Community wireless network
Posted by Lost_Engineering_phd@reddit | preppers | View on Reddit | 47 comments
I was over in the PirateRadio sub and a discussion came up that got me to thinking about going way beyond just my part 15 AM radio station. The discussion got me to thinking about the possibility of building a large community wireless mesh network (not meshtastic) but rather a fully built out intranet wifi system. For hardware my plan is to use OpenWrt routers with BATMAN mesh running backbone on 5.8, and I can build traveling wave slot antennas for long range 2.4 clients. I would want to power these nodes using Solar so they are entirely infrastructure free.
Regarding software, it would start with a captive portal, from there users could chose apps such as jellyfin video servers with around 1000 movies, epubs, wiki, boards, chat, sip VoIP, etc. This would go well beyond the prepper Disc raspberry Pi system. The goal is to have many miles of coverage across many nodes and provide service for thousands. I already have much of the backend built in my home lab and it is idle most the time. My thoughts are to build out a network that would be local and free of corporate or government controls. I would not plan on it having connection to the public Internet.
I am highly confident in my technical abilities to accomplish this, it would certainly be challenging im sure. My question is should I. Would it be any benefit to the community, this service would continue to function even if there was no Internet or grid power. It certainly seems like something that might fall into being a prep thing.
If you do think this is a good idea or you think I should add any service comments are welcome. Alternatively, if you see any problems or just think it is a waste of time let me know. I don't want to buy 20-50 routers and waste time building out a worthless network.
No-Language6720@reddit
Man I would love to collaborate to help with this. If we could build a second Internet totally funded by citizens for citizens outside of government or corporate control hat would be fucking awesome. if we could get even a small a portion of people at least in the US the ability to tap into it...idk if we could pull the funds to do anything meaningful though. If I we could pull it off we may be able to actually start a revolution without the fucking algorithms messing with things and the mass censorship happening.
I know there are some social media type of sites out there with similar concrepts to what I described above already but these are still tied to the traditional internet. For those from what I understand they are just self hosted in someone's house/garage/basement for decentralized hosting. Both this idea and yous would likely come with a heavy amount of security risks for hacking and the government would probably be looking to spy and get in when they find out about it.
I'm not a heavy network person but I do know enough about DNS over the traditional Internet and what CIDR blocks are and software coding around some of that. I can also build out APIs that can function across the traditional internet or even corporate private intranets that don't touch the public internet. I'm assuming what you would build would use similar protocols to the traditional internet or similar enough to learn. We could build out new ones that make more sense too. Just my rambling of thoughts. I have enough experience I should be able to help with some level of coding if needed to contribute.
I would also gladly help with funding if needed/wanted.
Feel free to reply or DM if you want to talk further.
Lost_Engineering_phd@reddit (OP)
It would be feasible to have disparate networks that could use SDWAN to link together. In a lot of ways we would be building a modern fidonet.. back before the internet many BBS's (mine included) were part of fidonet, and would allow sending mail from BBS to BBS.
There are a couple existing community wireless networks around, but most have focused on providing Internet. Guifi.net in Spain and Freifunk in Germany have built out extensive mesh networks. The problem with mesh networks is that they tend to collapse when trying to handle video. My idea for a network started as a video service first. I have over 1000 DVD's I have been converting to put on my jellyfin server. Then it occurred to me, I should add VoIP, next was the idea for a social and bulletin board. There's quite a few ready to go docker and VM packages already so not having to reinvent the wheel. My main innovation is to improve the RF performance. I think I can use aluminum downspout to make a traveling wave slot antenna. This should provide about 18dBi and a 120° pattern. By using downspout the cost can be drastically reduced. My idea is that a standard node would have 3 2.4 GHz radios and a 5.8 GHz interlink backbone. I would plan the 5.8 to use helical circular polarization antennas. I also plan for super nodes that would have a mirror of the jellyfin and geo-Dns users to the closest server to reduce network load for video service. Since this is not the Internet, we would have a captive portal start page so users can access services.
