CEMN - Community Engagement Mesh Network

Under served communities can benefit from the exchange of information through local mesh networks to exchange critical notifications

Similar projects worth following
There have been many text message based initiatives in developing countries recently to achieve health, agriculture, education and economic development goals. Most initiatives use bulk text messaging to broadcast information such as crop reports and health advisories to the general population. In countries where the infrastructure is under developed this can be the most effective way of communicating. Often this is one way communications (broadcast) but where two way communications is provided the cost of the communications relative to local incomes is high and the cost and availability of staff to respond remain a challenge. More local support and a collaborative sharing of skills and resources locally is a more viable solution but the cost of using the existing telecommunication infrastructure remains an issue. Creating a local mesh network for peer to peer communications between individuals in small under served communities is a way of facilitating that communications.

CEMN: Community Engagement Mesh Network

What started as a paper design in the first round of the HackADay prize to connect people in developing countries using either Peer-To-Peer networks on handsets augmented with a wireless  mesh network has evolved to a refined design that tries to address some on the realities in the communities of under developed countries: limit budgets for infrastructure, lack of access to advanced manufacturing techniques, lack of free exchange of information and a lack of trusted sources of information.

The CEMN is design to be  secure (encrypted) and authenticated messaging system that can be deployed cost effectively in undeveloped and disadvantaged regions. It is made up of off the shelf modules so it can be assembled locally. As many sub-systems as possible tried to use potentially available local materials so the construction could be done locally.

Text messaging is the most cost effective means of getting information out in developing countries by government and non-government organizations. Because of the cost of sending text messages by individuals it is often a one way conversation. CEMN provides a way for people in developing countries to have a two way conversation by having the community own the network.

The design used off the shelf hardware component modules so that the design could be a plug and play as possible. It relied heavily on open source software to implement the features such a fault tolerance of the mesh network and message server software to handle messaging. Being  able to freely exchange, trusted information is something we take for granted in developed nations. In some countries it is against regulations to import any technology that encrypts communications, so the designer chose not to use some of the open source encryption schemes and software add-ons. In addition as the network could be potentially be closed with no connection to the great World Wide Web, due to their being no third party infrastructure in the region, the use of trusted third party sources to validate hosts are problematic and many of those open source encryption schemes depended heavily on third party sources to create secure connections. The designer decided on a more simplistic scheme to encrypt the messages and to provide validation. The encryption can be implemented after the fact without importing an explicit encryption technology into a country. The encryption schemes were chosen to not be  overly elaborate, so programmer without encryption and security experience could implement them, but effective enough as to be considered secure.

The challenge of this project was to create a system that could be deployed affordably, be maintained by people with no advance technical training, be fault tolerant as it may be months before trained staff can be in a remote region and be self powering in the absence of centralized power sources. Fault tolerance is a key important element in an environment where ownership of the infrastructure by the locals is not guaranteed. As a result regular servicing is not always done. There are countless examples where clean water infrastructure has been left to fall apart due to the lack of regular care.

This was not a project about creating the newest and the greatest technology. It was a project that tried to understand the cultural and economic realities of the end users and then create a solution that provided real world services that are needed and desired to its end users.  The CEMN App user experience mirrors that of sending traditional text message and messages are limited to 117 characters with longer message broken into 117 characters chunks. The restriction on character length is due to the asymmetric encryption scheme used.  Everything was done to balance simplicity with a reasonable level of security. Methods of facilitating longer methods are available but the designers wanted the CEMN Message App to feel familiar to end users. ...

Read more »

Final Technical Specification for Community Engagement Mesh Network Initiative - Final Oct 20 2017.doc

The is the technical specification for the project with user and performance assumptions, detailed descriptions of the hardware and software components, parts list, instructions and documentation to build the solution.

msword - 23.47 MB - 10/21/2017 at 08:27


Bluetooth Range Limitations Research.doc

This document discuss some the factors that determine the effective range of Bluetooth links.

msword - 44.50 kB - 06/12/2017 at 07:14


Analysis of Supplementary Wireless Communication Modules.doc

Comparison of potential supplementary WiFi and Bluetooth modules for the system design.

msword - 58.50 kB - 06/12/2017 at 06:43


P2P Local Mesh Communications Network for Developing Countries Rev April 30 2017 D.doc

Basic overview of the system design and rationale for the design.

msword - 116.00 kB - 05/01/2017 at 10:08


  • 1 × 2N2222 Discrete Semiconductors / Transistors, MOSFETs, FETs, IGBTs

  • And Beyond

    hlew10/21/2017 at 09:44 0 comments

    In lookng at other developing region projects I ran across the R.A.C.H.E.L. (Remote Area Community Hotspots for Education and Learning) project.

