STAIRWAI to COSMOS: Wireless for AI (W4AI) Workshop

Help the STAIRWAI project build a plan for lowering the barrier to research on AI solutions for wireless communications and networks.

The first STAIRWAI workshop asked how AI can improve wireless systems. This one turns the question around: if AI applications become the dominant traffic and the dominant workload, what do wireless networks — and the testbeds we prototype them on — have to become?

Draft program

8:30 – 9:00 · Registration, light breakfast, poster set-up

9:00 – 9:10 · Welcoming remarks — Narayan Mandayam, Director, WINLAB

9:10 – 10:00 · Keynote 1: the vendor and operator view — Mikael Prytz, Ericsson

10:00 – 11:10 · Poster session and coffee

11:10 – 12:00 · Keynote 2: AI Glasses - Building Technology for Human Connection — Arpit Mehta, Meta

12:00 – 1:20 · Lunch

1:20 – 2:10 · Keynote 3: AI in the field, over real networks — Ranveer Chandra, Microsoft

2:10 – 3:00 · Keynote 4: Building Trust in Industrial AI at the Wireless Edge — Charif Mahmoudi, Siemens

3:00 – 3:10 · Framing the breakouts — Jorge Ortiz, Rutgers

3:10 – 4:20 · Parallel breakouts and coffee

4:20 – 4:40 · Summary, priorities, and the road to a full center proposal — closing remarks by Jorge Ortiz, Rutgers

Breakouts: A. Wireless-native AI — split inference, distributed and over-the-air learning, latency budgets · B. Verticals — what AgTech, manufacturing and XR actually need from the network · C. What makes a testbed AI-ready — datasets, benchmarks, reproducibility, tooling.

Verticals in focus

• Agriculture and the outdoor edge — intermittent connectivity, energy budgets, inference where backhaul is not an option.
• Advanced manufacturing — deterministic latency, closed-loop control, what a private network owes an AI controller.
• XR, wearables and AR glasses — splitting computation between a device that cannot get hot and an edge that cannot be far away.
• Smart streetscapes and mobility — multimodal street-level sensing, with the COSMOS West Harlem deployment as the live setting.

About STAIRWAI

STAIRWAI to COSMOS — Sensor-enabled Testbed for Advancing Innovative Research in Wireless+AI — is an NSF planning grant under the AI-Ready Test Beds program. Its job is to work out, with the research community, what a full center on Wireless and AI should build first: streamlined dataset collection, shared benchmarks, a framework for deploying and evaluating wireless-aware AI, and infrastructure for real-time acceleration at the edge.

The first workshop, AI for Wireless (AI4W), ran at WINLAB on 30 January 2026 with speakers from Ericsson, NVIDIA, AT&T and Qualcomm.

Call for posters and demos

The poster session is the heart of the morning and is open to students, postdocs and industry researchers. Send a title, authors and a short abstract to reserve a board — easels and posterboards are provided. Demos are welcome; tell us what you need and we will find you a table, power and spectrum.

Arriving Thursday?

The COSMOS³ Community Workshop on Next-Generation Testbed Capabilities runs the afternoon before — Thursday 17 September, 12:00–5:00 PM, same venue — on FR3, the digital twin and street-level sensing, followed by a joint reception open to both workshops. Register for both: eventbrite.com/e/1998936074609

Workshop dinner, Thursday evening: the off-site dinner is booked on the Thursday page only, as a separate free ticket. If you are coming Friday only and would like to join us Thursday evening, register on the Thursday page and add the dinner ticket there. Please do not book the dinner on both pages.

Organizers

Anand D. Sarwate (PI, Rutgers), Jorge Ortiz (Rutgers), Bo Yuan (Rutgers), Yingying Chen (Rutgers), Ivan Seskar (Rutgers/WINLAB), Narayan Mandayam (Director, WINLAB). Coordination: Taqiya Ehsan (Rutgers).

Supported by NSF award 2509233, Planning: AI Ready: STAIRWAI to COSMOS, hosted on the NSF PAWR COSMOS / COSMOS³ platform.