I spent yesterday on a site survey for a customer who needs a temporary connection, and needs it working from the first hour. Standing under the mast in front of me, the question was the one I get on nearly every survey: how do we get a better signal here? It is a fair question. It is also the wrong one to build on.

Here is the uncomfortable part. We can get that signal about as good as it gets: best network, best position, cleanest line to the tower. And it still does not solve the connectivity challenge, because a single signal, however strong, is one point of failure. Strength is not resilience. The moment that network congests, or has a bad backhaul day, or simply drops, your perfect signal is a dead connection.

Strength is not resilience

A strong signal on one network tells you how good things are right now, on that network, in that spot. It tells you nothing about what happens when that network has a bad hour. Optimising a single link is polishing a single point of failure, and for a deployment that has to work, that is the wrong thing to optimise.

Bond several networks instead

So I do not chase one signal. I bond several cellular networks into one connection: different carriers, different masts, combined into a single logical link. If one path wobbles, the others carry the load. Where it helps, we can bring Starlink into the same bond, so cellular and satellite back each other up rather than sitting as separate options someone has to switch between by hand.

The bond runs over an encrypted SpeedFusion tunnel to a cloud hub, and the hub is where the useful properties come from. Here is what that buys the customer:

  • A static public IP that lives at the cloud hub, so it stays put no matter which underlying links are live. Carriers can come and go and the address the customer is reached on does not change.
  • A routed range of public IP addresses across the same bonded connection, for deployments that need more than one address behind the link.
  • A priority path for video calls, with WAN smoothing so the picture holds together when a link starts to degrade instead of breaking up.
  • Failover built to hold the session up across a link change, rather than dropping the call and making everyone dial back in.
  • Remote visibility of the whole setup through InControl2, so we can see how each link is behaving and act on it without a site visit.

And because none of this depends on a fixed line, there is no waiting on an Openreach install date. The kit can stand up the same day it arrives on site.

We measure, we do not guess

The part I am proudest of is how we land on the configuration. We do not read a signal bar and make a judgement call. We built our own survey tool that plugs straight into the test kit on site, runs every connectivity option and every combination of carriers and paths, and hands back the configuration that performs best in that exact spot. The customer gets a setup proven on the ground at their location, not a best guess off a coverage map. That tool, and the lab work behind it, is what turns "the signal looks fine here" into "this is the config, and here is why".

Where this fits

This is the approach for a site, an event, or any temporary deployment that simply has to work and cannot wait for a fixed line: broadcast and live events, construction and remote sites, a pop-up office, an emergency standup. Anywhere the connection matters and the location is not permanent. So no, I am not going to spend the day chasing a stronger signal on one network. I am bonding several, letting the data pick the winner, and handing over something measured.