    Essentially they distribute Raspberry PI cached webservers pre-loaded with education content so no need for Worldwide Web access to get content.

    Mesh networks would be a great to extend the reach outside of libraries and classrooms to the community even in isolated regions with no WWW internet.

  • A Pause in the Journey

    hlew10/21/2017 at 08:01 0 comments

    As with any project it is interesting to see where we started and look at where we actually ended up. 

    As I post the Technical Specification document that covers all the user and performance assumptions,

    an overview of the solution and the instructions and documentation on how to build it, it is strange

    to see how this has changed from the simple idea I roughed out on a piece of paper

    in the conceptual round of the HackADay prize. 

    Many of the early assumptions did not prove out to be true. More open source component parts

    in terms of software already existed than I knew about. It was also interesting to see the different

    approaches people had already tried to bring communications infrastructure to under developed regions.

    Some were too grandiose and did not consider the local cultural and social norms. Some made totally

    different assumptions on what was needed. Ultimately it was good to build on the "shoulder of giants"

    because very little is totally new these days even in technology. Often they are just ideas that society

    is now ready to except (who knew people want to share so much of their lives on social networks). In other

    case it was a idea that technology finally made practical (CDMA was developed by the screen

    actress  Hedy Lamarr and pretty much everyone has some of her genius in their pockets).

    In light of all the natural disaster recently the importance of communications or the lack of it

    is front and center in many peoples mind. So this is not the end of journey but just a pause.

  • Video finally live!

    hlew10/21/2017 at 03:52 0 comments

    The video of the project is finally live. It is hard to put together an effective video using open source music and images. 

    It hopefully balance information with humor.

    All images in the video are from sources that provide open source license or are the property of the authors. 

    The attribute of all images and music will come next when there is time for the poor and tired engineers to rest

    after the HackADay prize deadline. 

    It feels like university and it is time to cram for exams.

  • Cantenna: The Poor Mans' Directional Antenna

    hlew10/20/2017 at 09:58 1 comment

    Yagi or parabolic antennaes are the most effective at concentrating the power of the WiFi transmitter in a specific direction but they are complex to design and construct. In order to simplify the construction process and make it easy to construct locally, a cantenna design was chosen. The cantenna is an abbreviation of the word “can” and antenna. In it simplest form it is literally is a can with a pickup mounted a set distance from the closed up of the can.

    The most complete information on the optimum parameters to use for a cantenna are covered in the book “Cantenna the Poor Wifi Hackers Choice” by Sven-Erik Zetterström. 1 The important parameters are the diameter of the tube used to form the antenna, distance of the pickup from the end of the tube and the length of the tube

    1 Sven-Erik Zetterström. “Cantenna the Poor Wifi Hackers Choice”. .

    There are a range of values each dimension can be for the cantenna but the approximate starting values are:

    •The length of the can, (L) should be longer than 3/4 of the wavelength

    •The diameter of the can (D) should be longer than 1/2 of the wavelength

    •The copper pickup element should be approximately 1/4 of the wavelength

    •The position of the copper pickup element (y) away from the back of the cantenna (rear standoff) is based off the overall diameter of the can and can be calculated using this on-line calculator:

    As mentioned above the copper pick up elements should be approximately ¼ of the wavelength of the carrier frequency. WiFi typically has 14 overlapping channels with most WiFi units are shipped set to channel 6. Some Wi-Fi channel numbers overlap with each other. Channel 1 uses the lowest frequency band and each subsequent channel increases the frequency slightly. Therefore, the further apart two channel numbers are, the less the degree of overlap and likelihood of interference. The three Wi-Fi channels 1, 6 and 11 have no frequency overlap with each other for 2.5 GHz WiFi

    Newer 802.11n and 802.11ac Wi-Fi networks also support 5 GHz wireless connections. These frequencies are much less likely to suffer from wireless interference issues in homes the way 2.4 GHz does. Additionally, the 5 GHz Wi-Fi channel choices available in most home network equipment have been pre-selected to choose only the non-overlapping ones.

