabric cuts a five-hop path to three and is reported to be over 20% more power efficient; hollow core fiber is said to reduce latency by about one-third. For customers, Azure Multicloud Interconnect provisions cloud cross-connects through APIs (AWS generally available; GCP and Lumen offerings in preview, AT&T coming soon). AKS Fleet Manager peers clusters with failover and shared policy, while Route Server route maps filter, summarize, and influence routes. Agent Grid applies identity-based controls without changing agent code; Azure Firewall Premium's AI add-on uses two inline small language models to detect malware and exploit requests. Adaptive Virtual Patching applies mitigations until patches arrive. The excerpt also describes Microsoft's use of over 100 network agents, which handled more than 30,000 cases and reduced backlog by about three quarters over the past year.

### Source excerpt

AI workloads change how traffic moves inside a data center. GPUs exchange data with other GPUs, servers, and storage, which shifts traffic from north-south to east-west. AI agents add another layer. They span compute, GPUs, the network, and data centers, and demand can move from one place to another overnight. At its Ignite conference this week, Microsoft detailed Azure networking updates for this shift, including new services for enterprises as well as how Microsoft itself is using agentic AI for networking. (It also previewed Windows Server 2028, its largest platform update since Windows Server 2025.) Networking-related announcements made at Ignite include: Azure Multicloud Interconnect: Provisions cross-connects between Azure and other clouds through APIs. Amazon Web Services (AWS) is generally available. Google Cloud Platform (GCP) is in preview, as is Azure Last Mile Interconnect with Lumen. AT&T is coming soon. Cross-cluster networking for Azure Kubernetes Service (AKS): Peers clusters through Azure Kubernetes Fleet Manager with failover and one network policy. Azure Route Server route maps: Filter and summarize routes and apply autonomous system (AS) path prepending. Agent Grid: Enforces identity-based microsegmentation, isolation, and traffic policies for AI agents on the network. Azure Firewall AI add-on: Runs two inline small language models on Azure Firewall Premium to detect malware and exploit requests. Azure Adaptive Virtual Patching: Maps published vulnerabilities to workloads on virtual machines and applies inline mitigations until the actual patch arrives. Microsoft points to the shifts in traffic, location, and security that AI creates. "So, with all of these shifts, the network is no longer about connectivity. It is really becoming the intelligent control for the agentic world," said Igor Sakhnov, corporate vice president and general manager of Azure Networking, in a session at Microsoft Ignite. Azure flattens its fabric and adds hollow core fiber

## Can Bandits Learn to Share a Channel? Distributed Multi-Armed Bandits for Slotted ALOHA

DevFeed: [Can Bandits Learn to Share a Channel? Distributed Multi-Armed Bandits for Slotted ALOHA](<https://devfeed.tech/articles/can-bandits-learn-to-share-a-channel-distributed-multi-armed-bandits-for-slotted-aloha-143001.md>)

Original publisher: [Read original article](<https://arxiv.org/abs/2611.24676>)

Author: Kamil Szczech, Maksymilian Wojnar, Krzysztof Rusek, Katarzyna Kosek-Szott, Szymon Szott

Published: 2026-11-19T05:00:00Z

Content type: article

Language: en

Sources: [arXiv: Networking](<https://devfeed.tech/sources/arxiv-networking.md>)

Topics: [Reinforcement learning](<https://devfeed.tech/topics/reinforcement-learning.md>), [Machine learning](<https://devfeed.tech/topics/machine-learning.md>), [AI Networking](<https://devfeed.tech/topics/ai-networking.md>)

Tags: [bandits](<https://devfeed.tech/tags/bandits.md>), [cs-ni](<https://devfeed.tech/tags/cs-ni.md>), [distributed](<https://devfeed.tech/tags/distributed.md>), [fairness](<https://devfeed.tech/tags/fairness.md>), [machine-learning](<https://devfeed.tech/tags/machine-learning.md>), [networks](<https://devfeed.tech/tags/networks.md>), [reinforcement-learning](<https://devfeed.tech/tags/reinforcement-learning.md>), [softmax](<https://devfeed.tech/tags/softmax.md>), [throughput](<https://devfeed.tech/tags/throughput.md>)

### AI overview

In a capture-free slotted ALOHA system, the authors evaluate whether independently operating multi-armed bandits can learn channel access online without pre-training, fixed schedules, signaling, or prior knowledge of network size. Their best-performing method, SALSA, uses a Softmax bandit to learn a near-optimal per-node transmission policy. The abstract reports throughput near the theoretical maximum for fair random access and performance comparable to a state-of-the-art competing solution, while also evaluating aggregate throughput and per-node fairness.

### Source excerpt

arXiv:2611.24676v1 Announce Type: new Abstract: Slotted ALOHA networks constitute a playground for research on machine learning (ML)-based channel access schemes. We consider ML agents deployed in such networks and operating under severe constraints: fully distributed (independent operation), fully online (no pre-training), aperiodic (no fixed schedule), and with no explicit coordination (no signaling). We ask whether a multi-armed bandit (MAB), a minimal reinforcement learning agent, can learn the optimal policy for a capture-free slotted ALOHA system. We evaluate several MAB algorithms on aggregate network throughput and per-node fairness. Our best-performing agent is based on a Softmax bandit, which we call SALSA (Softmax ALOHA Learning for Shared Access). SALSA learns a per-node transmission policy close to optimal, with throughput near the theoretical maximum for fair random access. We achieve performance comparable to a competing state-of-the-art solution, but without requiring a priori knowledge of the network size.

## CoSlice: Coordinated Multi-Agent Reinforcement Learning for 5G Network Slicing Allocation

DevFeed: [CoSlice: Coordinated Multi-Agent Reinforcement Learning for 5G Network Slicing Allocation](<https://devfeed.tech/articles/coslice-coordinated-multi-agent-reinforcement-learning-for-5g-network-slicing-allocation-142996.md>)

Original publisher: [Read original article](<https://arxiv.org/abs/2611.23909>)

Author: Pengxin Wang, Nakjung Choi, Jingdi Chen

Published: 2026-11-19T05:00:00Z

Content type: article

Language: en

Sources: [arXiv: Machine Learning](<https://devfeed.tech/sources/arxiv-machine-learning.md>), [arXiv: Networking](<https://devfeed.tech/sources/arxiv-networking.md>)

Topics: [Reinforcement learning](<https://devfeed.tech/topics/reinforcement-learning.md>), [Multi Agent Systems](<https://devfeed.tech/topics/multi-agent-systems.md>), [AI Networking](<https://devfeed.tech/topics/ai-networking.md>), [Operations research and optimization](<https://devfeed.tech/topics/operations-research-and-optimization.md>)

Tags: [5g](<https://devfeed.tech/tags/5g.md>), [agent](<https://devfeed.tech/tags/agent.md>), [algorithm](<https://devfeed.tech/tags/algorithm.md>), [application](<https://devfeed.tech/tags/application.md>), [budget](<https://devfeed.tech/tags/budget.md>), [control](<https://devfeed.tech/tags/control.md>), [cs-lg](<https://devfeed.tech/tags/cs-lg.md>), [cs-ma](<https://devfeed.tech/tags/cs-ma.md>), [cs-ni](<https://devfeed.tech/tags/cs-ni.md>), [efficiency](<https://devfeed.tech/tags/efficiency.md>), [environment](<https://devfeed.tech/tags/environment.md>), [evaluation](<https://devfeed.tech/tags/evaluation.md>), [execution](<https://devfeed.tech/tags/execution.md>), [experimental](<https://devfeed.tech/tags/experimental.md>), [fast](<https://devfeed.tech/tags/fast.md>), [framework](<https://devfeed.tech/tags/framework.md>), [game](<https://devfeed.tech/tags/game.md>), [global](<https://devfeed.tech/tags/global.md>), [global-network](<https://devfeed.tech/tags/global-network.md>), [learning](<https://devfeed.tech/tags/learning.md>), [multi-agent](<https://devfeed.tech/tags/multi-agent.md>), [network](<https://devfeed.tech/tags/network.md>), [reinforcement-learning](<https://devfeed.tech/tags/reinforcement-learning.md>)

### AI overview

CoSlice combines fast, preference-conditioned slice agents with a slower LLM coordinator that adjusts their objectives using global network observations to allocate O-RAN physical resource blocks. The authors frame allocation as a multi-objective partially observable stochastic game, report a Preference Correlated Equilibrium and a sublinear regret bound, and train the system through multi-objective pretraining followed by joint optimization. In an evaluation environment built with ORANSlice and RFsim, experiments across diverse slice workloads found lower average and tail SLA violations than centralized and multi-agent baselines. The evidence provided is an abstract, so it does not specify effect sizes or implementation details.

### Source excerpt

arXiv:2611.23909v1 Announce Type: new Abstract: Radio Access Network (RAN) slicing is a shared resource allocation problem: each slice seeks to satisfy its Service Level Agreement (SLA) with minimal radio-resource consumption, while all slices compete for a limited resource budget. This creates both an intra-slice trade-off between SLA assurance and resource efficiency and an inter-slice conflict over shared resources, calling for specialized yet coordinated control. In this work, we present CoSlice, a coordinated multi-agent reinforcement learning framework for inter-slice physical resource block allocation in Open Radio Access Network (O-RAN). CoSlice uses preference-conditioned slice agents for fast resource allocation and an LLM-based coordinator to adapt their objectives using global network observations at a slower timescale. We formulate the problem as a multi-objective partially observable stochastic game (mo-POSG), establish the existence of a Preference Correlated Equilibrium, and derive a sublinear regret bound for the learning procedure. We implement the algorithm by using a two-stage training procedure that consists of multi-objective pretraining and joint optimization. We build a network slicing evaluation environment using ORANSlice and its radio-frequency simulator (RFsim), supporting realistic protocol-stack execution and application traffic across multiple slices. Experimental results show that CoSlice reduces both average and tail SLA violations compared with other centralized and multi-agent baselines under diverse slice workloads.

## Nvidia expands AI infrastructure push into high-performance computing

DevFeed: [Nvidia expands AI infrastructure push into high-performance computing](<https://devfeed.tech/articles/nvidia-expands-ai-infrastructure-push-into-high-performance-computing-141282.md>)

Original publisher: [Read original article](<https://www.networkworld.com/article/4250303/nvidia-expands-ai-infrastructure-push-into-high-performance-computing.html>)

Author: Andy Patrizio

Published: 2026-11-18T17:10:21Z

Content type: news

Language: en

Sources: [Network World](<https://devfeed.tech/sources/network-world.md>)

Topics: [High-Performance Computing](<https://devfeed.tech/topics/high-performance-computing.md>), [Nvidia](<https://devfeed.tech/topics/nvidia.md>), [Supercomputing](<https://devfeed.tech/topics/supercomputing.md>), [CUDA-X](<https://devfeed.tech/topics/cuda-x.md>), [systems](<https://devfeed.tech/topics/systems.md>), [AI Infrastructure](<https://devfeed.tech/topics/ai-infrastructure.md>), [Nemotron](<https://devfeed.tech/topics/nemotron.md>), [NVLink](<https://devfeed.tech/topics/nvlink.md>), [AI Networking](<https://devfeed.tech/topics/ai-networking.md>)

