Emergency backup wifi network

There might come a day when the electricity or internet goes down in your area and you would like to still have some local communications.

Or maybe you just have a temporary camp site that you would like to set up a large Wifi network over.

Either way, below is an explanation of how to use cheap routers to create a basic emergency network spread over several hundred metres.

Here we contemplate a scenario where a major event has occurred and a network needs to be deployed quickly and without a high level of training.

The benefits of a local network in an emergency include:

  • being able send and receive voice calls or PTT message with nearby citizens using existing smartphones
  • being able to send and receive text messages and photos with nearby citizens (i.e call for help)
  • If Internet is available on the network, restoring communications for those able to reach the network
  • being able to set up IP Cameras to offer ad-hoc surveillance in the local area
  • Being able to make important files available to the local community (books, maps, entertainment, news posts)

Therefore it would optimal to have ready-to-deploy infrastructure stored and available for rapid deployment.

Design Goals

  1. Ideally the network should be able to use multiple nodes that will form a mesh with each other without having to explicitly configure each node in the network.
  2. Each unit should be able to source power from solar or easily replenished batteries
  3. The cost of each unit should be kept to a minimum


This implementation will:

  • Be based on Linksys WRT54G and WRT54GL routers (54Mbps)
  • Configured in WDS mode to present a single SSID network to connect to and to support roaming
  • Use off the shelf and available hardware
  • Use Solar power as the primary energy source
  • Store power using lead sealed acid batteries or 1860 Lithium batteries
  • Attempt to be durable against dust and water ingress and extreme heat

Ideally A mesh of ruggedized, battery backed up WRT54GL routers forming a bridged local network on a common SSID for local communications and video feeds using software such as Serval or similar.


Linksys WRT54GL routers are a (54Mbps) router that has been available since 2005 and remains in production. It known as extremely reliable and still sells millions of dollars’ worth of units a year indicating likelihood of ongoing manufacture. If manufacture were to cease, the second-hand market would still provide a viable source of units.

Current new price is AUD$25 (Umart).

Power draw is estimated at 5W max (tbc). (Max and minimum voltages to be analysed)

The WRT54GL includes two real omni-directional, replaceable antennas which are able to be positioned. This opens up the option to replace with higher gain or more directional antennas for longer hops in the network.


  • Current draw implementation would limit to ~50 units due to IPv4 Address range. Backplane bandwidth saturation would like limit to 10-20 units anyway.
  • Network Speed – The 54Mbps of the WRT54GL is slow by modern standards. Wirelessly bridging halves this bandwidth
  • Ideally include the ability to throttle heavy users – real issue if Internet is added to the mesh network.
  • Current draw – whilst low, unlikely to have ability to reduce further

Implementation / Building


Model A – Single box with WRT54GL, battery solution, and solar panel.

  • Monocrystaline Solar Panel would be smaller than Poly
  • Need antennas to be external – test if plastic causes wifi signal blockage
  • Housing to be IP68 or similar ($75 pelican clone)
  • Renogy 50W solar panel – external
  • Epoxy the slot for the power cable through the case
  • Weather-proof power connectors


Two brand new WRT54GLs were purchased to help expand the testing
This dust and water proof portable container was acquired for testing.
A quick mock up of the lay out of the router, power supply and antenna mast

External Overall Case Dimensions 430 x 380 x 154mm
Internal Overall Case Dimensions 395 x 320 x 150mm
Case Weight 2.7kg


  •  Match the solar array and battery storage solution to the current draw of the unit under normal operation (Panel > 40W appears to be needed).
  • Renology 50W mono panel – 63cm vs 54 cm, 4.4kg
  • Offer different battery solutions for different situations
  • SLA for cheaper solutions where weigh doesn’t matter
  • Switchmode 230V plug adaptor (e.g if generator or car based invertor is running)
  •  4x 18650 based 3D printable cradle. This design should allow additional cradles to be added in parallel for longer usage times. To work with the shape of the WRTGL – ideally to work with the inherent stack design.
  • Inexpensive PCBs for MPPT solar charges of 18650s exist
  • AA based battery cradle (10x AA or AAA).
  • Internal UPS (in the space available inside the case). Maybe using Lion batteries in the AA form factor.
  • USB to WRT54GL power solution (some kind of buck-boost 5V to 12V convertor)
  • Power consumption
  • Investigate options to save power in the WRT54GL.
  • Investigate options to setup an external controller (Arduino) to reduce power consumption.
  • Investigate on PCB changes to save power (i.e. replacing Power IC with more efficient version)
  • Case enclosure
  • Test the wrt54gl inside a structural polymer case (pelican)


There are many options for powering a WRT54GL.

Power Requirements

The following data was recorded from taking a normal WTR54GL running r14929 of DD-WRT at different voltages.

It would appear that 9V is about the lower stable voltage at WRT54GL will operate. The device used was an early device ~(2008) based on is low value MAC address.

VoltagemA minmA maxStatus

Therefore it could be run as low as 8.5V but 9 would be a safer bet.

AA Batteries

18650 Batteries

Sealed Lead Acid


  • Use benchtop power supplier to determine minimum voltage
  • Sacrifice a unit by testing maximum voltage
  • Pull apart and analyse power circuit to see if datasheet for ICs can be located to determine min/max voltage.
  • Website (+ github) for software configuration
  • Field test distance range
  • Field test antennas
  • Find cheap directional RP-TNC based antennas
  • Build test rig to monitor field condition solar charge capability

Stretch Goals

  • USB port to allow charging of phones from the same power source
  • USB stick and adaptor to allow Serval to be installed on Android phones
  • Include storage of handset inside weather-proof case whilst charging
  • Camouflage the box and panel
  • Roof friendly mount points
  • Option to use default antenna or to attach external antennas     

o   Store external antennas in the same case

Similar Projects

Currently there are excellent efforts underway to develop a robust mesh extender that would be deployable in an emergency. Unfortunately these are not ready yet. See: http://developer.servalproject.org/dokuwiki/doku.php?id=content:meshextender:2ng

Build Blog

7/7/19 – WDS setup Success

Success at configuring WDS on the routers after 2 days. It is worth noting that the WRT54GL routers will not allow WDS to work with modern versions of DDWRT firmware.

The version of DDWRT known to allow WDS on WRT54GL v1.1 routers is the 2010 version 14929.

It can be found here: ftp://ftp.dd-wrt.com/betas/2010/08-12-10-r14929/broadcom/

Note that this quite old version has known security issues and so should not be used for a regular “daily use” set up.

9/7/2019 – Battery Voltage Tests

The goal is to test a power solution based on 3 18650 Lipo or simlar batteries. These will be 3.5 -4.2V each. Therefore the minimum voltage should exceed the minimum required to operate the WRT54GL.

The power solution is this Battery Charger (MPPT) combined with a balance charger and 18650 protection:

An interesting review of this board is found here on YouTube:

Also looking to test this board from DF Robot:



Four out of five WDS nodes will work happily. 5/5 causes connectivity over the whole mesh.

Need to add “keep alive” setting to each WDS client router to make sure they power cycle if they lose connection to the host.

Solar Power Generation Test

Using two 10W panels, and the DF Robot Solar Management Board, 4.120 Ah (73.7Wh) was recorded as being collected and stored in the SLA battery.

The maximum current generated was .88A. The two 10W panels are wired in parallel with a by-pass diode (1N4007) and the recorded maximum power was 15.5W.

As the battery was fully charged, it is expected that approximately 15-20% more power could have been collected during this single day trail.

The design will now focus on a 9Ah SLA battery due to the testing.

Leave a Reply