    Choices vary by country, but in the United States these non-overlapping 5 GHz channels are most recommended: 36, 40, 44, 48, 149, 153, 157 and 161.

    Usable non-overlapping 5 GHz channels also exist between 48 and 149, specifically 52, 56, 60, 64, 100, 104, 108, 112, 116, 132, and 136. These channels fall into a specially regulated category where a Wi-Fi transmitter is required to detect whether other devices are already transmitting on the same channel and automatically changes its channel to avoid conflict.”1


    As a run of thumb for 2.54 GHz WiFi cantennae:

    Driven element length never, ever, changes, and is always 31 mm. You will see many instances of online advice that say different; they're wrong. To optimise it for Channel 1, you can make it 0.5 mm longer; to optimise for Channel 11, then 0.5 mm shorter. Otherwise, leave well alone. For practical purposes, it is always left at centre band optimum length, which is around 31mm.” 1

    The current design follow this rule of thumb without issues. Implementations should vary the pickup length depending on the needs for their particular system though. In most rural installation it should not be necessary to experiment...

    Read more »

  • Solar Charging Circuitry for the Nodes

    hlew10/12/2017 at 00:49 0 comments

    Each node in the mesh network is design to run independent  of AC power. Each node has a solar panel that powers a charging circuit that feeds into an array of 6 volt batteries.  The problems with solar power is that there will be times when not enough sun hits the panel to supply the current required to power the node. and at night the system needs to run fully off the batteries. The AMP-HOUR capacity of the batteries therefore need to be calculated for a worst case scenario so the system is never shut down due to a lack of power.

    In addition to that the charging circuit needs to designed so the that batteries are not overcharged, Overcharging the batteries can damage them and result in a dramatic reduction in the functional life of the batteries  The follow schematic applies to lead acid or NiCd batteries. As one of the objectives of this project is to ensure the parts can be sourced in developing countries the charging circuit is designed with discrete components are opposed to using a pre-packaged power module. All the components in this charging circuit have been around for decades  and there are many equivalent parts from many manufacturers that can be substituted.

    This circuit is a modified version of the circuit on this website:

    Nominally approximately 12 volts is produced by most large solar panels. The current from the solar panel passes  through the diode D1 to the voltage regulator, LM317.  The output of the voltage regulator is tuned by the variable resistor (potentiometer) VR to set the proper charging voltage for the battery. D1 prevents the solar cell from drawing power from the battery circuit when it is not actively producing current such as at night or on a very cloudy day. When the voltage on the battery reaches greater than 6.8 volts  (fully charged battery), the zener diode (zd) break down and turns on the NPN transistor T1 which will then shunt the current to ground through the resistor R1. The goal is to make sure the battery does not get overcharged. Overcharging the battery dramatically reduces the life of the battery as the elevated temperature due to excess current affects the chemicals in the battery.

    The web link above also provides a circuit board which can be used in the design. To save the cost of doing a minimum production run of the boards, a prototype board with soldered connections between components can also be used to implement the circuit. This was the method of fabrication for circuits before the existence of circuit boards. This gives the community implementing the system a more cost effective approach if they only need to add  one more node to their network.  The other parts of a node do not require that they be manufactured as the other components of a node are off the shelf modules that are connected though off the shelf cables, Our goal was to create a system that could be essentially plugged together as opposed to assembled through a complex process.

  • Poor Man's Directional Antennae

    hlew07/24/2017 at 07:36 0 comments

    For those that do not want to the effort to create a true Yagi antennae we found this two articles on creating a Cantennae, essentially a tin card cut to the appropriate length. The key is cut the length to 1/4 wave, half wave or full wave of the main carrier frequency of the signal you are trying to detect.

    Cantennae construction basic:

    Wireless 2.4 GHz Directional Antenna Calculator

    Enter the can diameter and get the length of where to place the pick to detect the peak of the standing wave. from the end of the can.