Tags: [agents](<https://devfeed.tech/tags/agents.md>), [ai](<https://devfeed.tech/tags/ai.md>), [ai-agents](<https://devfeed.tech/tags/ai-agents.md>), [ai-infrastructure](<https://devfeed.tech/tags/ai-infrastructure.md>), [ai-models](<https://devfeed.tech/tags/ai-models.md>), [ai-workloads](<https://devfeed.tech/tags/ai-workloads.md>), [artificial-intelligence](<https://devfeed.tech/tags/artificial-intelligence.md>), [artificial-intelligence-cpus-and-processors-industry-markets-networking-technology-industry](<https://devfeed.tech/tags/artificial-intelligence-cpus-and-processors-industry-markets-networking-technology-industry.md>), [computing](<https://devfeed.tech/tags/computing.md>), [cpu](<https://devfeed.tech/tags/cpu.md>), [cpus](<https://devfeed.tech/tags/cpus.md>), [cpus-and-processors](<https://devfeed.tech/tags/cpus-and-processors.md>), [cuda](<https://devfeed.tech/tags/cuda.md>), [data](<https://devfeed.tech/tags/data.md>), [data-processing](<https://devfeed.tech/tags/data-processing.md>), [edition-us](<https://devfeed.tech/tags/edition-us.md>), [engineering](<https://devfeed.tech/tags/engineering.md>), [gpu](<https://devfeed.tech/tags/gpu.md>), [gpus](<https://devfeed.tech/tags/gpus.md>), [high-performance](<https://devfeed.tech/tags/high-performance.md>), [high-performance-computing](<https://devfeed.tech/tags/high-performance-computing.md>), [industry](<https://devfeed.tech/tags/industry.md>), [infrastructure](<https://devfeed.tech/tags/infrastructure.md>), [language-en](<https://devfeed.tech/tags/language-en.md>), [markets](<https://devfeed.tech/tags/markets.md>), [networking](<https://devfeed.tech/tags/networking.md>), [nvidia](<https://devfeed.tech/tags/nvidia.md>), [performance](<https://devfeed.tech/tags/performance.md>), [storytype-news](<https://devfeed.tech/tags/storytype-news.md>), [technology-industry](<https://devfeed.tech/tags/technology-industry.md>)

### AI overview

Nvidia's SC26 announcements extend its push from GPU sales toward full HPC and AI infrastructure. Rubin HPC, expected in 2028, targets scientific simulations, engineering, and data processing with more than 50,000 CUDA cores, seventh-generation NVLink, and sparse-data optimizations. A rack-scale Vera CPU system combines 192 CPUs per rack with Quantum-X800 InfiniBand and SHARP, using NVL72 chassis, power, and cooling components. Nemotron for Science offers open models, datasets, evaluation tools, and post-training recipes for scientific AI agents that can interpret research, generate code, and use computational tools. CUDA-X updates include claimed speedups ranging from 3x to 900x for specific scientific tasks, with comparisons varying by library and baseline. Lambda is deploying Nvidia Photonics co-packaged optics switches to connect GB300 NVL72 racks containing over 10,000 GPUs; Nvidia presents lower power use, reliability, and scalability as intended benefits.

### Source excerpt

Perhaps anxious to talk about technology instead of its CEO drama, Nvidia unveiled a raft of HPC-related news at the Supercomputing '26 show taking place in Chicago. Nvidia's announcements at SC26 continue the company's efforts to expand beyond selling GPUs into supplying complete AI and high-performance computing infrastructure. Rubin HPC: A new GPU for scientific computing The news covers next-generation GPU and CPU platforms, networking and interconnect technology, accelerated scientific computing, and systems designed for large-scale AI workloads. Here is a breakdown of the major developments. Nvidia announced Rubin HPC, a new GPU within its Vera Rubin platform designed specifically for high-performance computing applications. Unlike GPUs primarily optimized for AI training and inference, Rubin HPC is designed to boost scientific simulations, engineering applications, and large-scale data processing. Rubin HPC has more than 50,000 CUDA cores on a single GPU. The core count for regular Roman GPUs has not been disclosed. It includes seventh-generation NVLink for accelerated GPU-to-GPU communication, sparse-data acceleration, and optimizations for sparse matrix operations, stencil computations, and sorting and reduction operations. Rubin HPC is expected to ship in 2028. Vera CPU platform expands into HPC Nvidia also announced a rack-scale version of its Vera CPU platform specifically designed for HPC. When it first introduced its own CPU, Nvidia positioned it as an alternative to traditional server processors from Intel and AMD. Now Nvidia is positioning its CPU for a specific market. The new configuration combines 192 Vera CPUs per rack plus Quantum-X800 InfiniBand networking, and SHARP in-network computing technology. It uses existing NVL72 rack components, including chassis, power distribution, and cooling infrastructure. Nemotron for Science: AI agents for scientific research Nemotron is a family of AI models developed by Nvidia, designed to power AI applicatio

## Arista expands its routing options to support secure AI inferencing

DevFeed: [Arista expands its routing options to support secure AI inferencing](<https://devfeed.tech/articles/arista-expands-its-routing-options-to-support-secure-ai-inferencing-138575.md>)

Original publisher: [Read original article](<https://www.networkworld.com/article/4249783/arista-expands-its-routing-options-to-support-secure-ai-inferencing.html>)

Author: Michael Cooney

Published: 2026-11-17T19:05:35Z

Content type: news

Language: en

Sources: [Network World](<https://devfeed.tech/sources/network-world.md>)

Topics: [AI Networking](<https://devfeed.tech/topics/ai-networking.md>), [AI Infrastructure](<https://devfeed.tech/topics/ai-infrastructure.md>), [Segment Routing over IPv6 (SRv6)](<https://devfeed.tech/topics/segment-routing-over-ipv6-srv6.md>)

Tags: [ai](<https://devfeed.tech/tags/ai.md>), [arista](<https://devfeed.tech/tags/arista.md>), [artificial-intelligence](<https://devfeed.tech/tags/artificial-intelligence.md>), [artificial-intelligence-data-center-enterprise-routers-network-switches-networking-networking-d](<https://devfeed.tech/tags/artificial-intelligence-data-center-enterprise-routers-network-switches-networking-networking-d.md>), [data-center](<https://devfeed.tech/tags/data-center.md>), [edition-us](<https://devfeed.tech/tags/edition-us.md>), [encryption](<https://devfeed.tech/tags/encryption.md>), [enterprise-routers](<https://devfeed.tech/tags/enterprise-routers.md>), [eos](<https://devfeed.tech/tags/eos.md>), [inferencing](<https://devfeed.tech/tags/inferencing.md>), [language-en](<https://devfeed.tech/tags/language-en.md>), [network-switches](<https://devfeed.tech/tags/network-switches.md>), [networking](<https://devfeed.tech/tags/networking.md>), [networking-devices](<https://devfeed.tech/tags/networking-devices.md>), [routing](<https://devfeed.tech/tags/routing.md>), [secure-ai](<https://devfeed.tech/tags/secure-ai.md>), [storytype-news](<https://devfeed.tech/tags/storytype-news.md>)

### AI overview

Arista's new Etherlink R4 routers pair high-capacity networking with per-port line-rate encryption for distributed AI infrastructure. The 7020HR4 and 7020R4 are hardened 1RU, 48-port platforms aimed at edge, inference, data center leaf, and service provider access roles; they use dynamic deep buffers and offer port choices spanning 1G to 400G. The 7282R4 and 7280R4 provide up to 25.6 Tbps of non-blocking bandwidth and support 100GbE, 400GbE, and 800GbE; the 2RU 7282R4 adds dual Supervisor Engines for redundant control planes. Across the R4 line, TunnelSec supports MACsec, IPsec, VXLANsec, and hardware-based Segment Routing over IPv6, while EOS is intended to streamline centralized security updates, which Arista says can often be applied without rebooting. The 7020HR4 is expected in early 2027; the other new models are available now.

### Source excerpt

Arista is unveiling a portfolio of networking devices designed to secure and improve the performance and resiliency of connections serving AI compute and storage infrastructure. Four new Etherlink R4 high-performance routing switches are aimed at large-scale distributed data center, AI back-end, and service provider environments where high route scale, deep packet buffering, and 400G connectivity are required. In addition, the new R4s integrate line-rate encryption into every port via TunnelSec, including MACsec, IPsec, VXLANsec, and hardware-based Segment Routing over IPv6, according to a blog post co-authored by Brendan Gibbs, Arista vice president, product line management, and Tyson Lamoreaux, senior vice president, Cloud/AI. All run Arista's Extensible Operating System, or EOS. "The initial phase of artificial intelligence (AI) buildouts concentrated compute nodes in massive, centralized clusters, but the need for greater scale has led operators to span training clusters across multiple data centers with AI scale-across solutions. Now, the operational reality of centralized inferencing has emerged as the new frontier requiring a distributed solution," Gibbs and Lamoreaux wrote. "Training can tolerate being tucked away near remote power grids, while inferencing cannot. As AI transitions from static prompts to real-time interaction, intelligence must be distributed across front-end networks spanning distinct geographies, regional hubs, and enterprise perimeters." The new product lineup features the hardened 1RU, 48-port 7020HR4 and 7020R4 compact routing platforms. Both are targeted at high-performance edge applications, distributed AI inference, secure data center leaf, service provider access routing, and high-end desktop applications. As part of the R4 family, 7020R4 Series platforms share a common forwarding architecture with dynamic deep buffers for deterministic, lossless forwarding performance and support scalable resources. Both are available with a choice

## PaletteAI and BE Verity: one workflow for GPUs and fabric | Spectro Cloud

DevFeed: [PaletteAI and BE Verity: one workflow for GPUs and fabric | Spectro Cloud](<https://devfeed.tech/articles/paletteai-and-be-verity-one-workflow-for-gpus-and-fabric-spectro-cloud-138176.md>)

Original publisher: [Read original article](<https://www.spectrocloud.com/blog/paletteai-be-verity-gpu-and-fabric-orchestration>)

Author: Saad Malik

Published: 2026-11-17T15:00:00Z

Content type: release

Language: en

Sources: [Spectro Cloud Feed](<https://devfeed.tech/sources/spectro-cloud-feed.md>)

Topics: [workflow automation](<https://devfeed.tech/topics/workflow-automation.md>), [AI Networking](<https://devfeed.tech/topics/ai-networking.md>)

Tags: [ai](<https://devfeed.tech/tags/ai.md>), [ai-factories](<https://devfeed.tech/tags/ai-factories.md>), [ai-factory](<https://devfeed.tech/tags/ai-factory.md>), [automatically](<https://devfeed.tech/tags/automatically.md>), [automation](<https://devfeed.tech/tags/automation.md>), [bare-metal](<https://devfeed.tech/tags/bare-metal.md>), [block](<https://devfeed.tech/tags/block.md>), [broadcom](<https://devfeed.tech/tags/broadcom.md>), [cluster](<https://devfeed.tech/tags/cluster.md>), [clusters](<https://devfeed.tech/tags/clusters.md>), [codex](<https://devfeed.tech/tags/codex.md>), [compute](<https://devfeed.tech/tags/compute.md>), [configuration](<https://devfeed.tech/tags/configuration.md>), [connect](<https://devfeed.tech/tags/connect.md>), [fabric](<https://devfeed.tech/tags/fabric.md>), [gpus](<https://devfeed.tech/tags/gpus.md>), [sc26](<https://devfeed.tech/tags/sc26.md>), [servers](<https://devfeed.tech/tags/servers.md>), [tenant](<https://devfeed.tech/tags/tenant.md>), [workflow](<https://devfeed.tech/tags/workflow.md>)

### AI overview

PaletteAI and BE Verity coordinate tenant setup across GPU compute and network fabric in one workflow. PaletteAI assigns bare-metal GPU servers, deploys Kubernetes and the AI stack; it then calls Verity to configure the tenant's network across the fabric and connect the servers, so compute and network are ready together and tenant traffic is segmented from the start. PaletteAI manages compute, while Verity orchestrates and observes multi-vendor Ethernet fabrics. The described integration uses fabric Verity has already built.

### Source excerpt

Spectro Cloud PaletteAI now integrates with BE Verity, so a new tenant's GPU servers and network fabric are provisioned in one workflow. See it at SC26.

