Transcription
I just reversed engineered a $20,000 military IP mesh radio using open source parts totaling only $106.23. Sure, the government has been known to overpay for things, but I could build about 200 of these for the price of a single unit of theirs. And last week I posted about this on Reddit, and a few days later the US military contacted me. Fortunately, not to shut me down, but just looking for advice. Which I gotta say, felt pretty good.
The true power of ATAC is being able to get situational awareness in the palm of your hand. Usually that was reserved to your talks, your tactical operations centers where commanders and planners could see the big picture. But when you enable warfighter, which is something I think that the current army and military, marines, air force, I mean they are trying to get there, but they're just also kind of, because of their own procurement processes, they're going about in a very very slow manner. Like you gotta test, and everybody has an opinion about some piece of kit. So it's very challenging in that aspect. But the way special operations used it, and now you can go onto a target and then you can see where everybody's at, you can coordinate fires. Hell, there's some apps in there where you can send through either link 16 or saddle straight to a fast mover to drop a bomb on target. We used it in other ways, such as helicopter interdictions, the bloodhound tool is very very useful, we know how long it would take us to get somewhere, we would be able to pull the metadata off of video feeds so that we know exactly where that vehicle was, and it would cut that time down. We also used it for what we call "mad minutes" where we had to shift target, we had to come up with a plan really quickly, usually it was a bunch of guys huddled up against a helo and coming up with a hasty plan. This, we could do it all from our phone, we could watch the video right in the palm of our hand. So when you network ATAC, it becomes extremely more powerful.
But what did I just build? And what exactly does it enable? And why is it ruffling so many feathers? Freedom tech is finally catching up. I'm talking about decentralized open source tools that you actually own, built by companies that encourage tinkering instead of suing you for it. That's something I can get behind. It's that southern engineering grit I admire, making gear tougher, smarter, and more useful by pushing it past the limits. And being able to sideload, extend, and overclock my tech to do things nobody thought is possible, that's exactly what this channel is all about. This device creates a private portable IP network, but let's see how far it can go and what bandwidth it can support. I'm heading into a signal dead zone and I'm bringing nothing but a starlink and two wi-fi Halow nodes. Can I create a fully functional long-range wireless internet network from scratch and actually use it for real-world practical uses like chat messages, GPS coordination, push-to-talk, ATAC, video streaming, and internet connections? This feels like reliving the glory days of walkie-talkies as a kid, except now it's upgraded to the latest wi-fi standards. We're talking 802.11s mesh and 802.11ah Halow. So instead of just goofing around in the backyard, you're running a legit long-range encrypted internet network and the DoD may or may not want a word with you. If this doesn't work, I may have wasted a lot of time and a lot of money, but if it does work, this changes everything.
But before breaking into it, just what the heck is this thing? Okay, let's keep this dead simple. It's a mainnet, which stands for mobile ad hoc network. It sounds complicated, but all it means is it's a self- contained portable node that links up with others just like it. Each one of these nodes is built around a small computer, usually a raspberry pi, running an open source router software called OpenWrt. That's the brains. On board, you still get traditional wi-fi, dual band 2.4 and 5 gigahertz, which gives you a couple hundred feet of coverage, which is fine if you're in the same cabin or campsite, but it's not enough if you need to reach miles down the trail. That's where Halow comes in. When short-range wi-fi can't reach, Halow links your nodes to other nodes and with clear line of sight, it can stretch several miles. That's the backbone that ties the whole mesh together. And here's the part most people miss. Once you have your radio, you don't really need special gear to use it. You just connect your phone over wi-fi the same way you'd connect on an airplane network. No SIM cards, no carrier, no big tech permission slip. And once you're in, you can run apps like ATAC, which automatically discovers other nodes, push-to-talk radios, chat relays, and if one node has an uplink, say Starlink, literally any other internet app can be used. And because these nodes run on 802.11s mesh, every node that joins strengthens and extends the network. Add more nodes and the coverage grows. Picture a convoy of trucks on an off-road trail, each carrying a node. Now the whole team has comms, live positioning, even internet with no towers, no ISP, and no monthly bill. That's a mainnet.