I think if a standard server package and network design were built that could be duplicated, we could have a true alternative.
No-Language6720@reddit
Man...you are awesome as hell. Please somehow keep me in the loop if you want some collaboration and/or help. I'm familiar with docker containers and could probably be of use if you decide to expand.
Lost_Engineering_phd@reddit (OP)
I fully welcome anyone that has anything to add to this. I am still investigating if the antenna design is feasible. Keeping the cost as low as possible is the key to whether this works or not. Making the antenna ourselves rather than purchasing is the only way possible. A commercial antenna like an L-Com 120′ 17dB antenna cost close to $800, And you would need 3 of them at every node. The 5.8 backbone could be done a bunch of ways since that would be a point 2 point or multipoint link. If cost were not an issue, 24GHz would be great for gigabit+ back haul, but those radios are not cheap.
As I have been giving this more thought I have actually come to the thought that I may need more than just a standard node and super-node, so I'm also looking into a mini-node that would be only mesh linked with a single radio.
I also need to look into the total system power requirements. I think each node should be entirely off grid solar powered so it is not reliant on any external systems. The idea is extreme fault tolerance at every level. I'm afraid my HP Z8's HA configuration will require quite a bit of solar to make that grid free though.
I may need to start a project page if there's much interest. There's a whole lot of things to design and work out before the first node goes live. I'm still on the fence if it should even provide Internet connectivity. If there were lots of these networks across the country some kind of interlinking would be amazing.
Austechprep@reddit
You should check out Reticulum, it can operate over any network interface, so it can do LoRa like meshtastic, but it can also do WiFi, ethernet, serial etc.
This is similar to a project I'm working on, but I plan on working on low bandwidth due to the difficulty of connecting my non-tech friends houses with WiFi etc, far easier to get non-technical people using an antenna that doesn't need perfect alignment after a storm that requires us to use the off-grid network.
I've been getting a few friends up and running, and I like Reticulum cos it's cheap enough for me to get them up and going, easy enough for them to use (although there is certainly some complexity in getting the network setup).
https://github.com/markqvist/Reticulum
You can make low-bandwidth web pages for people connected via lora or other radio, I've got a low-bandwidth web page as the "front page" with a few pages that give an explanation of how to use the low-bandwidth internet and then some areas for if you are on a high bandwidth.
Lots of options and if your technical you can develop your own software using this stack to do what you want.
Hardest part about doing what you want is finding people willing to be a part of it, unless you're able to get access to public infrastructure you will have a hard time finding someone who is willing to stick an antenna on their roof so that they can text another very small number of people.
But goodluck, I've managed to get 4 other people on board, decent reception across most of my town which is very fun, but not everyone has backup power for their nodes so it's not a perfect network.
Lost_Engineering_phd@reddit (OP)
I had not looked into Reticulum, the android app looks pretty neat. I will have to educate myself on it. I'm hoping that offering access to my movies would be the key to getting rooftop nodes. Alot of ideas to work out for sure.
Austechprep@reddit
I think like others have suggested, getting people keen on an off-grid version of it would be the difficult part. I've been able to do the comms and little website/forums stuff using Reticulum because it's my hobby, I'll pay for it and they don't really need to do anything. I already share my movies too just using Plex, it allows my friends to stream the content I have on my home server already and a reasonably good interface (although it keep suggesting stuff not on my server, but external sources which is confusing to some of the older people on my Plex).
I always figured that if I was ever close enough for a WiFi meshed network to work in whatever community I was in, that I'd try to join all the main buildings etc by ethernet anyway and leave the wireless to where it's absolutely required.
Directional WiFi isn't terribly expensive but it would add up very quick if you've got a lot of people getting involved, let alone the need to point it at the next dish in the line, so everyone would need a sending, receiving or repeating dish, a lot of organization required.