## Delay-Aware Reinforcement Learning for O-RAN Control under Stochastic Reward Feedback

DevFeed: [Delay-Aware Reinforcement Learning for O-RAN Control under Stochastic Reward Feedback](<https://devfeed.tech/articles/delay-aware-reinforcement-learning-for-o-ran-control-under-stochastic-reward-feedback-138859.md>)

Original publisher: [Read original article](<https://arxiv.org/abs/2611.19344>)

Author: Xingqi Wu, Junaid Farooq, Tao Li, Juntao Chen, Ying Wang

Published: 2026-11-17T05:00:00Z

Content type: article

Language: en

Sources: [arXiv: Machine Learning](<https://devfeed.tech/sources/arxiv-machine-learning.md>)

Topics: [Reinforcement learning](<https://devfeed.tech/topics/reinforcement-learning.md>), [AI Networking](<https://devfeed.tech/topics/ai-networking.md>), [Latency](<https://devfeed.tech/topics/latency.md>)

Tags: [actions](<https://devfeed.tech/tags/actions.md>), [ai](<https://devfeed.tech/tags/ai.md>), [asynchronous](<https://devfeed.tech/tags/asynchronous.md>), [control](<https://devfeed.tech/tags/control.md>), [convergence](<https://devfeed.tech/tags/convergence.md>), [credit-assignment](<https://devfeed.tech/tags/credit-assignment.md>), [cs-lg](<https://devfeed.tech/tags/cs-lg.md>), [cs-ma](<https://devfeed.tech/tags/cs-ma.md>), [decision](<https://devfeed.tech/tags/decision.md>), [deep](<https://devfeed.tech/tags/deep.md>), [delay](<https://devfeed.tech/tags/delay.md>), [faster](<https://devfeed.tech/tags/faster.md>), [feedback](<https://devfeed.tech/tags/feedback.md>), [framework](<https://devfeed.tech/tags/framework.md>), [latency](<https://devfeed.tech/tags/latency.md>), [learning](<https://devfeed.tech/tags/learning.md>), [long-short-term-memory](<https://devfeed.tech/tags/long-short-term-memory.md>), [lstm](<https://devfeed.tech/tags/lstm.md>), [markov](<https://devfeed.tech/tags/markov.md>), [module](<https://devfeed.tech/tags/module.md>), [reinforcement-learning](<https://devfeed.tech/tags/reinforcement-learning.md>), [reward](<https://devfeed.tech/tags/reward.md>)

### AI overview

In O-RAN control, stochastic delays can separate an action from the reward used to train the controller, misaligning credit assignment and destabilizing learning. The proposed off-policy deep RL framework predicts delay with an LSTM and realigns rewards over time. In experiments using delay traces from the Open AI Cellular (OAIC) testbed, it reportedly converges faster and reduces latency violations and packet drops compared with conventional RL that ignores feedback delay. The supplied excerpt does not provide quantitative results or further implementation details.

### Source excerpt

arXiv:2611.19344v1 Announce Type: cross Abstract: Reinforcement learning (RL) is increasingly employed for real-time control in Open Radio Access Networks (O-RANs). However, the asynchronous nature of O-RAN feedback introduces stochastic delays between actions and their observable rewards, leading to misaligned credit assignment and unstable training. This letter formulates the O-RAN control problem as a Markov decision process (MDP) with delayed and stochastic rewards, where the true reward corresponding to an action is received after an unknown delay. To address this, we propose a delay-aware deep RL framework that integrates a long short-term memory (LSTM)-based delay predictor and a temporal reward realignment module within an off-policy learning pipeline. Using empirical delay traces collected from the Open AI Cellular (OAIC) testbed, we demonstrate that the proposed framework achieves faster convergence and significantly reduces latency violations and packet drops compared to conventional RL methods that ignore feedback delay. The results confirm that reward-delay compensation is critical for reliable AI-driven O-RAN control.

## HPE Lux AI Cluster Debuts at No. 14 on the TOP500 With 220.79 Petaflops on AMD Instinct MI355X GPUs

DevFeed: [HPE Lux AI Cluster Debuts at No. 14 on the TOP500 With 220.79 Petaflops on AMD Instinct MI355X GPUs](<https://devfeed.tech/articles/hpe-lux-ai-cluster-debuts-at-no-14-on-the-top500-with-220-79-petaflops-on-amd-instinct-mi355x-gpus-136079.md>)

Original publisher: [Read original article](<https://www.storagereview.com/news/hpe-lux-ai-cluster-top500-no-14-220-79-petaflops-ornl-genesis-mission>)

Author: Harold Fritts

Published: 2026-11-16T19:00:00Z

Content type: news

Language: en

Sources: [StorageReview.com](<https://devfeed.tech/sources/storagereview-com.md>)

Topics: [AI Infrastructure](<https://devfeed.tech/topics/ai-infrastructure.md>), [Supercomputing](<https://devfeed.tech/topics/supercomputing.md>), [rocm](<https://devfeed.tech/topics/rocm.md>), [AI Networking](<https://devfeed.tech/topics/ai-networking.md>), [GPU optimization](<https://devfeed.tech/topics/gpu-optimization.md>)

Tags: [ai](<https://devfeed.tech/tags/ai.md>), [ai-infrastructure](<https://devfeed.tech/tags/ai-infrastructure.md>), [ai-training-and-inference](<https://devfeed.tech/tags/ai-training-and-inference.md>), [amd](<https://devfeed.tech/tags/amd.md>), [amd-epyc](<https://devfeed.tech/tags/amd-epyc.md>), [amd-instinct-mi355x](<https://devfeed.tech/tags/amd-instinct-mi355x.md>), [cluster](<https://devfeed.tech/tags/cluster.md>), [computing](<https://devfeed.tech/tags/computing.md>), [cpus](<https://devfeed.tech/tags/cpus.md>), [data-center](<https://devfeed.tech/tags/data-center.md>), [deep](<https://devfeed.tech/tags/deep.md>), [deep-learning](<https://devfeed.tech/tags/deep-learning.md>), [department-of-energy](<https://devfeed.tech/tags/department-of-energy.md>), [enterprise](<https://devfeed.tech/tags/enterprise.md>), [general](<https://devfeed.tech/tags/general.md>), [gpus](<https://devfeed.tech/tags/gpus.md>), [hpe](<https://devfeed.tech/tags/hpe.md>), [oak-ridge-national-laboratory](<https://devfeed.tech/tags/oak-ridge-national-laboratory.md>), [performance](<https://devfeed.tech/tags/performance.md>), [supercomputing](<https://devfeed.tech/tags/supercomputing.md>), [top500](<https://devfeed.tech/tags/top500.md>)

### AI overview

HPE says Lux, an AI cluster built with AMD for Oak Ridge National Laboratory, measured 220.79 petaflops on HPL and debuted at No. 14 on the TOP500, ahead of systems designed for traditional supercomputing workloads. It uses direct liquid-cooled HPE ProLiant XD685 servers with AMD Instinct MI355X GPUs, EPYC CPUs and Pensando DPUs, plus 800G Ethernet switches. HPE says network load balancing and congestion management are intended to improve GPU utilization; AMD says the system runs ROCm. Lux provides multi-tenant, cloud-style access for researchers and is the first dedicated AI system online for DOE's Genesis Mission. Half of its 3.5 million annual node-hours are reserved for that mission and half for proprietary use. The excerpt gives no node or GPU count.

### Source excerpt

HPE today announced that Lux, the AI cluster it built with AMD for Oak Ridge National Laboratory (ORNL), has debuted at No. 14 on the TOP500 list with 220.79 petaflops of measured HPL performance. HPE says the system, which was designed for AI training and inference, placed ahead of machines built for traditional supercomputing workloads. The post HPE Lux AI Cluster Debuts at No. 14 on the TOP500 With 220.79 Petaflops on AMD Instinct MI355X GPUs appeared first on StorageReview.com.

## Lambda announces the world's first large-scale CPO deployment of NVIDIA Photonics switches

DevFeed: [Lambda announces the world's first large-scale CPO deployment of NVIDIA Photonics switches](<https://devfeed.tech/articles/lambda-announces-the-world-s-first-large-scale-cpo-deployment-of-nvidia-photonics-switches-136925.md>)

Original publisher: [Read original article](<https://lambda.com/blog/worlds-first-large-scale-deployment-of-nvidia-photonics-switches-with-cpo>)

Author: Lambda

Published: 2026-11-16T16:03:52Z

Content type: release

Language: en

Sources: [The Lambda deep learning blog](<https://devfeed.tech/sources/the-lambda-deep-learning-blog.md>)

Topics: [AI Networking](<https://devfeed.tech/topics/ai-networking.md>), [Nvidia](<https://devfeed.tech/topics/nvidia.md>), [NVIDIA Blackwell](<https://devfeed.tech/topics/nvidia-blackwell.md>)

Tags: [announcements](<https://devfeed.tech/tags/announcements.md>), [blackwell](<https://devfeed.tech/tags/blackwell.md>), [blackwell-ultra](<https://devfeed.tech/tags/blackwell-ultra.md>), [co-packaged-optics](<https://devfeed.tech/tags/co-packaged-optics.md>), [cooler](<https://devfeed.tech/tags/cooler.md>), [cpo](<https://devfeed.tech/tags/cpo.md>), [deployment](<https://devfeed.tech/tags/deployment.md>), [first](<https://devfeed.tech/tags/first.md>), [gpus](<https://devfeed.tech/tags/gpus.md>), [integrity](<https://devfeed.tech/tags/integrity.md>), [lambda](<https://devfeed.tech/tags/lambda.md>), [large](<https://devfeed.tech/tags/large.md>), [latency](<https://devfeed.tech/tags/latency.md>), [networking](<https://devfeed.tech/tags/networking.md>), [nvidia](<https://devfeed.tech/tags/nvidia.md>), [nvidia-blackwell](<https://devfeed.tech/tags/nvidia-blackwell.md>), [nvidia-gb300](<https://devfeed.tech/tags/nvidia-gb300.md>), [nvidia-gb300-nvl72](<https://devfeed.tech/tags/nvidia-gb300-nvl72.md>), [optics](<https://devfeed.tech/tags/optics.md>), [photonics](<https://devfeed.tech/tags/photonics.md>), [scale](<https://devfeed.tech/tags/scale.md>), [switches](<https://devfeed.tech/tags/switches.md>), [world](<https://devfeed.tech/tags/world.md>)

### AI overview

Lambda says it deployed NVIDIA Photonics co-packaged optics (CPO) switches to connect more than 10,000 NVIDIA Blackwell Ultra GPUs at its Quebec campus. CPO places optical engines beside the switch ASIC, shortening electrical paths from inches to millimeters; the announcement says this reduces signal loss and removes the need for retimer chips. In benchmarks across rack, scalable-unit and pod configurations, CPO matched conventional pluggable optics in throughput, showed more stable signal integrity, and had lower latency, with the latency advantage increasing from 3.8% at rack to 10.1% at pod. Median optic temperature was 36°C versus 51°C for pluggable optics. The announcement says cooler, more stable optics support up to 15 times higher reliability. The switches have 144 800Gb/s ports, liquid cooling and 18 removable external light sources. These are company-reported findings; no benchmark methodology or independent validation is provided in the excerpt.