Everything I'm showing here, gear setup, config, it's all in a field manual I put together linked below. I'm calling it the Haven IP Mesh Radio. It's an open source alternative to radios like the MPU5, except instead of $20,000 and a procurement officer, you just need a raspberry pi and some grit. Now I've had this guide live for about a week and people are already building with it. But starting today for the next 30 orders, I'm including a bonus 30 minute one-on-one call with me to go over your setup, answer questions, or just talk shop. I'm capping this at 30 because I literally can't take more calls without melting my calendar. So if you're watching this and the link's still active, jump in. And super quick, if you like the tech but don't want to build it out yourself, Trey over at MoroseX is building these to order. These are IP Mesh Radios, very similar to what we're talking about here. And he's super knowledgeable and the design is, I mean, you just can't beat this design. It's incredible.
So now that we've got our two nodes set up, we're going to want to see what kind of data rate and bandwidth we can get. So Halow supports different channel widths, very similar to spreading factor with Meshtastic, which you can choose right in the OpenWRT UI. All right, so I do want to run you through the anatomy of this thing pretty quickly here. So it's pretty kind of modular in the sense that like any component of this you could probably swap out for something else, but I'll run you through what I have here. So the motherboard here is a Raspberry Pi 4. We use the 4, not the 5, because some of the OpenWRT firmware doesn't quite work on the 5 yet. As soon as it does, we'll start using the 5, because the 5 is more powerful, frankly. And then we have a hat for a mini PCIe card. And the PCIe card or the mini PCIe card that we want to use here is what introduces the Halow chip. This thing's not expensive. It's about 15 bucks for the card there, and then it's maybe like 20 bucks for the hat. And so we add that on to the Raspberry Pi. You can use any sort of antenna that is sub-gigahertz. Right now I'm just playing around with this big fancy antenna, but frankly, you can get away with antennas like this. You can get away with whip antennas. Any of your mesh-tastic antennas will work because they're in the same range. And then it just uses an SMA to U.FL connector here, so you can really do that however you want. And then for power, just to make it mobile, I'm using this WaveShare hat that actually relies on pogo pins. You can see them right there. So you just put it on top, and then you put these batteries in. These are 21700s. Each one is 5000 milliamperes, so a total of 10,000 milliamperes capacity, which is pretty solid, actually. And you really don't have to move this around. If you want to charge it, you just plug in the USB-C, but then you have your kind of self-contained unit here. And then the only other component...this is a Wi-Fi dongle. This brings 2.4 and 5 gigahertz Wi-Fi to the setup. Technically, the Raspberry Pi can support that on board, but it can't do Halow with that, unfortunately. So as kind of an intermediary solution, we could use something like this. In the future, I want to use a proper hat with a real Wi-Fi chip. This is an Intel chip, and you could put real 2.4 and 5 gigahertz antennas, and it could be a lot cleaner of a setup. Frankly, it's such early days that I don't even have a proper case for this. The individual who's working on OpenMainnet is working on a 3D printing design case, so we could use that. Or you can kind of just get creative with what you want. But that's the basic setup of this thing, so we're not talking a whole lot of money to get this thing up and running.
So I'm actually getting Wi-Fi out here, but it's not coming across the standard Wi-Fi connection. It's coming across Ethernet, and it's coming across the 915 megahertz antenna. You can see it's pretty slow, but it is loading, and really we just wanted to do some tests, so we'll run those tests. Okay, so I'm watching one of my own YouTube videos, and I'm doing it by way of this 915 megahertz sub-gigahertz antenna that is connected to a Halow chip on a Raspberry Pi, and then going into my laptop over Ethernet. But what's interesting is the router that this is connecting to is actually way, way up there, and it's able to pass my home wireless connection to this device because it's using the Halow antenna with a one megahertz channel width, which means it can go super far. Here's what I achieved with my performance tests. At 8 megahertz, I measured about 15 megabits per second, both up and down, which is right on the theoretical ceiling for a Wi-Fi Halow at that channel width. At 4 megahertz, throughput landed at 7.3 megabits, again matching the expected max. At 2 megahertz, I actually clocked 4.4 megabits, just a hair above textbook 3.9, well within margin for how iPerf reports traffic. And at the narrowest 1 megahertz channel, I got 1.8 megabits, which lines up almost exactly with the 1.95 megabits theoretical limit.