  • Primer for Yagi Antenna design

    hlew07/18/2017 at 06:19 0 comments

    It has been a long time since I took electro-magnetics courses in university. It is time to take

    a primer on Yagi antenna design. Yagi antenna are very directional but it gives you a better

    signal to noise ratio and allows you to send and receive longer distances than other antenna

    designs because the directionality focuses all the power in a narrow cone. As with all

    antenna it is more effective if there are no obstructions between the sending

    and receiving antenna. Rain and other airborne particles can also cause scattering

    lessening the effective distance. Unobstructed line of sight is the rule of thumb if

    you do not want to do the calculations.

    Here is a good reference on how to construct Yagi antenna for a few dollars of materials.

    It just requires that you measure accurately.

  • Open Source Mesh Network Initiatives Compared

    hlew06/12/2017 at 08:57 0 comments

    It has been interesting to compare the PIC Mesh network architect described in the original specification document for this project with other open mesh network architectures that were designed to have more capability but require a specific operating system (Linux or x86) and a library of drivers to support it.

    If anyone knows of other open mesh network initiative that should be on that list let me know.

  • Addendum to Analysis of the Effective Range of Bluetooth Links

    hlew06/12/2017 at 08:48 0 comments

    I forgot to add that the discussion is in a document in the "Files" section for this project.

  • Analysis of the Effective Range of Bluetooth Links

    hlew06/12/2017 at 07:28 0 comments

    To reduce the complexity of what needs to be implemented to facilitate the actual messaging process on the handsets for the users, we decide to use an existing application call "Firechat". Firechat is interesting in that it can use either Bluetooth or WiFi infrastructure to deliver the message to the intended recipient. It also have the ability to use handsets in it local environment as a relay if it is not close enough to a Bluetooth or WiFi node.

    There are limitation on the effective distance a handset can be away from an active node. This update looks at some of the factors affecting the effective distance of a handset from a node for a Bluetooth link. The practical range of Bluetooth enabled handsets from the node determined the node spacing in the original overall specification document.

View all 13 project logs

  • 1
    Overview of project and location on instructions

    For simplicity of document control all instructions and component parts list are included in the "Technical Specification" document included in the files for this project.

    In addition all software components are available in the GitHub repository listed in the Technical Specification document and the project description.

    This is an open source project and the designs are free to be copied, modified and improved upon with appropriate attribution.

View all instructions

Enjoy this project?



zyndram wrote 07/22/2023 at 09:40 point

  Are you sure? yes | no

kewball wrote 05/26/2017 at 22:02 point have been working in this area for quite a while

  Are you sure? yes | no

hlew wrote 05/29/2017 at 06:09 point

Very cool. What has been the unmet challenges of mesh networks that have still not been addressed in the field? One of the comments below pointed out the problem with low densities per square mile/kilometer as one issue. 

  Are you sure? yes | no

Jean Pierre Le Rouzic wrote 05/17/2017 at 17:48 point

I have three remarks:
1) A Wi-Fi node every 30m means a density of one thousand of active users per square kilometers, a situation one can expect in a urban area like San-Diego but probably not in rural areas.
2) The floor plan of Wi-Fi is -80dB, when one could use -120dB and expect ranges in the km, more useful in rural areas.
3) Why using Wi-Fi at all? It is a mess with a spec having one thousand pages and nearly every feature protected by patents, instead you can use unlicensed 3G or another possibility, like what this article describes:

  Are you sure? yes | no

hlew wrote 05/29/2017 at 06:25 point

Lots of interesting points.

I have been looking at some articles on long range WiFi. 

At -120dB does the throughput suffer due to increased errors rates due to the lower signal to noise ratio? It has been a few decades since I took my communications courses in university so I might be phrasing the question wrong.

I have noticed a few papers that suggest that throughput suffers at longer ranges due to the longer propagation delays and delays in waiting for the error detection and re-transmission process to be complete. As the error rate goes up the throughput goes down as the system waits for the error to be confirmed and the re-transmission process to be complete. Intel created a system where scheduled time windows are used for the error detection process to be complete and the re-transmission to begin to cut down on the wasted bandwidth.

The unlicensed 3G look interesting but it looks like such systems are still at the mercy of National Regulators and it looks like not many regulators have a policy in place yet. While one could run an unauthorized system it would be a risk on the investment if regulators change their mind. 

