Mesh networking when the cell network fails

Interesting article about a new plan for mesh networking Android phones if the cell network fails. I point this out because of another blog post of mine from 2005 on a related proposal from Klein Gilhousen that he was pushing after Katrina.

The wifi mesh has the problem that wifi range is not going to get much better then 30-40m, and so you need a very serious density of phones to get a real mesh going, especially to route IP as this plan wishes to. Klein's plan was to have the phones mesh over the wireless bands that were going unusued when the cell networks were dead (or absent in the wilderness.) The problem with his plan was that phone tranceivers tend to not be able to transmit and receive on the same bands, they need a cell tower. He proposed new generations of phones be modified to allow that.

But it hasn't happened, in spite of being an obviously valuable thing in disasters. Sure there are some interference issues at the edges of legitimate cell nets, but they could be worked out. Cell phones are almost exclusively sold via carriers in the many countries, including the USA. They haven't felt it a priority to push for phones that can work without carriers.

I suspect trying to route voice or full IP is also a mistake, especially for a Katrina like situation. There the older network technologies of the world, designed for very intermittent connectivity, make some sense. A network designed to send short text messages, a "short message service" if you will, using mesh principles combined with store and forward could make sure texts got to and from a lot of places. You might throw in small photos so trapped people could do things like send photos of wounds to doctors.

Today's phones have huge amounts of memory. Phones with gigabytes of flash could store tens to hundreds of millions of passing (compressed and encrypted) texts until work got out that a text had been delivered. Texts could hop during brief connections, and airplanes, blimps and drones could fly overhead doing brief data syncs with people on the ground. (You would not send every text to every phone, but every phone would know how many hops it has been recently from the outside, and you could send always upstream.) A combination of cell protocols when far and wifi when close (or to those airplanes) could get decent volumes of data moving.

Phones would know if they were on their own batteries, or plugged into a car or other power source, and the ones with power would advertise they can route long term. It would not be perfect but it would be much better than what we have now.

But the real lament is that, as fast as the pace of change is in some fields of mobile, here we are 7.5 years after Katrina, having seen several other disasters that wiped out cell nets, and nothing much has changed.


Such a system could also be useful for the unnatural disaster of your local authorities deciding to censor cell/broadband networks.

To the extent an app can pick a wifi SSID for you -- probably a safe bet on Android, maybe not on iOS? -- it seems an app plus set of standard conventions could make this work.

Namely, decide a consensus SSID string for emergency situations, a sort of "Free Public WiFi" that's not a glitch. Perhaps even codify that it's not to be used for non-emergency situations, like CB Channel 9. (Maybe it's even, "911".)

Have an app that when launched, switches your device to that SSID, finds any neighbors, and both receives and forwards short text messages, including public bulletins, requests for help, and local routing info (including hop-counts-to-the-mothernet).

There doesn't *need* to be anything else, but with proper hardware support this app might also put things in an extreme low-power (or intermittently high-power) mode, for longevity or distance. Mass-market access-point hardware could in the future always look for and provide throttled relaying for this SSID. (Perhaps, unthrottling when a reliable indicator of crisis arrives, like losing its own uplink.)

This is the important point -- that the plan be for intermittent, store and forward hop connectivity, the way USENET and E-mail used to work in the UUCP days, rather than live connectivity. I mean if it sees it can do live connectivity, great, but you should plan to do as much as possible without it. An efficient text message store and forward plan seems best.

As noted, you can try to hop to neighbours. It's best if the phones do this on their own. People in a crisis will not have pre-installed an emergency mode app or remember to turn it on. In fact you might just have the phones start doing occasional pings for emergency mode when they detect the main cells signals are gone -- but you don't want them doing this just because they are inside buildings.

You could have signed "an emergency is in progress" packets that propagate out to some phones and somehow tell other phones they should go in this mode. Or even a simple broadcast unit that can be taken to an emergency and flown around to set this status.

After that I think drones that can do flights over all the areas low and slow to exchange messages with any phones they see on the ground is the right thing to do. You can get to everybody, tell them that they can send texts to loved ones and the emergency crews, and to wait for drones and get their phone outdoors if they hear a drone, unless the mesh is already working for them. Tell them to get their phone on the car charger etc.

Right, you'd want it to be built-in and automatic for true emergency utility. I'm just thinking how we could do baby-steps to test the protocol and increase awareness.

An app would let some hobbysts try an ad-hoc store-and-forward net, perhaps in remote environments or as part of drills or other group activities: campouts, festivals, football halftimes.

Then maybe some random confluence of events or minor disaster would let the app version demonstrate its value in a way that generates news stories. Then, it could become a bundled install and promoted through official channels, and get the necessary OS hooks for launchless operation in a pinch.

Fly-over repeater drones are good, but maybe also just airdrops of cheap repeaters, or drones that park someplace once they find a spot that patches a big gap.

I don't think "mesh network" is the right word for this if it's designed specifically for emergency purposes. A mesh network implies constant, active, data agnostic connections between the devices. An active network that could be used for voice communication without the use of cell towers also makes the telcos nervous (that is, if it lets you make calls based on phone number routing).