I think if people are keen to join a off-grid network, Reticulum should tick every box, comms, websites, forums etc, it covers all bases, movies can be transferred by USB anyway.
Lost_Engineering_phd@reddit (OP)
Time is the commodity that always seems to be in short supply for sure.
Some years back I built out an extensive rural wireless broadband company, I merged and sold the company off. I know how fast equipment cost can add up, for sure. In my experience CPE and install was the hardest part to fund.
I will certainly need to find cost cutting measures everywhere I can. Much of this system plan hinges on my theory of using cheap downspout as a traveling waveguide slot antenna. The 2x3 is too close to the critical frequency at 2.4Ghz, but the 3X5 should work. I intend on doing a test build soon and verifying with the VNA that it works as expected. For the 5.8Ghz directional antennas I plan on using a very special helix antenna. I was fortunate enough to have met one of the designers of the Russian Lunokhod lunar probe and he shared some antenna secrets. The helix antennas I have built based on this have been amazing. There's quite a few cheap AP's I can get that will run OpenWrt and BATMAN mesh.
It is certainly an ambitious project, with an uncertain outcome. My thoughts are that alot of people will soon need to be making tough decisions about what they spend on and having a free community service with movies could be a big draw.
Austechprep@reddit
Thats awesome congrats, I'm in a similar boat except I am hoping to sell my company in the next 1-2 years. Once sold and I've got some disposable funds, I have every intention of upgrading the mesh network to whatever the hell I want haha.
5G cellular is apparently cheap enough to run, theres a few DIY cellular solutions out there and with eSIMs being common you don't even need to extra hardware of SIMs. I've not done much work with VNA but I've worked alongside a colleague using one for tuning our antennas, but I ended up finding a really reliable supplier for my antennas anyway, far easier.
Licenced bands would be fun to play with and be far more user friendly if you can get onto a 4G/5G band so you can have your own private cellular network, it would then do everything you're looking to do.
Next warning would be to be careful about how amitious you are with your movies incentive, offering a movies service would quickly draw the attention of licencing problems, there are plenty of legal ways to be shutdown for sharing entertainment, just keep it simple and create a comms high speed meshed comms network.
Sounds like you want to create incentive, free WiFi/internet is always a good enough incentive for most people, if your off-grid network can still connect to the internet, but just retains your custom services when the internet goes down, then that's probably going to get enough people on your network as it is, keep the entertainment freebies out of it, free data is plenty.
Eazy12345678@reddit
more realistic is get starlink and hope its up in an emergency
Lost_Engineering_phd@reddit (OP)
The primary idea is to not have corporate and government control and censorship. But SDWAN interlink via starlink between cities might be a good option.
93brunocardoso@reddit
Communication will be gold in shtf situation
Lost_Engineering_phd@reddit (OP)
This idea puts the community in communication,lol. As preppers we usually focus on Ham / GMRS / CB etc. I do have a lot of fun with each of those. Especially digital modes on HF! I think adding public coms, like my AM station and a CWN (community wireless network) could be beneficial too. In a SHTF situation I could go up the coverage of my AM well beyond part 15 limits.
shikkonin@reddit
There are already a large number of networks doing exactly that. Why build another one instead of extending the ones already in place?
Lost_Engineering_phd@reddit (OP)
Can you provide examples? (Not meshtastic) I know there's nothing like this in my city. I am familiar with Guifi.net in Spain, Freifunk in Germany as well as NYC mesh. Unlike NYC mesh, I am not looking to provide community Internet.
shikkonin@reddit
There you go.
Lost_Engineering_phd@reddit (OP)
If you are just going to be a contrarian, at least point out issues that pertain to the discussion.
You are aware that many people live outside these areas right? To have a 2000km link would require a height of 250,000m above the ground. This is orbital, so you need a massive constellation of satellites. The cost would be in the hundreds of billions of dollars.
This is completely against the idea of a local service network that is isolated from the Internet to provide service in the event of corporate or government censorship or grid down.