### Source excerpt

Deployment connects 10,000+ NVIDIA Blackwell Ultra GPUs and demonstrates lower latency, cooler optics, and improved signal integrity with co-packaged optics networking

## PF-Exec: Large-Scale UAV Relay Load Balancing with an Anchored Potential Direction Decider and a Per-Dimension Gain-Learning Executor

DevFeed: [PF-Exec: Large-Scale UAV Relay Load Balancing with an Anchored Potential Direction Decider and a Per-Dimension Gain-Learning Executor](<https://devfeed.tech/articles/pf-exec-large-scale-uav-relay-load-balancing-with-an-anchored-potential-direction-decider-and-a-per-dimension-gain-learning-executor-135786.md>)

Original publisher: [Read original article](<https://arxiv.org/abs/2611.17492>)

Author: Liheng Zeng, Hanzhang Jia, Yiran Li, Yi Gao, Bo Ma

Published: 2026-11-16T05:00:00Z

Content type: article

Language: en

Sources: [arXiv: Machine Learning](<https://devfeed.tech/sources/arxiv-machine-learning.md>)

Topics: [Artificial Intelligence](<https://devfeed.tech/topics/ai.md>), [Load Balancing](<https://devfeed.tech/topics/load-balancing.md>), [AI Networking](<https://devfeed.tech/topics/ai-networking.md>)

Tags: [balancing](<https://devfeed.tech/tags/balancing.md>), [cs-it](<https://devfeed.tech/tags/cs-it.md>), [cs-lg](<https://devfeed.tech/tags/cs-lg.md>), [dimension](<https://devfeed.tech/tags/dimension.md>), [eess-sp](<https://devfeed.tech/tags/eess-sp.md>), [executor](<https://devfeed.tech/tags/executor.md>), [large](<https://devfeed.tech/tags/large.md>), [learning](<https://devfeed.tech/tags/learning.md>), [load](<https://devfeed.tech/tags/load.md>), [load-balancing](<https://devfeed.tech/tags/load-balancing.md>), [math-it](<https://devfeed.tech/tags/math-it.md>), [relay](<https://devfeed.tech/tags/relay.md>)

### AI overview

PF-Exec separates large-scale UAV relay load balancing into an anchored potential direction decider and a gain-learning executor. The decider supplies closed-form targets; the executor learns a positive gain for each dimension from local observations and clips actions to physical single-step limits. Unit gains make its initial behavior equivalent to the pure potential policy. Bandwidth targets are closed form, while position targets use the demand-weighted centroid of served users. In the highest-pressure burst tests, the fixed-step comparison loses both fairness and mean load on some seeds; PF-Exec keeps fairness near one. Of ten comparison methods, only the greedy heuristic survives that setting, with much higher load variance. Fairness also stays high with stale global state and position noise in the two largest scenarios. Learned and fixed-step executor flight energy is nearly identical. The abstract reports proofs of convergence properties for both layers, but provides no detailed numerical results or proof assumptions.

### Source excerpt

arXiv:2611.17492v1 Announce Type: new Abstract: Relay load balancing in large-scale UAV networks requires joint decisions on association, bandwidth, power and hovering position, while the dimension grows with the number of UAVs and each UAV sees only its local state. This paper splits the decision into two layers. An anchored potential direction decider emits a closed-form target at every instant and returns its direction only. A gain-learning executor maps a local observation into a positive gain per dimension, composes gain and target, and clips the action against the physical single-step bound; the two layers form PF-Exec. The unit gain is the default start, so the executor reduces to the pure potential policy before training. The load is first-order homogeneous in bandwidth, making the bandwidth target closed form, and the position target is the demand-weighted centroid of the served users. The system uses a two-hop access and millimetre-wave backhaul link with soft association, a 3GPP UMa-AV channel with two shadowing layers, an OU-type social-force crowd and a quadrotor physical layer. Evaluation spans all fleet sizes, two crowd scenarios, independent seeds, the published baselines in their own settings and a degraded-information line. In the highest-pressure burst setting the fixed-step comparison breaks down on part of the seeds, losing fairness and mean load at once, while PF-Exec holds fairness near one. Among the ten comparison methods only the greedy heuristic survives that setting; its load variance stays far higher. Under stale global state and position noise the fairness index stays high at both largest-scale scenarios. Flight energy is nearly identical between the learned and the fixed-step executor, and three constants are fixed once across all settings. The convergence properties of both layers, from the truncated gain iteration to the perturbed tracking bound, are proved in the companion analysis.

## Evaluating Polling Schemes for OFDMA Uplink Scheduling in Time-Sensitive Wi-Fi Networks

DevFeed: [Evaluating Polling Schemes for OFDMA Uplink Scheduling in Time-Sensitive Wi-Fi Networks](<https://devfeed.tech/articles/evaluating-polling-schemes-for-ofdma-uplink-scheduling-in-time-sensitive-wi-fi-networks-134918.md>)

Original publisher: [Read original article](<https://arxiv.org/abs/2611.17898>)

Author: Douglas Dziedzorm Agbeve, Ferhat Van Herck, Andrey Belogaev, Jeroen Famaey

Published: 2026-11-16T05:00:00Z

Content type: comparison

Language: en

Sources: [arXiv: Networking](<https://devfeed.tech/sources/arxiv-networking.md>)

Topics: [scheduling](<https://devfeed.tech/topics/scheduling.md>), [AI Networking](<https://devfeed.tech/topics/ai-networking.md>)

Tags: [access-point](<https://devfeed.tech/tags/access-point.md>), [buffer](<https://devfeed.tech/tags/buffer.md>), [channel](<https://devfeed.tech/tags/channel.md>), [concurrent](<https://devfeed.tech/tags/concurrent.md>), [cs-ni](<https://devfeed.tech/tags/cs-ni.md>), [delay](<https://devfeed.tech/tags/delay.md>), [extended-reality](<https://devfeed.tech/tags/extended-reality.md>), [head-tracking](<https://devfeed.tech/tags/head-tracking.md>), [industrial-iot](<https://devfeed.tech/tags/industrial-iot.md>), [learning](<https://devfeed.tech/tags/learning.md>), [load](<https://devfeed.tech/tags/load.md>), [networks](<https://devfeed.tech/tags/networks.md>), [packets](<https://devfeed.tech/tags/packets.md>), [performance](<https://devfeed.tech/tags/performance.md>), [scheduled](<https://devfeed.tech/tags/scheduled.md>), [scheduling](<https://devfeed.tech/tags/scheduling.md>), [status](<https://devfeed.tech/tags/status.md>), [throughput](<https://devfeed.tech/tags/throughput.md>), [traffic](<https://devfeed.tech/tags/traffic.md>), [video](<https://devfeed.tech/tags/video.md>), [wi-fi](<https://devfeed.tech/tags/wi-fi.md>)

### AI overview

The study finds that no uplink polling scheme performs best across all time-sensitive Wi-Fi workloads; results depend on traffic and load. For large teleconferences, activity-based and learning-based polling perform best, though contention-based access remains competitive at light load. In industrial IoT, scheduled access delivers the most deadline-bound packets under heavy load, while contention-based access performs best under lighter load. For XR, OFDMA schemes sustain video throughput that contention-based access cannot; activity-based polling has the fewest head-tracking deadline misses but longer video delay tails. The evidence is from an abstract describing three evaluated scenarios.

### Source excerpt

arXiv:2611.17898v1 Announce Type: new Abstract: IEEE 802.11ax introduced orthogonal frequency division multiple access (OFDMA) to improve channel utilization through concurrent multi-user transmissions. For effective uplink (UL) scheduling, the access point (AP) requires timely knowledge of buffered traffic at the stations (STAs). Explicit buffer-status polling provides this, yet each polling round can query only as many STAs as there are resource units (RUs). Scheduled-access, random-access, activity-based, and learning-based schemes address this buffer-status acquisition problem differently and have generally been evaluated in isolation under different conditions, so it remains unclear how they compare across time-sensitive workloads. We evaluate common schemes from the Wi-Fi standard and literature under three time-sensitive environments: a large-scale teleconferencing session, an industrial IoT deployment, and an extended reality (XR) service. The relative performance of the schemes depends strongly on traffic characteristics and offered load. In conferencing, activity-based and learning-based polling perform best, while conventional contention-based access remains competitive under light load. In the industrial scenario, scheduled-access most effectively delivers time-sensitive packets within their deadlines under heavy load, whereas contention-based access performs best under lighter load. In XR, OFDMA-based schemes achieve high video throughput that contention-based access cannot sustain; among them, activity-based polling achieves the lowest deadline-miss ratios for uplink head-tracking traffic, at the expense of longer video delay tails. These findings highlight the importance of selecting channel access and scheduling schemes according to traffic requirements and operating conditions.

## Cisco forecasts $9 billion in AI infrastructure revenue in FY27

DevFeed: [Cisco forecasts $9 billion in AI infrastructure revenue in FY27](<https://devfeed.tech/articles/cisco-forecasts-9-billion-in-ai-infrastructure-revenue-in-fy27-133395.md>)

Original publisher: [Read original article](<https://www.networkworld.com/article/4248551/cisco-forecasts-9-billion-in-ai-infrastructure-revenue-in-fy27.html>)

Author: Michael Cooney

Published: 2026-11-13T19:28:44Z

Content type: news

Language: en

Sources: [Network World](<https://devfeed.tech/sources/network-world.md>)

Topics: [Cisco](<https://devfeed.tech/topics/cisco.md>), [AI Infrastructure](<https://devfeed.tech/topics/ai-infrastructure.md>), [AI Networking](<https://devfeed.tech/topics/ai-networking.md>), [ai security](<https://devfeed.tech/topics/ai-security.md>), [data centers](<https://devfeed.tech/topics/data-centers.md>)

Tags: [ai](<https://devfeed.tech/tags/ai.md>), [ai-infrastructure](<https://devfeed.tech/tags/ai-infrastructure.md>), [artificial-intelligence](<https://devfeed.tech/tags/artificial-intelligence.md>), [artificial-intelligence-industry-markets-network-security-networking-security-technology-indus](<https://devfeed.tech/tags/artificial-intelligence-industry-markets-network-security-networking-security-technology-indus.md>), [cisco](<https://devfeed.tech/tags/cisco.md>), [cloud](<https://devfeed.tech/tags/cloud.md>), [data-center](<https://devfeed.tech/tags/data-center.md>), [edition-us](<https://devfeed.tech/tags/edition-us.md>), [hyperscalers](<https://devfeed.tech/tags/hyperscalers.md>), [industry](<https://devfeed.tech/tags/industry.md>), [language-en](<https://devfeed.tech/tags/language-en.md>), [markets](<https://devfeed.tech/tags/markets.md>), [network-security](<https://devfeed.tech/tags/network-security.md>), [networking](<https://devfeed.tech/tags/networking.md>), [revenue](<https://devfeed.tech/tags/revenue.md>), [security](<https://devfeed.tech/tags/security.md>), [storytype-news](<https://devfeed.tech/tags/storytype-news.md>), [technology-industry](<https://devfeed.tech/tags/technology-industry.md>)

### AI overview

Cisco reported Q1 revenue of $18.5 billion, up 24% year over year, and 28% growth in networking product orders. CEO Chuck Robbins projected AI infrastructure revenue of $9 billion in FY27, more than double FY26, citing hyperscaler deployments and 18 AI design wins over five quarters. He pointed to five potential growth drivers: hyperscale demand, early-stage campus refreshes, customer planning prompted by post-Mythos security concerns, demand for sovereign infrastructure, and enterprise investment in hybrid data centers. Security also contributed: core security revenue excluding Splunk rose 15%, firewall orders grew more than 30% for a third consecutive quarter, and Cisco said its AI Defense platform protects two trillion tokens monthly. The excerpt reports management's rationale and results; it does not establish whether projected growth will continue.

