Building Reliable Networks in Impossible Locations: Lessons from a Live Event

Most network planning happens with the benefit of time — you can survey a building, run cabling, and test before anyone relies on it. Every so often, you get a project with none of that runway. One of mine was a live auction event held inside a sports arena in Detroit, MI that was in the process of closing down. Point-of-sale needed to run on 250 iPads. The building had no existing, usable network connectivity, and its concrete construction made getting a signal through the space genuinely difficult.

The core problem: concrete doesn't cooperate

Dense concrete structures are notoriously hostile to wireless signal — the material absorbs and reflects radio waves in ways that make consistent coverage hard even in buildings designed with networking in mind. This arena wasn't designed with that in mind at all, and it was on its way to being decommissioned, so there was no institutional network infrastructure left to lean on.

Building for the event, not for permanence

Because this was a one-time event rather than an ongoing installation, the goal wasn't a polished, permanent network — it was a reliable, temporary one that would hold up for the duration of the auction. That reframes the whole approach: prioritize getting a stable connection to where the point-of-sale devices actually were, rather than full-building coverage.

What made it work

• On-site testing, not assumptions. With a building this unpredictable, I tested signal strength and stability directly at the point-of-sale locations rather than planning purely on paper.
• Positioning equipment around the concrete, not through it. Where signal had to pass through structure, coverage suffered. Positioning to work with the building's layout — line of sight where possible — mattered more than raw equipment power.
• Redundancy where it counted. For an event processing live payments, "mostly working" isn't good enough. The network needed to hold up reliably for every transaction, not just most of them.

The payoff: no manual batching

The alternative to a working live connection is manual card batching — processing transactions offline and reconciling them later, which is slower, more error-prone, and a worse experience for both staff and attendees. Because the network held up, all 250 point-of-sale devices processed payments in real time throughout the event, with checkout staying completely seamless.

The broader lesson

Not every networking project happens in a building built for networking. Sometimes the job is making a space that was never designed for reliable connectivity work anyway, on a deadline that doesn't move. That kind of problem-solving under real constraints is exactly the situation I'm built for.