The group is about "prepping" for abnormal situations and or disasters. I'm not sure there's a "every thing is awesome" group that encourages normalcy bias. But there's a great song for that in the LEGO movie at least.
shikkonin@reddit
Doesn't make the technology bad and doesn't constitute a reason to do things differently from what has already been proven in real life.
Nobody is talking about that, stop inventing things nobody said.
You can be connected to the internet in normal times and still provide isolated service when the grid goes down.
No shit Sherlock. You know what works best in abnormal situations or disasters? Stuff that you have fucking experience with. Hence providing utility so that people use it in normal times as well is an immense bonus for preparedness.
preppers-ModTeam@reddit
Your comment has been removed for breaking our rules on civility and trolling.
Lost_Engineering_phd@reddit (OP)
As for having experience as you put it, I have built and sold off a rural wireless ISP some years ago. Not my first rodeo with large wireless networks. My longest P2P backbone link was over 20 miles. I have worked with 900Mhz nlos, 2.4G, 5.8, licensed 8, 10, 11, 18, 23 GHz 24 and 60 GHz P2P and even free space optical. I have installed and designed hundreds of microwave links. I won't even go into my experience with HF or my contract work in highly adverse conditions and locations, So yeah experience.
I clearly mentioned the BATMAN protocol (Better Approach to Mobile Ad-hoc Networking) this was first used by guifi and freifunk. I'm not reinventing the wheel here. The primary improvements I'm looking to deploy are adding geo-DNS to distribute the network load for video servers and improved RF design. The per node cost is also quite high for both of those networks, fortunately for them, they both are in Socializt countries and receive significant government funding. They are also deployed in areas with much higher population density. So for use where I am will require a very different network design. Networks are bespoke to the environment they exist in.
All of the projects I am looking at hosting on my servers are mature and deployed across thousands of platforms. I have experience with many of these packages some at scale. I am not opposed to having some of the services Internet facing to allow external access. Especially the forums and social servers. But there is no need for another ISP.
You may not be aware but there's a whole shit ton of stuff you have to do if you build an ISP now, even a free public one. If you are providing VoIP that connects to the world (PHONE), you must be e911 compliant this requires having each user's info and location. You have to log access under the DMCA. You now also must comply with the protective acts that have passed. It is crazy now, just running a Tor node can get you in trouble. However if your network only provides intranet services within itself you can avoid all this.
shikkonin@reddit
What a wall of text without any relevance whatsoever. Nic ergst you have experience, are you planning on being the only user of this network?
There is a reason why it's not used anymore...
Oh come on. You can't really be that dense.
preppers-ModTeam@reddit
Your comment has been removed for breaking our rules on civility and trolling. Insulting other users is not acceptable on r/preppers.
Lost_Engineering_phd@reddit (OP)
I have built and deployed networks with thousands of endpoints.
I will give you a point in that the network and microwave links I have built were built as P2P and P2MP fully managed. A mesh network is a very different beast.
That is quite a claim that BATMAN is not used anymore. If you said OLSR, sure but libremesh and others are using BATMAN.adv Seriously if you know a better scalable mesh protocol I would be very interested in learning about it. There's certainly room for improvement in route discovery.
I suppose I should elaborate and be more accurate by calling Germany and Spain democratic socializist. I am not saying anything about the method of government the people there have chosen. I very much enjoyed Germany while I was there. I can not say how it is there today, but when I was there the German people I knew seemed a lot more comfortable and happy than their American counterparts. Russia has also gone with the plutocratic oligarchy method of government we seem to be courting.
If you do have knowledge in the field of mesh network design I truly welcome your insight and experience. I know from my experience it can be very difficult to have optimized lowest cost routing in a fixed structure network, a dynamic mesh is an exponentially more difficult challenge.
Potential-Load9313@reddit
don't get pissy just because someone challenges your idea
Lost_Engineering_phd@reddit (OP)
In the Socratic tradition a challenge must pertain to the discussion and be factual. If a person is levying a technical critique they should have an understanding of the material.