Out of the box, these Halow radios ship at about 21 dbm. That's roughly 125 milliwatts. Pretty modest, not much more than a home router. But with the OpenMainnet firmware, which is the open source package that this is based on, that ceiling gets raised to 27 dbm, which is around 500 milliwatts. That's almost four times the wattage, which makes a huge difference in link stability and range. And just for context, the legal limit for unlicensed 900 megahertz gear in the US is 30 dbm, which is the equivalent of one full watt. That's the absolute max you can transmit without a special license. So at 27 dbm, we're sitting right near the top of what's possible for a civilian build, just shy of the legal ceiling. You get the benefits of serious range and resilience while staying inside of the regulatory box. Now, one quick myth buster. Bandwidth isn't set by the antenna. It's set by the modulation and channel width in the radio. But the antenna does matter for range. A higher gain antenna can focus the signal, letting you hold a link further out or punch through interference. So swapping antennas won't make your network faster, but it can make it go a whole lot farther. And OpenWRT gives you a full suite of diagnostic throughput testing, range tests, and live packet stats. So the takeaway is simple. In real world conditions, my test basically hit the ceiling of what 802.11aH can do. Wide channels gives you max bandwidth. Narrow channels stretch your range.
So now what can we do with this? That's where it gets interesting. The most obvious application that comes to mind is ATAC, which is Android Tactical Awareness Kit, which is used by the military and domestic search and rescue teams for operatives, which requires several team members staying connected in highly dynamic situations. Now, there's a number of different ways to leverage ATAC, but with this setup, I was able to get the devices connected from mainnet one to mainnet two without using any sort of fancy TAC servers. There was a little bit of finessing on the network layer, and if you're trying to get that working, just don't think you're not going to learn about subnetting in the process. But I can provide some of those details in my guide, and I was able to send ATAC messages from a device connected to mainnet one over Halow to a device connected to mainnet two. I was also able to do a couple other fancy things like drop images onto the map and share those between nodes. And I was able to do that using something called multicasting, which is essentially making use of the Wi-Fi network to send those communications, but there was no sort of centralized remote or even local TAC server. So it sends it directly from device to device by IP address on the network. And I gotta say, I really like that setup because everything is confined within your private network and you're able to push communications to each other, track nodes on a map, and have those sort of off-grid comms that you usually only normally get with a satellite connection.
But what about voice and push to talk? So I actually set up mumble on both of the devices, and I didn't go too crazy with it. I used a remote mumble server, but I was easily able to push voice from node to node using one of those public servers. Now you would never actually want to do that. You would want to set up a mumble server on one of your raspberry pies, but that should be easy enough. And the latency was pretty low and the quality was pretty high. But anyways, this is push to talk. It works just fine. It's using one of the uplinks, connecting to a remote server on the web. But the signal is still going from mainnet node number one to mainnet node number two, and they're only connected by Wi-Fi Halow. So you could go long range with it and you can see the clarity and the latency is more than adequate.
Next is video feeds. So I didn't go too crazy with it, but again, I was able to set up a video stream from a device that was also on the network, which is this Xiao here. This was actually also using Halow. You don't necessarily have to use Halow. You could use something like IP cameras, or you could set up a stream on a laptop or a drone or something like that. But as long as it's using a standard either UDP, RTSP, HTTP, one of those live video streams and a standard protocol, you can pull it up on your ATAC device, provided it's on the same network. Okay.
And then the big test was going to a remote off-grid area. So I went down to Crandon Park, which is basically a key off of Miami. Now I had both of my mainnets, but I also had something else, an uplink. So I just purchased a DISHI, which is a standard star link, and I plugged into node number one, which means that my whole network had internet, but just how far could I push it? So I took node number two out for a stroll and we just kept going periodically checking the connection. Okay. So now we're trying to get the Halow mainnet plugged into the star link router, and this will be node number one. So we have the Halow chip with the Halow antenna, which is going up here. And it's using the star link wifi router. So that should just bring Halow to this star link network here. All right. So we're trying to get star link online here. Looks like it might work. I had to use a life PO4 battery with a special, I don't know, sort of like regulator, governor of some type. And then the star link itself is right up here. Cause I'm not using a star link mini. I'm actually using the standard one because it gets better, better throughput. But it's a little bit harder to power. But I think this is going to work. It looks like the app is still thinking about it right now, determining alignment adjustment, maybe necessary. So it looks like it might work though. So we'll see. Okay. So it says we're online. So that actually wasn't too painful. Let's see real quick what we're getting in the way speed tests. Right. So let's go run one of these. All right. So not too bad. I'll take it. So 22 down, three up, but it looks like the software is still updating. So we'll see if perhaps that gets improved. Okay.