### Source excerpt

Cisco continues to benefit from a "networking supercycle," reporting record revenue--driven by demand for AI infrastructure, security, and campus refreshes from hyperscalers and enterprise customers--for its fiscal first quarter ended October 24. "In Q1, we delivered record quarterly revenue of $18.5 billion, up 24% year over year, and record non-GAAP earnings per share of $1.40, up 40% year over year. These are our strongest top and bottom-line growth rates in over 15 years," Cisco CEO Chuck Robbins said on the vendor's quarterly financial call. "Networking product orders grew 28% in Q1, delivering our ninth consecutive quarter of double-digit growth. In fact, we saw double-digit growth across every product family: service provider routing, compute, enterprise routing, wireless, data center switching and campus switching," Robbins said. "These broad-based results support our view that we are in the early stages of a multiyear, multibillion-dollar networking supercycle." Enterprise product orders were up 15% year-over-year, on top of 4% growth a year ago, with double-digit growth across all geographies and in networking and security, Robbins added. "We continue to see robust demand for our campus networking solutions in Q1 with 14% year-over-year growth in product orders. Our next-generation portfolio continues to ramp faster than prior product launches, and we believe we are still very early in this refresh cycle," he said. Product orders from service provider and cloud customers continue to be broad-based, growing 47% year-over-year, Robbins said. "Overall, our hyperscaler business continues to show strong momentum with AI infrastructure revenue projected to grow to $9 billion in fiscal year '27, more than double fiscal year '26 levels. Hyperscalers are now deploying our scale-across networking architecture and production environments as they extend coordinated AI workloads across data centers." As for whether or not this growth is sustainable over time, Robbins cit

## Janus: One Agent, Two Roles for Digital Twin Construction and Network Orchestration

DevFeed: [Janus: One Agent, Two Roles for Digital Twin Construction and Network Orchestration](<https://devfeed.tech/articles/janus-one-agent-two-roles-for-digital-twin-construction-and-network-orchestration-131574.md>)

Original publisher: [Read original article](<https://arxiv.org/abs/2611.14920>)

Author: Peijun Hou, Yue Zhang, Yi Wei, Nan Cen

Published: 2026-11-13T05:00:00Z

Content type: article

Language: en

Sources: [arXiv: Networking](<https://devfeed.tech/sources/arxiv-networking.md>)

Topics: [AI Networking](<https://devfeed.tech/topics/ai-networking.md>), [Network Operations](<https://devfeed.tech/topics/network-operations.md>), [llm-reasoning](<https://devfeed.tech/topics/llm-reasoning.md>), [AI Architecture](<https://devfeed.tech/topics/ai-architecture.md>), [Large Language Model](<https://devfeed.tech/topics/llm.md>), [Artificial Intelligence](<https://devfeed.tech/topics/ai.md>)

Tags: [agent](<https://devfeed.tech/tags/agent.md>), [agentic](<https://devfeed.tech/tags/agentic.md>), [ai](<https://devfeed.tech/tags/ai.md>), [architecture](<https://devfeed.tech/tags/architecture.md>), [association](<https://devfeed.tech/tags/association.md>), [cs-ni](<https://devfeed.tech/tags/cs-ni.md>), [digital-twin](<https://devfeed.tech/tags/digital-twin.md>), [graph](<https://devfeed.tech/tags/graph.md>), [large-language-models-llms](<https://devfeed.tech/tags/large-language-models-llms.md>), [network](<https://devfeed.tech/tags/network.md>), [orchestration](<https://devfeed.tech/tags/orchestration.md>), [overhead](<https://devfeed.tech/tags/overhead.md>), [reasoning](<https://devfeed.tech/tags/reasoning.md>), [routing](<https://devfeed.tech/tags/routing.md>), [scalability](<https://devfeed.tech/tags/scalability.md>)

### AI overview

Janus combines digital-twin construction and wireless-network orchestration in a closed loop. Its NetTwin agent uses one agent for two roles: building a twin from partial observations and making task-driven network decisions. A planner, modeler and coder structure the work, while a unified prompt module is designed to improve graph reasoning. In evaluations on networks of up to 200 nodes, covering user association and SAGIN routing, Janus reportedly achieved over 90% digital-twin accuracy and performance gaps within 10.53% of optimal, with moderate computational overhead. The supplied text is an abstract and does not provide further evaluation details.

### Source excerpt

arXiv:2611.14920v1 Announce Type: new Abstract: Next-generation wireless networks are increasingly large-scale, heterogeneous, and dynamic, posing significant challenges to network modeling and control. Digital twins offer a promising solution, but existing approaches typically rely on static models and decouple twin construction from network orchestration, limiting adaptability. Meanwhile, large language models (LLMs) provide strong reasoning capabilities but struggle with graph-structured data and are highly sensitive to prompt design, making inference unreliable. In this paper, we propose Janus, an agentic LLM-based framework that unifies digital twin construction and network orchestration within a closed-loop system. At the core of Janus is a novel NetTwin agent, which introduces a new one-agent, two-role paradigm, enabling a single agent to simultaneously construct the digital twin from partial observations while performing task-driven network decision-making. To realize this, we design a modular architecture with three functional roles, including planner, modeler, and coder, along with a newly designed unified prompt module that further enhances graph reasoning robustness. Extensive evaluations on wireless networks with up to 200 nodes and two representative tasks, including user association and SAGIN routing, show that Janus achieves over 90% digital twin accuracy and near-optimal performance with gaps within 10.53%, while maintaining moderate computational overhead. These results demonstrate the effectiveness, scalability, and generalizability of Janus, paving the way for AI-native autonomous wireless network orchestration.

## Human Grounded Evaluation of Large Language Models for Optical Network Automation

DevFeed: [Human Grounded Evaluation of Large Language Models for Optical Network Automation](<https://devfeed.tech/articles/human-grounded-evaluation-of-large-language-models-for-optical-network-automation-132813.md>)

Original publisher: [Read original article](<https://arxiv.org/abs/2607.18068>)

Author: Kiarash Rezaei, Omran Ayoub, Paolo Monti, Carlos Natalino

Published: 2026-11-13T05:00:00Z

Content type: article

Language: en

Sources: [arXiv: Machine Learning](<https://devfeed.tech/sources/arxiv-machine-learning.md>)

Topics: [LLM evaluation / benchmarking](<https://devfeed.tech/topics/llm-evaluation-benchmarking.md>), [Large Language Model](<https://devfeed.tech/topics/llm.md>), [AI Networking](<https://devfeed.tech/topics/ai-networking.md>)

Tags: [artificial-intelligence](<https://devfeed.tech/tags/artificial-intelligence.md>), [cs-ai](<https://devfeed.tech/tags/cs-ai.md>), [cs-lg](<https://devfeed.tech/tags/cs-lg.md>), [cs-ni](<https://devfeed.tech/tags/cs-ni.md>), [evaluation](<https://devfeed.tech/tags/evaluation.md>), [large-language-models](<https://devfeed.tech/tags/large-language-models.md>), [model-selection](<https://devfeed.tech/tags/model-selection.md>), [network-automation](<https://devfeed.tech/tags/network-automation.md>), [optical](<https://devfeed.tech/tags/optical.md>)

### AI overview

HuGLEN compares candidate language models for optical network automation by combining LLM-as-a-judge evaluations with a small set of expert ratings, then ranking models by a quality efficiency score (QES). In the demonstrated task, models turn explainable AI outputs for optical network quality-of-transmission estimation into operator-friendly explanations. The reported 12B-parameter model achieved the highest QES, the best observed trade-off between explanation quality and inference efficiency. The abstract says the pipeline reduces human-labeling burden and supports reproducible model selection; it does not provide details of the rating procedure or results beyond this example.

### Source excerpt

arXiv:2607.18068v2 Announce Type: cross Abstract: Large language models (LLMs) are increasingly adopted for network automation, yet their output quality and inference cost can vary substantially across LLM families. We present HuGLEN, a stepwise evaluation pipeline that uses an LLM-as-a-judge together with a small set of expert ratings to enable scalable and reproducible comparison of candidate LLMs, and to rank them using a quality efficiency score (QES). We demonstrate HuGLEN for translating outputs from an explainable artificial intelligence (XAI) model for the optical network quality of transmission (QoT) estimation task into operator-friendly explanations. Our results show that a medium-sized LLM (12B parameters) achieves the highest QES, indicating the best trade-off between explanation quality and efficiency. Overall, HuGLEN reduces the human-labeling burden while supporting consistent model selection for operator-facing automation tasks.

## Machine Learning for Scalable Multi-AP Coordination in IEEE 802.11bn Networks

DevFeed: [Machine Learning for Scalable Multi-AP Coordination in IEEE 802.11bn Networks](<https://devfeed.tech/articles/machine-learning-for-scalable-multi-ap-coordination-in-ieee-802-11bn-networks-131587.md>)

Original publisher: [Read original article](<https://arxiv.org/abs/2611.16634>)

Author: Maksymilian Wojnar, Szymon Szott

Published: 2026-11-13T05:00:00Z

Content type: article

Language: en

Sources: [arXiv: Machine Learning](<https://devfeed.tech/sources/arxiv-machine-learning.md>), [arXiv: Networking](<https://devfeed.tech/sources/arxiv-networking.md>)

Topics: [AI Networking](<https://devfeed.tech/topics/ai-networking.md>), [Machine learning](<https://devfeed.tech/topics/machine-learning.md>), [Wi-Fi](<https://devfeed.tech/topics/wi-fi.md>)

Tags: [access-points](<https://devfeed.tech/tags/access-points.md>), [bandits](<https://devfeed.tech/tags/bandits.md>), [cs-lg](<https://devfeed.tech/tags/cs-lg.md>), [cs-ni](<https://devfeed.tech/tags/cs-ni.md>), [distributed](<https://devfeed.tech/tags/distributed.md>), [flow-matching](<https://devfeed.tech/tags/flow-matching.md>), [graph-neural-networks](<https://devfeed.tech/tags/graph-neural-networks.md>), [hardware](<https://devfeed.tech/tags/hardware.md>), [ieee](<https://devfeed.tech/tags/ieee.md>), [ieee-802-11bn](<https://devfeed.tech/tags/ieee-802-11bn.md>), [machine-learning](<https://devfeed.tech/tags/machine-learning.md>), [networks](<https://devfeed.tech/tags/networks.md>), [scaling](<https://devfeed.tech/tags/scaling.md>), [simulation](<https://devfeed.tech/tags/simulation.md>), [testbed](<https://devfeed.tech/tags/testbed.md>), [throughput](<https://devfeed.tech/tags/throughput.md>), [wi-fi](<https://devfeed.tech/tags/wi-fi.md>)

### AI overview

In dense Wi-Fi deployments, IEEE 802.11bn multi-access point coordination (MAPC) lets neighboring APs cooperate, but selecting simultaneous transmitters and configurations is computationally difficult. The paper develops a progression of machine learning methods, from lightweight hierarchical multi-armed bandits (H-MABs) to a generative pipeline using graph neural networks and flow matching. It reports simulation and hardware testbed evaluations showing higher throughput than legacy access and coordination scaling to dozens of APs, where analytical and search-based methods become intractable. The supplied text is an abstract and does not give detailed performance figures or method comparisons.

### Source excerpt

arXiv:2611.16634v1 Announce Type: new Abstract: As Wi-Fi deployments become denser, fully distributed random channel access no longer guarantees the required quality of service. The multi-access point coordination (MAPC) framework introduced in IEEE~802.11bn lets neighboring access points (APs) cooperate, but choosing which APs transmit together and how to configure them remains an open, computationally hard problem. We develop and validate a stack of machine learning (ML) coordination methods, progressing from lightweight hierarchical multi-armed bandits (H-MABs) to a generative pipeline built on graph neural networks (GNNs) and flow matching (FM). The methods are evaluated in simulation and on a hardware testbed, improving throughput over legacy access and scaling coordination to dozens of APs where analytical and search-based methods become intractable.