If critical points are more along the lines of social or general principals that is also welcome for discussion. And for large scope subjects like that details and evidence can be assumed. IE "if you build it will the come?"
As for accepting criticism, If you look at the thread you will see a critical discussion where the throughput limitations could be problematic for video service. The proposed solution is to have super nodes that mirror the primary video server and use Geo-DNS direction. This may still be inadequate on its own. Configuration to have a higher speed hub and spoke P2P backbone may alleviate this. But only build out testing will prove it one way or the other.
halcyon4ever@reddit
I've been kicking this idea around for nearly a decade, and have built a similar system. The issue with wifi mesh is it is slow and the range is crap (unless you are using really high priced dedicated mesh AP's, but I'm building this with recycled junk). We did the same thing, using the 5gz radio as the dedicated mesh backhaul and 2.4 for clients, which got better coverage. For things like Wikipedia (locally hosted on kwix) it worked great, but video quickly choked the bandwidth.
Not disparaging it, just sharing our current bottleneck.
We needed a node at nearly every house (5ghz doesn't have the greatest range). And having the nodes outside meant they could get line of sight. So we built the prototypes into garden gnomes that could sit on front porches.
David_C5@reddit
There are long range WiFi. It'll cost less overall because you need less of them. They'll transmit for miles.
Lost_Engineering_phd@reddit (OP)
Thanks for sharing your experience.
My plan on optimizing the network for video is by using geo DNS and mirroring the jellyfin servers at super nodes, then rsync. My video library is also mostly SD DVD rips. I have a pretty strong background in RF and microwave, I had planned on using downspout pipe to build high gain traveling wave slot antennas on the cheap. I have to use the larger 3”X4” for the waveguide to be above the critical frequency for 2.4Ghz. 5.8 might be best using sector panel antennas for point 2 point.
It would be interesting to know how many hops are stable between the user and the video server.
halcyon4ever@reddit
But I'm going to pick up some openWRT routers and try your method. We've currently been looking at a vpn mesh using wireguard. Since short of a grid down I assume internet will still be common place but highly controled, so utilizing vpn's to connect the nodes and isolate the traffic was our current direction.
Lost_Engineering_phd@reddit (OP)
I had not thought about using VPN mesh. I was limiting my ideas to a full gapnet with no Internet connections, I'm mulling over having a proxy or gateway. But not looking to be an ISP, been there done that, sold the T-shirt and the hat.
In OpenWrt, you can dedicate the wl.1 5.8 radio to a port and use very high gain directional antennas as backbone.
I think it would work to build a mesh link node using a bit of downspout for the 2.4Ghz slot antenna, then have the radio at the top and have 5.8 helical directional antennas for the backbone. This should keep clients off the backbone and provide a far higher speed connection.
I will work on cutting a bit of spout, and testing with the VNA to see how It performs on 2.4. I'll let you know. The key is to keep it cheap.
halcyon4ever@reddit
We started looking at the VPN approach because censorship is becoming more of an issue than full loss of internet. Also because our primary users were more geographically distant. But I think our goals are simmilar, creating an easy to build node that can extend the network in both infrastructure and content.
halcyon4ever@reddit
Main thing we hit was capacity. 1 worked fantastic, but some where around 5 streams it crapped out. Which maths out to to match the bit rate of a video at 10mbps x5, hitting was the max of a wireless G at 54mbps.
The problem with wifi, is it essentially operates like a hub not a switch. Every packet from every user within range has to be processed and discarded, so the pipe saturates even if the data isn't for you. So this is more an inherent limitation of the technology and not of just meshing. I run into this with businesses that have gone all wifi for their office and a few zoom calls will start to max out the wifi.
Lost_Engineering_phd@reddit (OP)
I have also seen iphone clients with poor signal going down to 2mbps and severely hurting network performance on wifi many times.
