A wireless network design is a prediction. The expensive part begins after the access points, radios, antennas, and cables are where the drawing said they should be.
Eino’s Site Survey launch this week exposes that gap. Its survey kit measures private cellular, public cellular including DAS, and Wi-Fi, then puts those measurements onto the same 3D digital twin used for design and monitoring. The useful idea isn’t the backpack full of radios; it’s closing the loop between promised performance and delivered performance.
I spent years around automated test and measurement systems, so this feels familiar. A specification without a verification method is partly a wish, something hardware teams learned long ago and software products tied to physical infrastructure occasionally relearn after deployment.
Acceptance criteria belong in the model
Wireless design tools predict coverage from geometry, materials, antenna placement, transmit power, interference assumptions, and propagation models. Then reality gets a vote.
A wall may be modeled with the wrong material, an antenna can land somewhere different from the drawing, furniture moves, equipment appears, and firmware changes. Meanwhile, a beautifully colored heatmap doesn’t complain.
Eino’s approach matters because measured performance lands back on the site’s digital twin instead of becoming a disconnected survey artifact that has to be reconciled later. Its documentation puts validation beside simulation, heatmaps, field surveys, monitoring, and reporting, making the design model something closer to a testable product contract.
For a product manager, I’d write acceptance criteria in the same coordinates as the promise: location, device class, network technology, workload, and minimum performance. “Good Wi-Fi” is mush. A requirement tied to a warehouse zone, a specific workload, and an AGV’s connectivity needs can actually fail a test and produce evidence someone can act on.
Watch the handoffs
Enterprise infrastructure often has a design tool, an installation workflow, a survey tool, and then a monitoring system. Handoffs hide product debt.
Each one can quietly change identifiers, assumptions, coordinate systems, ownership, or even what “passing” means after the equipment is bolted into place and the installer leaves. If the planned access point is AP-17 in design, a field engineer shouldn’t have to reverse-engineer which installed radio it became before comparing prediction with measurement.
A failed survey shouldn’t disappear into a PDF. The discrepancy should survive into monitoring, attached to the same place, asset, requirement, and history that created the original prediction.
There is a counterweight: one integrated model can become its own source of false confidence when the underlying site data is stale. Eino addresses part of that problem with LiDAR and drone-based site data that can be refreshed on a schedule, but freshness still has to be treated as data, not decoration.
Models age quietly, and I’ve written before that data products break when meaning changes; physical infrastructure adds another failure mode because the world itself changes underneath the model.
A design tool becomes more valuable when it can prove where its design was wrong. That’s a slightly uncomfortable product feature, and exactly why verification belongs beside prediction.
Comments
Comments are powered by GitHub Discussions through Giscus.