## Network-Co-Designed Dynamic Sparsification for Dependable Distributed Robot Learning over 3GPP Networks

DevFeed: [Network-Co-Designed Dynamic Sparsification for Dependable Distributed Robot Learning over 3GPP Networks](<https://devfeed.tech/articles/network-co-designed-dynamic-sparsification-for-dependable-distributed-robot-learning-over-3gpp-networks-127840.md>)

Original publisher: [Read original article](<https://arxiv.org/abs/2611.13282>)

Author: G\\'eza Szab\\'o, Boldizs\\'ar Bunda

Published: 2026-11-11T05:00:00Z

Content type: article

Language: en

Sources: [arXiv: Networking](<https://devfeed.tech/sources/arxiv-networking.md>)

Topics: [AI Networking](<https://devfeed.tech/topics/ai-networking.md>), [distributed-training](<https://devfeed.tech/topics/distributed-training.md>), [Simulation](<https://devfeed.tech/topics/simulation.md>), [distributed algorithm](<https://devfeed.tech/topics/distributed-algorithm.md>)

Tags: [ai](<https://devfeed.tech/tags/ai.md>), [asynchronous](<https://devfeed.tech/tags/asynchronous.md>), [bandwidth](<https://devfeed.tech/tags/bandwidth.md>), [bounded](<https://devfeed.tech/tags/bounded.md>), [communication](<https://devfeed.tech/tags/communication.md>), [compression](<https://devfeed.tech/tags/compression.md>), [control](<https://devfeed.tech/tags/control.md>), [cs-ni](<https://devfeed.tech/tags/cs-ni.md>), [delay](<https://devfeed.tech/tags/delay.md>), [distributed](<https://devfeed.tech/tags/distributed.md>), [federated](<https://devfeed.tech/tags/federated.md>), [feedback](<https://devfeed.tech/tags/feedback.md>), [framework](<https://devfeed.tech/tags/framework.md>), [implement](<https://devfeed.tech/tags/implement.md>), [latency](<https://devfeed.tech/tags/latency.md>), [learning](<https://devfeed.tech/tags/learning.md>), [links](<https://devfeed.tech/tags/links.md>), [mapping](<https://devfeed.tech/tags/mapping.md>), [network](<https://devfeed.tech/tags/network.md>), [networks](<https://devfeed.tech/tags/networks.md>), [robot](<https://devfeed.tech/tags/robot.md>)

### AI overview

The paper proposes dynamic sparsification that adjusts robot model-update compression to network byte budgets and time constraints, aiming to reduce communication under volatile uplink conditions. It describes two approaches: a standards-aligned SA6/5GC mapping using CAPIF with NEF, PCF and NWDAF plus AI-specific semantics; and a VASEAL-style role-reversed loop in which the learning system shares learning-state feedback and receives bounded, time-limited control intents. The mechanism was implemented in asynchronous federated reinforcement learning with heterogeneous simulated quadrupeds and emulated uplink impairments. The reported results suggest substantial communication reduction without material loss of learning progress in that studied setting; the abstract does not establish performance beyond it.

### Source excerpt

arXiv:2611.13282v1 Announce Type: new Abstract: Robot fleets performing distributed learning often face volatile uplink conditions like bandwidth variation, packet loss, and latency spikes that delay or destabilize training. We propose a network-co-designed dynamic sparsification loop that couples model-update compression to network-facing byte budgets and time constraints. We describe two realizations: (i) a standards-aligned SA6/5GC mapping that leverages CAPIF with NEF/PCF/NWDAF, enriched with AI-specific semantics; and (ii) a VASEAL-style role-reversed loop where the learning system exposes learning-state feedback and receives bounded, time-limited control intents. We implement the mechanism in an asynchronous federated reinforcement learning framework with heterogeneous simulated quadruped robots and reproducible uplink impairments via emulated network links. Results indicate that substantial communication reduction is achievable under dynamic constraints without materially degrading learning progress in the studied setting.

## Resilient Integrated Access and Backhaul Network Planning via Graph Neural Network Curriculum Learning

DevFeed: [Resilient Integrated Access and Backhaul Network Planning via Graph Neural Network Curriculum Learning](<https://devfeed.tech/articles/resilient-integrated-access-and-backhaul-network-planning-via-graph-neural-network-curriculum-learning-127845.md>)

Original publisher: [Read original article](<https://arxiv.org/abs/2611.14042>)

Author: Jie Zhang, Haochen Sun, Mingzhang Zhou, Paul D. Mitchell, Hamed Ahmadi, Qiang Ni

Published: 2026-11-11T05:00:00Z

Content type: article

Language: en

Sources: [arXiv: Networking](<https://devfeed.tech/sources/arxiv-networking.md>)

Topics: [Reinforcement learning](<https://devfeed.tech/topics/reinforcement-learning.md>), [mesh networking](<https://devfeed.tech/topics/mesh-networking.md>), [Neural Network](<https://devfeed.tech/topics/neural-network.md>), [graph theory](<https://devfeed.tech/topics/graph-theory.md>), [AI Networking](<https://devfeed.tech/topics/ai-networking.md>), [Artificial Intelligence](<https://devfeed.tech/topics/ai.md>)

Tags: [actor](<https://devfeed.tech/tags/actor.md>), [attention](<https://devfeed.tech/tags/attention.md>), [backhaul](<https://devfeed.tech/tags/backhaul.md>), [budget](<https://devfeed.tech/tags/budget.md>), [connectivity](<https://devfeed.tech/tags/connectivity.md>), [constraint](<https://devfeed.tech/tags/constraint.md>), [coverage](<https://devfeed.tech/tags/coverage.md>), [cross](<https://devfeed.tech/tags/cross.md>), [cs-ni](<https://devfeed.tech/tags/cs-ni.md>), [decision](<https://devfeed.tech/tags/decision.md>), [deployment](<https://devfeed.tech/tags/deployment.md>), [develop](<https://devfeed.tech/tags/develop.md>), [efficiency](<https://devfeed.tech/tags/efficiency.md>), [failures](<https://devfeed.tech/tags/failures.md>), [fault-tolerance](<https://devfeed.tech/tags/fault-tolerance.md>), [fiber](<https://devfeed.tech/tags/fiber.md>), [graph](<https://devfeed.tech/tags/graph.md>), [learning](<https://devfeed.tech/tags/learning.md>), [network](<https://devfeed.tech/tags/network.md>), [neural-network](<https://devfeed.tech/tags/neural-network.md>), [planning](<https://devfeed.tech/tags/planning.md>)

### AI overview

The proposed GATv2 actor-critic with PPO uses multi-stage curriculum learning to plan resilient Integrated Access and Backhaul deployments across different fiber-connected donor layouts. It treats coverage and multi-hop transport feasibility as coupled constraints, accounting for hop-by-hop rate-budget degradation and minimum parent connectivity for fault tolerance. In a 60 GHz urban simulation across multiple donor layouts, the policy met the 98% minimum coverage requirement with fewer deployed IAB nodes than learning and heuristic baselines, maintained higher coverage during progressive backhaul failures, and transferred across donor topologies without retraining. These results are reported for the stated simulation setting.

### Source excerpt

arXiv:2611.14042v1 Announce Type: new Abstract: Integrated Access and Backhaul (IAB) is a key enabler for cost-effective millimeter-wave (mmWave) densification, yet practical deployments are strongly constrained by the spatial availability of fiber-connected IAB donors. Variations in donor topologies reshape both the initial coverage footprint and the feasible multi-hop backhaul expansion patterns, causing planning policies trained on a single topology to generalize poorly. This paper studies cross-topology generalization for resilience-aware IAB deployment under coverage and transport feasibility requirements. We formulate IAB deployment as a constrained Markov decision process that couples access coverage satisfaction with multi-hop transport feasibility, including hop-by-hop rate-budget degradation and a minimum parent-connectivity requirement for fault tolerance. To solve the resulting large discrete planning problem, we propose a graph-based actor-critic approach that uses a Graph Attention Network v2 (GATv2) encoder with Proximal Policy Optimization (PPO), and we develop a multi-stage curriculum learning strategy that progressively increases constraint difficulty and scenario diversity for stable training and transferable policies. In a 60 GHz urban simulation over a 1000 m x 1000 m area with multiple donor layouts, the trained policy satisfies the minimum coverage requirement of 98% with fewer deployed IAB nodes than learning and heuristic baselines, and maintains higher coverage under progressive backhaul failures. The results demonstrate that curriculum-trained graph-based reinforcement learning can improve deployment efficiency and resilience while transferring across heterogeneous IAB donor topologies without retraining.

## Agentic AI demands a stronger, more assured network foundation

DevFeed: [Agentic AI demands a stronger, more assured network foundation](<https://devfeed.tech/articles/agentic-ai-demands-a-stronger-more-assured-network-foundation-124197.md>)

Original publisher: [Read original article](<https://blogs.cisco.com/networking/agentic-ai-demands-a-stronger-more-assured-network-foundation>)

Author: Grant Shirk

Published: 2026-11-09T20:00:14Z

Content type: article

Language: en

Sources: [Cisco Blogs](<https://devfeed.tech/sources/cisco-blogs.md>)

Topics: [AI Networking](<https://devfeed.tech/topics/ai-networking.md>), [Agentic AI](<https://devfeed.tech/topics/what-is-agentic-ai.md>), [Network Operations](<https://devfeed.tech/topics/network-operations.md>), [ThousandEyes](<https://devfeed.tech/topics/thousandeyes.md>), [Cryptography](<https://devfeed.tech/topics/cryptography.md>), [Wi-Fi](<https://devfeed.tech/topics/wi-fi.md>)

Tags: [2026](<https://devfeed.tech/tags/2026.md>), [about](<https://devfeed.tech/tags/about.md>), [access-point](<https://devfeed.tech/tags/access-point.md>), [act](<https://devfeed.tech/tags/act.md>), [agentic](<https://devfeed.tech/tags/agentic.md>), [agentic-ai](<https://devfeed.tech/tags/agentic-ai.md>), [ai](<https://devfeed.tech/tags/ai.md>), [anchor](<https://devfeed.tech/tags/anchor.md>), [autonomous](<https://devfeed.tech/tags/autonomous.md>), [change](<https://devfeed.tech/tags/change.md>), [cisco-secure-networking](<https://devfeed.tech/tags/cisco-secure-networking.md>), [foundation](<https://devfeed.tech/tags/foundation.md>), [machine](<https://devfeed.tech/tags/machine.md>), [more](<https://devfeed.tech/tags/more.md>), [need](<https://devfeed.tech/tags/need.md>), [network](<https://devfeed.tech/tags/network.md>), [networking](<https://devfeed.tech/tags/networking.md>), [operations](<https://devfeed.tech/tags/operations.md>), [post-quantum-cryptography-pqc](<https://devfeed.tech/tags/post-quantum-cryptography-pqc.md>), [speed](<https://devfeed.tech/tags/speed.md>), [technology](<https://devfeed.tech/tags/technology.md>), [transparency](<https://devfeed.tech/tags/transparency.md>), [trust](<https://devfeed.tech/tags/trust.md>)

### AI overview

Cisco describes network capabilities intended to support trusted AgenticOps: ThousandEyes agents and synthetic tests are automatically provisioned on Wi-Fi 6E and Wi-Fi 7 access points, while Experience Metrics extends telemetry to application performance. Agentic Actions combines AI agents, reasoning, and workflows in Cloud Control, showing recommendations and results and letting operators control how actions run; it is in beta or controlled availability. New MCP servers aim to standardize access to cloud and on-premises network intelligence. Security updates include quantum-ready secure boot and communications, including over-the-air encryption on supported platforms, and Cisco's CW9176ISEC Wi-Fi 7 access point, aligned to CNSA 2.0. Cisco also expands Live Protect runtime vulnerability shields to supported wireless controllers, Campus Gateways, and Secure Routers, allowing protection while permanent fixes are tested and deployed without an immediate upgrade or reboot.

### Source excerpt

When technology change is advancing at machine speed, we need a stronger foundation in the network to establish transparency, anchor trust, and enable autonomous operations.