Making sure the back haul links could sustain through put will certainly be a challenge.
ttkciar@reddit
It does seem like a good idea, but be prepared for nobody else in your neighborhood showing interest.
You might want to add an OpenStreetMaps service to the list.
If you don't already have something lined up for the wiki/forums/realtime-chat, check out Fossil-SCM. It has all of those (and more), is trivial to configure and manipulate (via http or its cli tool), and is a simple C program which compiles to a single binary. Its backing store is a single sqlite file.
I love it because I can make changes to Fossil-SCM instances on my laptop, and rsync the sqlite file to my colo server to "deploy" my changes.
Lost_Engineering_phd@reddit (OP)
Streetmaps is a great service to add, I had not thought about that one.
I will have a look at fossil. Might be nice to not have a whole stack of services that are not integrated.
Paranormal_Lemon@reddit
Especially if it involves flashing firmware to routers
halcyon4ever@reddit
Yeah, my neighbors had zero interest, but were at least willing to let me plop a garden gnome shaped router on their front porch for testing.
MM_K16@reddit
may not be in your neighborhood , but this is definitely something that would benefit certain close knit communities
JawnZ@reddit
Check out AREDN too It's for Ham radio operators but pretty cool with the distances you can get. You could find s;me ideas on stealth nodes (get them up high on the side of a hill) and really expand your reach
Lost_Engineering_phd@reddit (OP)
I have messed with AREDN, the newer OpenWrt with batman mesh is light years ahead. The cool thing with AREDN was the extra 3Ghz gear, but that is gone now.
When I built a WISP years back I had a 20Mile P2P link to get to a rural town, that was wild.
Swmp1024@reddit
What you are doing is pretty similar to AREDN networks on ham radio. I have a few mikrotik 2.4 routers I've hacked to do this. Great idea and could be useful. BATMAN is a more modern routing protocol that is better but a lot of infrastructure already exists with AREDN and you might even have it in your area.
Lost_Engineering_phd@reddit (OP)
I have not seen any ARDEN nodes in my area. Most of the hams in the club are of the opinion that a Ft-101zd is a modern radio since it only has tube finals. (Not hating on the 101ZD, I have one)
besquared2@reddit
The problem with things like this is that they are a good idea for peppers because we can see the need for something like this in a grid down scenario. To everyone else, it's not a good idea until they need it and there will not (likely) be an interest in it until it is needed, except by you of course.
Lost_Engineering_phd@reddit (OP)
That has been the main thing I have seen as a hold up too. I'm wondering with all the corporate censoring if that could become a driving force. Obviously if the grid is down and my network is the only one showing up on people's phones they will find out about it.
MM_K16@reddit
this is something I would like to discuss. what kind of system do you have running now?
Lost_Engineering_phd@reddit (OP)
I have a couple HP Z workstations with dual xenon's set up as VM and docker host running in HA redundancy. I am also using HP switches (3500yl and 1920s) currently I have two mikrotik CCR1009 routers in HA redundancy.
It is very similar to my network at work, so I can test changes. Better to crash my home lab than to have +1K people pissed off the network is down. I am considering getting Tp-link EAP225 outdoor ap's since they support OpenWRT and can run , they are also going real cheap now.
For my AM old time radio station I use a talking house Part 15 transmitter with some improvements for the audio, and outdoor antenna.
I have been loading my jellyfin server with my movie collection, I have about 1000 DVD's in my collection. I am looking into deployment of PeerTube and mastodon, mirotalk and phpBB servers to add social as well. I also plan to have internal email and office apps too.
My idea is a full feature non internet network that could empower the community. I may need a couple more solar panels to power the servers when under load. The plan is to have a reliable disconnected network that is even available if power is unavailable.
Secret_Enthusiasm_21@reddit
why? What is the benefit it offers, compared to the already existing networks? If the answer is "it's denser and better and faster", then yeah... it's not like all the other ones are opposed to having a denser, better, faster mesh. The limiting factor is the amount of contribution from community members