Using Wi-Fi has additional problems. What kind of disaster would knock out the cell towers yet leave enough power infrastructure running that Wi-Fi hotspots in homes and businesses would still be able to be powered? Sure, some people might have their Wi-Fi router on a UPS or a generator, but A) Those are lower priority devices to have drawing that power, B) only a small percentage of people in an average population even has those backup power systems, and C) the fact that they use wall wart transformers makes them inefficient for the power that they're drawing. Wi-Fi is also far from being the most energy efficient wireless protocol for ad hoc networks between phones.

(As an aside, I have a UPS and I have my DSL modem and Wi-Fi router plugged into it. On the rare occasion of a power outage I'll power down everything else plugged into the UPS. The DSL usually stays up, since it's based on the telco's infrastructure which is designed to be not as dependent on the local power grid. Then I'll pull out my iPad and get two to three times the battery life of any laptop, with the DSL modem and Wi-Fi router getting comparable lifetimes out of the UPS.)

What I think would be more useful is an emergency protocol built in to Bluetooth 4.0 chipsets and firmware. It would be as low power as possible, designed to skip the standard pairing between devices, and all it would do is store and forward text (only) messages between them. Its range would be shorter than Wi-Fi, but most likely if you're in Wi-Fi range of another person's device (congregating on a street corner or something) then you'll be within range for Bluetooth 4.0, too. The lower range tradeoff is a significant decrease in power requirements, and no base stations would be necessary. Telcos would be less worried about it because it would be a purpose built emergency message system, and not a scary mesh network that can route voice calls between phone numbers and could grow into something outside of their control.

This purpose built emergency system would also avoid the security issues of rogue agents setting up Wi-Fi routers spoofing the emergency SSIDs and sniffing the packets that they see.

Once a drone flies overhead or a temporary cell tower is set up nearby all the stored messages on a device would be sent out to it using the standard wireless phone system control packets.

What's important is that this system would need to be standardized across handsets and manufacturers, and built-in to the OS so that hooks are built-in to the native text messaging apps to switch to this system in an emergency. Anything requiring the user to install and launch a special emergency app, and have or allow it to be constantly running in the background is destined to fail, because there just won't be enough people participating to make it work.

It would also be good to have the system set up to be able to forward the messages via IP over Wi-Fi, if the user does happen to come into range of a hotspot that has an active Internet connection. This adds complexity, though, requiring some kind of intermediate gateway server, when the point is to allow users to send messages using the built-in SMS apps and have them routed transparently between the ad hoc low power store-and-forward emergency network and the standard cell carrier infrastructure so that neither the sender nor the recipient needs to be running any special software.

Actually, wifi hotspots are not part of this, though they take little power and can run a long time on a battery or UPS.

It's phones that this system plans to use. Those usually have a day's worth of battery, and if you plug them into a car, they have a very long amount of battery -- effectively the full length of the disaster, particularly if there is gasoline.

The best thing to do would be to put the phones outside, but if you put them on the cars, people might steal them. You don't want to put them in the houses or in the cars, though you may have to. You might put them somewhere secure and outside, waiting for links to other phones or to drones.

Of course a portable cell tower radio in an airplane also works, does not have to be a drone. After Katrina and the Tsunamis you would think we would have something like this ready to go and be anywhere in the world in 12 hours.

Right now phones are not configured to do ad hoc Wi-Fi networking, which is why they'd need base stations for software-only solutions on existing devices. On the other hand they're all already capable of ad hoc Bluetooth connections between devices.

The main power draws on a mobile phone's battery are the screen and the wireless radios. The cellular radio is the most power hungry, the Wi-Fi radio is second, and the Bluetooth radio is last (especially with Bluetooth 4.0). A phone's battery life could probably be doubled/stretched to two+ days or more if the cellular and Wi-Fi radios are turned off, even with the Bluetooth radio still running (see: battery life of an iPod Touch vs. an iPhone or other smart phone).

Lots of people don't have cars (and/or phone chargers in the cars), UPS systems, generators, etc. As long as a system like this is being worked on, it should be done to universally maximize its usefulness and availability. Cell phones are becoming vital communications tools across the world and are much more likely to be possessed by someone before they own a car, a Wi-Fi router, a UPS, etc. This not only applies to people in developing nations, but to people's children and low-income citizens in the developed world.

Also, after Katrina how many of the people who even owned cars in the stricken areas had access to them, or had them in a good enough condition to be able to plug their phones into them, even if they had the cables and adaptors? What kinds of major environmental disasters are there besides earthquakes that would take down the cell networks but are likely to leave cars relatively untouched and accessible? (And even then walls can fall on them, they can be inaccessible due to collapse of parking structures, buildings above underground lots, etc.) Hurricanes, floods, tornadoes, major storms, fires, volcanoes... You can't count on getting power from cars after any of those.

But you don't have a lot of choices. But yes, the phone would seek the lowest power method of communicating. If it can reach people by BT, it should reach by BT, and use wifi if it can't. You just can't reach very far by BT.

I am not sure what you mean about "phones are not configured to do ad-hoc" because that was the point of the paper I link to at the top of the article, they had written new code to make them do that.

Add new comment