## Online Causal Configuration for Networked Systems via Doubly Robust Steady-State Learning

DevFeed: [Online Causal Configuration for Networked Systems via Doubly Robust Steady-State Learning](<https://devfeed.tech/articles/online-causal-configuration-for-networked-systems-via-doubly-robust-steady-state-learning-123439.md>)

Original publisher: [Read original article](<https://arxiv.org/abs/2611.09665>)

Author: Yuli Liu, Zhiheng Zhang

Published: 2026-11-09T05:00:00Z

Content type: article

Language: en

Sources: [arXiv: Machine Learning](<https://devfeed.tech/sources/arxiv-machine-learning.md>)

Topics: [Machine learning](<https://devfeed.tech/topics/machine-learning.md>), [AI Networking](<https://devfeed.tech/topics/ai-networking.md>), [Reinforcement learning](<https://devfeed.tech/topics/reinforcement-learning.md>)

Tags: [accounting](<https://devfeed.tech/tags/accounting.md>), [audit](<https://devfeed.tech/tags/audit.md>), [benchmark](<https://devfeed.tech/tags/benchmark.md>), [certificate](<https://devfeed.tech/tags/certificate.md>), [certificates](<https://devfeed.tech/tags/certificates.md>), [configuration](<https://devfeed.tech/tags/configuration.md>), [context](<https://devfeed.tech/tags/context.md>), [controllers](<https://devfeed.tech/tags/controllers.md>), [coverage](<https://devfeed.tech/tags/coverage.md>), [cs-lg](<https://devfeed.tech/tags/cs-lg.md>), [deployment](<https://devfeed.tech/tags/deployment.md>), [distribution](<https://devfeed.tech/tags/distribution.md>), [error](<https://devfeed.tech/tags/error.md>), [exploration](<https://devfeed.tech/tags/exploration.md>), [growth](<https://devfeed.tech/tags/growth.md>), [horizon](<https://devfeed.tech/tags/horizon.md>), [independent](<https://devfeed.tech/tags/independent.md>), [learning](<https://devfeed.tech/tags/learning.md>), [online](<https://devfeed.tech/tags/online.md>), [robust](<https://devfeed.tech/tags/robust.md>), [stat-ml](<https://devfeed.tech/tags/stat-ml.md>), [systems](<https://devfeed.tech/tags/systems.md>)

### AI overview

CAUSAL-UCB targets the value a network configuration would have without measurement exploration, which can change both controller action distributions and the steady-state context distribution. It selects configurations using optimistic causal-value bounds and an interference-aware, clipped doubly robust evaluator that issues predecision certificates separating statistical variation from calibrated identification, transport, and temporal-dependence errors. The paper gives finite-horizon inference and deployment-loss lower bounds, regret bounds under valid certificates and selected-information growth, and an anytime finite-sample certificate for a restricted known-local-interference model. In a high-mismatch benchmark (24 paired seeds, 200 rounds), target regret was 1.732 for CAUSAL-UCB versus 6.326 for certified IPS-UCB; deployment gaps were 30.607 versus 23.217, respectively. Audit coverage tests and trace-driven semi-synthetic studies assess certificate coverage and deployment trade-offs.

### Source excerpt

arXiv:2611.09665v1 Announce Type: new Abstract: Measurement exploration reveals how interacting local controllers behave but can alter the steady-state regime a network operator seeks to optimize. Exploratory rewards may therefore fail to identify a configuration's no-exploration value: under full exploration, identical logged distributions can conceal different target values. Identification requires accounting for both action-law mismatch and shifts in the steady-state context distribution. We introduce CAUSAL-UCB, which selects configurations using optimistic bounds on target causal values. Its interference-aware clipped doubly robust evaluator produces predecision certificates separating statistical variation from calibrated identification, transport, and temporal-dependence errors. For any finite horizon and predetermined schedule of measurement-exploration rates $\{\eta_t\}$, we prove lower bounds of $\Omega((1+\sum_t\eta_t)^{-1/2})$ on minimax inference error and $\Omega(\sum_t\eta_t)$ on measurement-induced deployment loss in a stationary, potentially slow-mixing family. We bound regret under valid certificates and selected-information growth, and establish an anytime finite-sample certificate with a shared radius for estimation, selection, and reporting in a restricted known-local-interference model. In a high-mismatch benchmark with 24 paired seeds and 200 rounds, CAUSAL-UCB attains target regret of $1.732$ versus $6.326$ for certified IPS-UCB, with deployment gaps of $30.607$ and $23.217$, respectively. Independent-audit coverage tests and trace-driven semi-synthetic studies further assess certificate coverage and deployment trade-offs.

## How Cisco Built an Enterprise AI Factory: Scaling Economics, Performance, and Security

DevFeed: [How Cisco Built an Enterprise AI Factory: Scaling Economics, Performance, and Security](<https://devfeed.tech/articles/how-cisco-built-an-enterprise-ai-factory-scaling-economics-performance-and-security-121718.md>)

Original publisher: [Read original article](<https://blogs.cisco.com/cisco-on-cisco/how-cisco-built-an-enterprise-ai-factory>)

Author: Jon Woolwine

Published: 2026-11-06T19:35:41Z

Content type: article

Language: en

Sources: [Cisco Blogs](<https://devfeed.tech/sources/cisco-blogs.md>)

Topics: [Cisco IT](<https://devfeed.tech/topics/cisco-it.md>), [Cisco](<https://devfeed.tech/topics/cisco.md>), [ai security](<https://devfeed.tech/topics/ai-security.md>), [observability](<https://devfeed.tech/topics/observability.md>), [Production Engineering](<https://devfeed.tech/topics/production-engineering.md>), [AI Networking](<https://devfeed.tech/topics/ai-networking.md>), [Enterprise Architecture](<https://devfeed.tech/topics/enterprise-architecture.md>), [enterprise data protection](<https://devfeed.tech/topics/enterprise-data-protection.md>), [LLM observability](<https://devfeed.tech/topics/llm-observability.md>), [scheduling](<https://devfeed.tech/topics/scheduling.md>), [Load Balancing](<https://devfeed.tech/topics/load-balancing.md>)

Tags: [agentic-workflows](<https://devfeed.tech/tags/agentic-workflows.md>), [agents](<https://devfeed.tech/tags/agents.md>), [ai](<https://devfeed.tech/tags/ai.md>), [ai-data-center](<https://devfeed.tech/tags/ai-data-center.md>), [ai-defense](<https://devfeed.tech/tags/ai-defense.md>), [ai-factory](<https://devfeed.tech/tags/ai-factory.md>), [amd](<https://devfeed.tech/tags/amd.md>), [analytics](<https://devfeed.tech/tags/analytics.md>), [cisco](<https://devfeed.tech/tags/cisco.md>), [cisco-it](<https://devfeed.tech/tags/cisco-it.md>), [cisco-on-cisco](<https://devfeed.tech/tags/cisco-on-cisco.md>), [cisco-silicon-one](<https://devfeed.tech/tags/cisco-silicon-one.md>), [cost](<https://devfeed.tech/tags/cost.md>), [economics](<https://devfeed.tech/tags/economics.md>), [enterprise](<https://devfeed.tech/tags/enterprise.md>), [enterprise-ai](<https://devfeed.tech/tags/enterprise-ai.md>), [infrastructure](<https://devfeed.tech/tags/infrastructure.md>), [it](<https://devfeed.tech/tags/it.md>), [mission](<https://devfeed.tech/tags/mission.md>), [performance](<https://devfeed.tech/tags/performance.md>), [scaling](<https://devfeed.tech/tags/scaling.md>), [security](<https://devfeed.tech/tags/security.md>), [splunk](<https://devfeed.tech/tags/splunk.md>), [support](<https://devfeed.tech/tags/support.md>)

### AI overview

Cisco says it moved from isolated AI clusters to a pooled, multi-tenant factory across its data centers as MyAgent and Cisco IQ workloads grew. A high-bandwidth network fabric pools GPU capacity; UCS-X servers with NVIDIA and AMD GPUs support workload-specific compute; and rack-mounted rear-door heat exchangers doubled compute density in the existing footprint. Cisco reports average GPU utilization rising from 15% to about 50%, active GPUs growing from roughly 250 to over 1,568 in under two years, and 99.99%+ availability for MyAgent and Cisco IQ. It says 80% of steady-state inference is most cost-efficient on-premises. Security uses AI Defense runtime guardrails, while Hypershield segmentation and Nexus Live Protect are described as forthcoming capabilities. Splunk tools provide infrastructure metrics, traces, logs, and event management; Agent Observability and Tokenomics are still being expanded. Next steps include agent-focused monitoring, dynamic resource pooling, GPU reclaims, and adaptive networking.

### Source excerpt

Discover how Cisco IT modernized its infrastructure to create a secure, scalable, and cost-efficient enterprise AI factory to support mission-critical AI.

## SpaceX Teams Up With Qualcomm to Build a Nationwide 5G Network

DevFeed: [SpaceX Teams Up With Qualcomm to Build a Nationwide 5G Network](<https://devfeed.tech/articles/spacex-teams-up-with-qualcomm-to-build-a-nationwide-5g-network-121762.md>)

Original publisher: [Read original article](<https://hackernoon.com/spacex-teams-up-with-qualcomm-to-build-a-nationwide-5g-network?source=rss>)

Author: Bruce Li

Published: 2026-11-06T19:00:05Z

Content type: news

Language: en

Sources: [HackerNoon](<https://devfeed.tech/sources/hackernoon.md>)

Topics: [5G](<https://devfeed.tech/topics/5g.md>), [Qualcomm](<https://devfeed.tech/topics/qualcomm.md>), [Network](<https://devfeed.tech/topics/network.md>), [Networking hardware](<https://devfeed.tech/topics/networking-hardware.md>), [AI Networking](<https://devfeed.tech/topics/ai-networking.md>), [huawei](<https://devfeed.tech/topics/huawei.md>)

Tags: [5g](<https://devfeed.tech/tags/5g.md>), [build](<https://devfeed.tech/tags/build.md>), [hackernoon-top-story](<https://devfeed.tech/tags/hackernoon-top-story.md>), [network](<https://devfeed.tech/tags/network.md>), [open-ran](<https://devfeed.tech/tags/open-ran.md>), [qualcomm](<https://devfeed.tech/tags/qualcomm.md>), [satellite-internet](<https://devfeed.tech/tags/satellite-internet.md>), [spacex](<https://devfeed.tech/tags/spacex.md>), [starlink](<https://devfeed.tech/tags/starlink.md>), [telecommunication](<https://devfeed.tech/tags/telecommunication.md>)

### AI overview

SpaceX plans to build a nationwide terrestrial 5G network under the Starlink Mobile brand, with tower deployments starting in 2027; no terrestrial launch date is announced. Qualcomm will supply Dragonwing radio and baseband platforms and Layer 1 software, while SpaceX is expected to build the equipment and develop higher network layers and core software. SpaceX has assembled about 80 MHz of spectrum, far less than T-Mobile's nearly 400 MHz, making efficient use of spectrum--especially 64T64R FDD massive MIMO--a central technical challenge. The plan pairs towers for urban capacity with satellites for broader coverage and aims for seamless handoffs. The article identifies major hurdles: potentially tens of billions in build costs, tower leases, permits, crews, backhaul, phone certification and competition. Qualcomm's approach and the overall technology are presented as plausible, but execution and commercial timing remain uncertain.

### Source excerpt

SpaceX put thousands of routers in orbit is now coming down to earth: steel towers, rooftop leases, zoning boards and all for 5G land network.