  Are you sure? yes | no

Jean Pierre Le Rouzic wrote 05/29/2017 at 07:19 point

Thanks for your reply, and I really like a project like yours. Bravo for starting it!

Below are more precise answers. My point is mainly that by using Wi-Fi as a telecoms system means, you will encounter many roadblocks.
Wi-Fi was invented  as a low cost wireless LAN system for the 1995 office, it as incredibly evolved by borrowing ideas from telecoms networks (MIMO, advanced coding and modulation protocols) but vendors are selling very, very poor implementations (and customers like it!). Proper telecoms systems should have real antennas, not short wires, and there are horrors stories about Wi-Fi hardware. Simply using a decent antenna would bring a huge lap in quality.

More about it:
* Indeed you cannot plan to emit at -120dB, I was pointing at the noise floor which is too high in Wi-Fi if you plan to use it as a telecoms tools. Managing SNR is one of the most important thing, a too high SNR means wasting usable bandwidth, too low indeed is also inefficient.  Staying with noise at -6dB below the receiving level is a good compromise, but it is something that nobody can guaranty. Now people works on pathway models and evaluation methods (probing it with sending a dedicated frame, evaluating SNR at the receiver, sending a report to the sender).
* 3G systems can have cells that are 20km wide.
* The system of window is already part of the IEEE 802.11 standard. But the root problem in Wi-Fi is the collision mechanism, there two or three similar "managed" mechanisms in the 802.11 spec:
* "unlicensed" 3G does not mean not "illegal" 3G. Is it about running 3G standard in unlicensed bands. Some Telcos plan to do it anyway.
* Wi-Fi works already in unlicensed bands, which means you have *no* guaranties at all. Unlicensed bands are the dump bins of the radio spectrum.

  Are you sure? yes | no

Jean Pierre Le Rouzic wrote 05/29/2017 at 07:45 point

Another point is that the meaning of "unlicensed band" may vary from country to country as it is a legal concept.
For example in Japan or Europe there are not exactly the same ISM bands as in US, so do not assume that a casual Wi-Fi dongle can be used anywhere without authorization.
Some states do not like the idea of independent telecoms networks, it is a good idea to check the legal status for the countries where you want to deploy your project.

  Are you sure? yes | no

Michael Doody wrote 05/08/2017 at 22:48 point

hlew - 

Please read my project "Device for Seismic Noise Analysis" - one of the other semifinalists. Maybe we could ask Hackaday if we could merge our two projects into one killer project?

Michael d.

  Are you sure? yes | no

hlew wrote 05/15/2017 at 09:14 point

Hi Michael, 

I am on a killer travel schedule right now for my day job for the next few weeks. I literally just realized I made it through this round of the judging for the HackADay prize. I am literally back on the road in a few days and I just returned. 

I will try to look at your project later in the week. It looks like you are trying to predict earthquakes and then require a mesh network to distribute the information either to a central system for analysis or to broadcast an alert to  citizens in a specific population center?

  Are you sure? yes | no

Michael Doody wrote 05/16/2017 at 21:15 point

The idea would be to have a robust gossip network that gets information from multiple seismic sensors to a central computer where it is processed and then back to the community as a warning if a seismic event is predicted. There would need to be rules about what kind of events are passed on to the gossip community - made easier by the fact that my seismometer devices actually probability data as a primary output. The world is changing - even homes in very poor villages are now gaining access to the internet. In 5 years internet access will be almost universal. We could do this!

  Are you sure? yes | no

hlew wrote 05/29/2017 at 06:34 point

We might want to combine and re-enter in the "Open Design" category.   It would easier than asking for a change at this late date. I know you can enter the same design in multiple categories. I am going to try to implement the actual design for the project over the next few weeks but I am on a busy travel schedule so it might be touch and go whether I will be ready for the "Open Design" category deadline. I will let you know if I make the progress required to make a decent submission.

It looks like the new American federal budget calls for the elimination of funding for the research of an early warning earth quake system for the West coast. I was surprised that the system is only intended only gives a 45 second advanced warning. The idea is it gives enough warning to shut off gas and other critical systems to limit damage.  It is not enough to warn people but it is enough to limit infrastructure damage.

  Are you sure? yes | no

Similar Projects

Does this project spark your interest?

Become a member to follow this project and never miss any updates