I'm going to give you a little rundown of everything we have going on here. So this is going to be a test of our main net. So mobile ad hoc network, node number one, mobile ad hoc network, node number two, primarily talking over wifi Halow, but I also wanted to bring an uplink to the party. So let me go ahead and show you what we have in terms of the configuration here. So down here, you're going to see so down here you see the Raspberry PI. This is our wifi Halow board. This is the main net essentially, right? It's ethernet into the star link router. I'm doing some crazy stuff with batteries because this is not a star link mini. This is a star link standard. So, but it seems to be working and it didn't give me too much trouble. So in terms of the wifi Halow antenna, that is right here. This is magnetized. So I can just pop it on like that. Our dishy is right there. And the uplink works has internet connection. It's not super fast right now, but I'll worry about that later. So now what I'm going to try to do is get the second main net going and see how far I can push it. Cause we have a lot of space here. So that's my objective right now. Okay.
So we actually have everything working now. This right here is node number two. I have the main net right there and I'm running ethernet into my computer and we can see that we have internet, even though my computer is not on wifi. So we have about 14, 15 down, which tracks perfectly because the internet's coming over wifi Halow at an eight megahertz width. And the theory theoretical max for that is like 17 megabits per second. And it's got the same down. So symmetric link. So everything's actually working now. Normally you would connect your main net to your end user devices, like like your ATAC devices. You would connect that over wifi, but that's just an extra step that I didn't want to get into right now because I'm still awaiting that piece. That's just a little wifi dongle. So you would just keep your radio on you and then you'd have a local wifi network and then Halow for the long distance stuff. But the question is now, how far can we still get a signal, right? And at what data lengths? And so I'm going to give that a shot. Okay.
So I have my little radio station here. I'm on the move here. I just did a test. This is node number two and I got six megabits per second down and 14 megabits per second up. And the second node is that white car way, way, way, way over there. But I was watching YouTube. I'm watching YouTube because of the, the, uh, sub gigahertz wifi Halow radio. So now we're going to go even further. Okay. So we're right here. This is our main net here. We're way, way further out, probably come up on a mile maybe. And I'm still getting internet here. Although it was beginning to break down a little bit. Let's see if we can do a speed test. Whoa, here we go. All right. So now we're not even quite hitting a megabit per second, but close, close to a megabit down. See what we get for up. For some reason, the uplink is better. And a couple megabits per second is still, it's still quite a bit to work with in terms of a voice low rate video, stuff like that. So I'd actually be curious if we can get any significant video here. So I'm going to put this in one of my hands, which I really don't have a free hand here. And then we're going to go back to YouTube. Let's see. See if anything comes through. This microchip can extend your Wi-Fi over 10 months. So we're watching YouTube videos using the 915 antenna. And my star link is way out there. So we're in a remote area. We're near a beach. We have no fiber or anything like that. We just have a star link. I'm about a mile away and we're still watching YouTube videos. Call me impressed. And now we've got proof that this little box can keep up with the big boys. So at 2000 feet, I was getting shy of one megabit per second down and almost three megabits per second up, which interestingly was enough for me to continue watching YouTube, which just felt surreal. I was 2000 feet away from my star link and still able to watch YouTube all due to Wi-Fi. Halow.
Now, to be clear, this is the Civvie version. It's absolutely usable in the field, but it's not military spec. It's not guaranteed to survive a Humvee rollover or keep sinking after a week in the desert. Think of it as the civilian alternative, open source, flexible and cheap enough to actually build yourself. Also, I'm not here to push more screens into more places, but I've been out where phones die and help's not coming. And that's where tools like this might have a place. And if the tech encourages people to get outside and explore, then I support that too. Now I know what you're all thinking. What about security? But rest assured, Halow radios use WPA3-SAE, the same encryption handshake you'll find on modern enterprise Wi-Fi. It's strong, it's proven, and yes, everything running on this mesh is locked down. So this isn't some open wild west signal that anyone can hijack. It's encrypted end to end, just like you'd expect.
Now you might be thinking this looks like LoRa or Meshtastic, but here's the difference. LoRa is brilliant for ultra low bandwidth over long range with very low power draw. It's perfect for text and GPS pings, but it tops out at kilobits per second. It's like walkie talkie, but for data. Wi-Fi Halow, on the other hand, runs in the megabits per second range. That's the difference between just sending a ping and streaming a video feed. And the big kicker is Halow supports full IP networking. This isn't a custom packet form, it's real internet. You can run ATAC, push-to-talk apps, webcams, even route traffic straight through Starlink. It's literally plug and play networking, just stretched for range. So if Meshtastic is your digital walkie talkie, Halow is your portable router for the middle of nowhere. Look at me, I am the ISP now. For more killer tech, click here.