## Worst-Case Discovery and Runtime Protection for RL-Based Network Controllers

DevFeed: [Worst-Case Discovery and Runtime Protection for RL-Based Network Controllers](<https://devfeed.tech/articles/worst-case-discovery-and-runtime-protection-for-rl-based-network-controllers-120016.md>)

Original publisher: [Read original article](<https://arxiv.org/abs/2605.04373>)

Author: Hongyu H\\\`e, Minhao Jin, Maria Apostolaki

Published: 2026-11-06T05:00:00Z

Content type: article

Language: en

Sources: [arXiv: Artificial Intelligence](<https://devfeed.tech/sources/arxiv-artificial-intelligence.md>), [arXiv: Networking](<https://devfeed.tech/sources/arxiv-networking.md>)

Topics: [AI Networking](<https://devfeed.tech/topics/ai-networking.md>), [Network](<https://devfeed.tech/topics/network.md>), [rule-engine](<https://devfeed.tech/topics/rule-engine.md>)

Tags: [congestion-control](<https://devfeed.tech/tags/congestion-control.md>), [controllers](<https://devfeed.tech/tags/controllers.md>), [cs-ai](<https://devfeed.tech/tags/cs-ai.md>), [cs-ni](<https://devfeed.tech/tags/cs-ni.md>), [cs-sy](<https://devfeed.tech/tags/cs-sy.md>), [discovery](<https://devfeed.tech/tags/discovery.md>), [eess-sy](<https://devfeed.tech/tags/eess-sy.md>), [formal-verification](<https://devfeed.tech/tags/formal-verification.md>), [framework](<https://devfeed.tech/tags/framework.md>), [gap](<https://devfeed.tech/tags/gap.md>), [inference](<https://devfeed.tech/tags/inference.md>), [intervene](<https://devfeed.tech/tags/intervene.md>), [lightweight](<https://devfeed.tech/tags/lightweight.md>), [logic](<https://devfeed.tech/tags/logic.md>), [lower-bound](<https://devfeed.tech/tags/lower-bound.md>), [network](<https://devfeed.tech/tags/network.md>), [networking](<https://devfeed.tech/tags/networking.md>), [protection](<https://devfeed.tech/tags/protection.md>), [rl](<https://devfeed.tech/tags/rl.md>), [runtime](<https://devfeed.tech/tags/runtime.md>), [runtime-protection](<https://devfeed.tech/tags/runtime-protection.md>)

### AI overview

ReGuard finds conditions where an RL network controller performs poorly despite better performance being achievable under the same conditions, then adds runtime protection without retraining. It frames discovery as bilevel regret maximization to certify a lower bound on the worst-case performance gap, and converts discovered counterfactual trajectories into lightweight logic rules that intervene only when a risky state is detected. Across Pensieve, Sage, and Park, it finds scenarios with 43-64% performance loss, exposes gaps 57% to six times larger than the strongest baselines, and reduces gaps by 79-85% in the first protection iteration. The abstract says protection also improves performance across a wider range of network conditions.

### Source excerpt

arXiv:2605.04373v2 Announce Type: replace-cross Abstract: RL-based controllers achieve strong average-case performance in networking tasks such as congestion control and adaptive bitrate streaming. Yet their performance can degrade severely under network conditions where strong performance is still achievable. Identifying such conditions and quantifying the resulting performance gap is intractable by enumeration, while the sequential and closed-loop nature of RL controllers makes formal verification methods impractical. We present ReGuard, a framework that discovers worst-case scenarios for a given RL controller and protects it against them at inference time without retraining. ReGuard formulates scenario discovery as a bilevel regret-maximization problem, certifying a lower bound on the controller's worst-case performance gap. ReGuard then analyzes the discovered counterfactual trajectories and compiles them into lightweight logic rules that intervene only when a risky state is detected, leaving the controller's behavior unchanged otherwise. We evaluate ReGuard across three RL-based network controllers: Pensieve, Sage, and Park. ReGuard discovers scenarios in which each controller suffers a 43-64% performance loss relative to what is achievable under the same conditions. ReGuard exposes performance gaps 57% to 6 times larger than those found by the strongest baselines and shrinks those gaps by 79-85% in its first iteration via lightweight rule-based protection, iteratively driving the remaining gap toward nominal levels. ReGuard's protection extends beyond the scenarios it discovers, improving performance across a wide range of network conditions.

## Invariant Discovery for Networked Systems

DevFeed: [Invariant Discovery for Networked Systems](<https://devfeed.tech/articles/invariant-discovery-for-networked-systems-120064.md>)

Original publisher: [Read original article](<https://arxiv.org/abs/2607.22944>)

Author: Hongyu H\\\`e, Alexander Krentsel, Sylvia Ratnasamy, Maria Apostolaki

Published: 2026-11-06T05:00:00Z

Content type: article

Language: en

Sources: [arXiv: Artificial Intelligence](<https://devfeed.tech/sources/arxiv-artificial-intelligence.md>), [arXiv: Machine Learning](<https://devfeed.tech/sources/arxiv-machine-learning.md>), [arXiv: Networking](<https://devfeed.tech/sources/arxiv-networking.md>)

Topics: [AI Networking](<https://devfeed.tech/topics/ai-networking.md>), [Automated reasoning](<https://devfeed.tech/topics/automated-reasoning.md>), [data-engineering](<https://devfeed.tech/topics/data-engineering.md>)

Tags: [ai](<https://devfeed.tech/tags/ai.md>), [applications](<https://devfeed.tech/tags/applications.md>), [automatic](<https://devfeed.tech/tags/automatic.md>), [cs-ai](<https://devfeed.tech/tags/cs-ai.md>), [cs-lg](<https://devfeed.tech/tags/cs-lg.md>), [cs-ni](<https://devfeed.tech/tags/cs-ni.md>), [cs-sc](<https://devfeed.tech/tags/cs-sc.md>), [data-types](<https://devfeed.tech/tags/data-types.md>), [discovery](<https://devfeed.tech/tags/discovery.md>), [false-positives](<https://devfeed.tech/tags/false-positives.md>), [formal](<https://devfeed.tech/tags/formal.md>), [full](<https://devfeed.tech/tags/full.md>), [grammar](<https://devfeed.tech/tags/grammar.md>), [guarantees](<https://devfeed.tech/tags/guarantees.md>), [induction](<https://devfeed.tech/tags/induction.md>), [input](<https://devfeed.tech/tags/input.md>), [inspect](<https://devfeed.tech/tags/inspect.md>), [invariant](<https://devfeed.tech/tags/invariant.md>), [invariants](<https://devfeed.tech/tags/invariants.md>), [llm](<https://devfeed.tech/tags/llm.md>), [logic](<https://devfeed.tech/tags/logic.md>), [logical](<https://devfeed.tech/tags/logical.md>), [metadata](<https://devfeed.tech/tags/metadata.md>), [systems](<https://devfeed.tech/tags/systems.md>)

### AI overview

Autogram discovers network telemetry invariants by using an LLM to infer a compact grammar from variable metadata, then conducting a deterministic statistical search within that grammar. If the grammar cannot recover trusted invariants in held-out telemetry, the LLM iteratively widens it. Logical and statistical checks verify candidates, making results auditable with formal guarantees within the induced grammar. Evaluations on public and production WAN telemetry reportedly recovered expert-derived invariants with full recall and low false positives. The abstract frames fully open-ended discovery as an unresolved direction.

### Source excerpt

arXiv:2607.22944v2 Announce Type: replace-cross Abstract: Invariants, the relations expected to hold among measured signals of a network, underpin applications from verification to traffic generation, telemetry imputation, and input validation. Yet, writing them by hand demands rare expertise in both formal logic and networking. Testing all possible combinations of telemetry variables statistically is intractable at network scale, while existing automatic tools still require a human-authored grammar of admissible invariants and learn only exact, "hard" rules that break under real-world noise. Our key idea is to use an LLM to inspect variable metadata, classify counter data types, and predict candidate syntactic relationships that form a compact grammar, followed by a deterministic statistical search within that grammar. If the initial grammar fails to recover trusted known invariants on held-out telemetry, Autogram instructs the LLM to iteratively widen the grammar until recovery succeeds. By confining the LLM to grammar induction and re-induction while delegating candidate verification to logical and statistical checks, Autogram turns non-deterministic AI proposals into auditable invariants with formal guarantees within the induced grammar. We design and prototype Autogram and evaluate it on both public and production WAN telemetry, recovering expert-derived invariants with full recall and low false positives. We close with a discussion of open problems on the path toward fully open-ended invariant discovery.

## Deep reinforcement learning-based optimization of multi-RAT handovers

DevFeed: [Deep reinforcement learning-based optimization of multi-RAT handovers](<https://devfeed.tech/articles/deep-reinforcement-learning-based-optimization-of-multi-rat-handovers-121324.md>)

Original publisher: [Read original article](<https://arxiv.org/abs/2611.07390>)

Author: Pavan Kumar Mangipudi, Satya Krishna Pothapragada, Alina Zare, Janise McNair

Published: 2026-11-06T05:00:00Z

Content type: article

Language: en

Sources: [arXiv: Networking](<https://devfeed.tech/sources/arxiv-networking.md>)

Topics: [Reinforcement learning](<https://devfeed.tech/topics/reinforcement-learning.md>), [Artificial Intelligence](<https://devfeed.tech/topics/ai.md>), [AI Networking](<https://devfeed.tech/topics/ai-networking.md>), [Neural Network](<https://devfeed.tech/topics/neural-network.md>)

Tags: [5g](<https://devfeed.tech/tags/5g.md>), [case-study](<https://devfeed.tech/tags/case-study.md>), [cs-ni](<https://devfeed.tech/tags/cs-ni.md>), [deep-reinforcement-learning](<https://devfeed.tech/tags/deep-reinforcement-learning.md>), [delay](<https://devfeed.tech/tags/delay.md>), [markov](<https://devfeed.tech/tags/markov.md>), [offloading](<https://devfeed.tech/tags/offloading.md>), [sdn](<https://devfeed.tech/tags/sdn.md>), [testbed](<https://devfeed.tech/tags/testbed.md>), [throughput](<https://devfeed.tech/tags/throughput.md>)

### AI overview

ReFlexMRS uses an offline-trained Deep Q-Network to select multi-RAT handover targets using network measurements rather than relying primarily on signal strength. Evaluated on an SDN-enabled 5G-WiFi testbed, it retained about 90-95% of RSSI-based baseline throughput while reducing total handovers by up to 18% and WiFi-to-5G handovers by up to 13%, with improved link delay, stability, and WiFi offloading while meeting required RSSI levels. These results are from the described case study.

### Source excerpt

arXiv:2611.07390v1 Announce Type: new Abstract: A major evolutionary goal and challenge for next-generation wireless systems is seamless multiple radio access technology (multi-RAT) mobility. In a multi-RAT environment, handover decisions must account for both user quality of service and network priorities, including signal quality, resource availability, energy consumption, latency, and security. Although WiFi/cellular mobility is well explored in research and standardization, many conventional handover mechanisms continue to rely primarily on received signal strength. Such decisions may overlook link delay, congestion, and the distinct operating characteristics of heterogeneous RATs. In this work, we present the Reinforced Flexible Multi-RAT Selection framework (ReFlexMRS), a configurable Deep Reinforcement Learning (DRL)-based approach for multi-criteria handover target selection. ReFlexMRS is evaluated using an SDN-enabled multi-RAT testbed. The decision problem is formulated as a Markov Decision Process with a finite action space and is addressed using a Deep Q-Network (DQN) with convolutional layers that extract relationships among network measurement features. The agent is trained offline from collected network observations using reward guided Q-value updates, then used to select handover targets during inference. A 5G-WiFi case study demonstrates that ReFlexMRS can trade a small amount of throughput for substantial improvements in link delay, handover stability, and WiFi offloading. Compared with RSSI-based selection, the proposed formulations retain approximately 90-95% of the baseline throughput while reducing total handovers by up to 18% and WiFi-to-5G handovers by up to 13%, while maintaining the required RSSI levels.

[Next page](<https://devfeed.tech/topics/ai-networking.md?cursor=WyIyMDI2LTExLTA2VDA1OjAwOjAwKzAwOjAwIiwgIjU1NDVhMDFjLTJmNmUtNDEwOC1hOGY0LTQyZWVlZDg3NmZjOCJd>)