← All posts

November 5, 2026 · Technology

Designing Around the Link You Can Count On

By Sam Amyett

Most of the conversation around tactical communications right now is about bandwidth, and for good reason. The radios that get the headlines, Silvus being the obvious example after Motorola paid $4.4 billion for the company last year, are built to push full-motion video across a mesh, and they do it well. The catch is that video sets the floor for everything else - the waveform, the power budget, the size of the radio, and ultimately a price tag that tends to land in the five figures per node.

A little background on why I look at this differently. I started out as a CSO aboard the E-6B, where the mission was keeping a communications link alive in a post-nuclear environment. The data rates we relied on were a small fraction of what a modern MANET radio pushes, and nobody treated that as a weakness, as our tactics were built around survivability first. From there, I used that operational experience to lead engineering teams at the PMA, where SWaP-C (size, weight, power, and cost) was always a concern, and where I saw firsthand how every requirement you add shows up in how many of something you can actually field.

So my argument, and I know it's a slightly unpopular one, is that if you strip video out of the requirement, a whole segment of the communications world opens up that doesn't get nearly as much attention as the sexy stuff. LoRa moves anywhere from a few hundred bits to a few tens of kilobits per second, and FLRC, its faster cousin on Semtech's newest chips, tops out around 2.6 megabits per second. That sounds laughable next to a MANET radio, but it is plenty for what a small unit needs to say most of the time: position reports, text, sensor alerts, status updates, and command messages, nearly all of which fit in a few hundred bytes.

On the hardware side, a LoRa radio module costs tens of dollars, runs for days on a small battery, and fits in a pocket. When the cost of a node drops by two or three orders of magnitude, the way you think about the network changes. You stop protecting each radio like a crown jewel and start treating nodes as consumables, so you can scatter relays across a ridgeline, leave sensors behind as tripwires, and accept that some percentage of them will be lost, jammed, or captured. That's the same logic that made cheap drones so effective over the last few years, and it's the logic behind efforts like Replicator, where quantity itself becomes a form of resilience.

Of course, there's no free lunch here. Low data rates mean long time-on-air, which limits how many nodes can share a channel before transmissions start colliding, and a narrowband signal sitting on a known frequency isn't exactly hard to find. Range and data rate also trade directly against each other, so the faster FLRC modes give back a good chunk of the distance that makes LoRa attractive in the first place. And some missions genuinely need video, so the high-end radios still have a job to do.

What I think the community is missing is the tier underneath them. If we wrote requirements around the constraint instead of around the most demanding use case, we could field a layer of communications cheap enough to buy by the thousands and simple enough to hand to anyone, which makes it survivable in a way an expensive radio never will be, because no single node matters. The drone world figured this out by accepting less capability per airframe in exchange for far more airframes. I think the communications world is overdue for the same trade... affordable mass, applied to the network.