Project Canopy: A Community-Owned Mesh Network for Rural Connectivity

Project Canopy was born from a simple frustration: rural communities in our region pay exorbitant prices for spotty, throttled internet from distant providers who have no incentive to improve. The solution, we believed, was not to lobby the incumbents but to build our own infrastructure—open, modular, and owned by the people who use it.

The core idea is a self-forming mesh network where every household or community center can host a node made from off-the-shelf hardware (Raspberry Pi 4, long-range Wi-Fi antennas, and a solar panel where needed). The nodes run custom open-source firmware based on OpenWrt and BATMAN-adv, automatically discovering neighbors and routing traffic without a central authority. A lightweight portal handles local authentication and bandwidth sharing via a cooperative membership model—each participant contributes a node and gets access in return.

We envisioned a phased rollout: first a pilot in one village to prove the concept, then a cooperative legal structure to manage the network collectively, and finally a mesh that could grow organically as new nodes join. The software stack would be fully open, documented in plain language, so that any community with basic technical skills could adapt it. We even designed a 3D-printed enclosure and a simple assembly guide, aiming to turn the project into a replicable kit.

Project Canopy was never about competing with commercial ISPs—it was about resilience, dignity, and the right to communicate. It was a bet that small, local networks could be more robust and equitable than centralized ones. That bet never got placed, but the plan remains a blueprint for anyone who wants to try.

Rural internet access is a persistent failure of the market. Incumbent ISPs cherry-pick dense urban areas, leaving large geographic swaths with slow, expensive, or nonexistent connectivity. When service exists, it's often a single bottlenecked DSL line or a satellite link with high latency and data caps. Communities have no recourse—they are customers, not participants. This power imbalance stifles education, healthcare, and economic opportunity. The digital divide is not just a gap; it's a wall. Existing solutions like municipal broadband or cellular backhaul require massive capital and political will. Project Canopy aimed to bypass both by building something small, cheap, and owned locally.

• Deploy a functional 10-node mesh in a single rural village within 6 months of starting.
• Develop and release all firmware as open-source (GPLv3) with documentation in English and Spanish.
• Form a legally recognized cooperative that holds the network's assets and governance in trust for members.
• Train at least 5 local technicians per node community to install, maintain, and troubleshoot the hardware.
• Establish a sustainable funding model through member contributions and small grants, covering hardware and spectrum fees.
• Create a replication kit (CAD files, assembly video, firmware image) so any community can start their own canopy.

• Not a replacement for commercial ISPs in areas where they already work well.
• Not a free public Wi-Fi for non-members; participation requires either a node or a small fee.
• Not a single centralized firmware that cannot be forked; we encourage community-specific modifications.
• Not a platform for advertising, surveillance, or data mining.
• Not a dependency on proprietary hardware; all components must be available from multiple sources.
• Not a project that scales to a city-wide network; focused on small, resilient clusters.

Hardware:

• Single-board computer: Raspberry Pi 4 (2GB+), with PoE hat or solar charger.
• Wi-Fi radios: 2x TP-Link AC1200 USB adapters (for 2.4/5 GHz mesh links) or long-range Ubiquiti LiteBeam for backhaul.
• Antennas: 5 dBi omnidirectional or 15 dBi directional panel antennas (weatherproof).
• Power: 50W solar panel + 12V battery + charge controller for off-grid nodes; mains power with UPS for village hubs.
• Enclosure: 3D-printed PETG case with vents and waterproofing, or repurposed electrical junction box.

Software:

• Operating system: OpenWrt (custom build with kmod-batman-adv).
• Mesh routing: BATMAN-adv with V7 compatibility and B.A.T.M.A.N. gateway detection.
• Management interface: luci-app-canopy (custom web UI for node status, bandwidth usage, and membership).
• Authentication: CoovaChilli with custom RADIUS backend for membership-based access.
• Firmware toolkit: Buildroot-based system to generate reproducible images.

Tools:

• CAD: FreeCAD for enclosure design.
• Documentation: GitBook with Markdown and embedded 3D model viewer.
• Communication: Matrix (homeserver on a pilot node) for community chat and alerts.

Project Canopy is a decentralized mesh network. Each node acts as both a client and a router. The topology is flat: every node within radio range of another can form a link, and the mesh routing protocol (BATMAN-adv) dynamically chooses the best path to the internet gateway. There is no central server—the network is fully distributed. Internet access is provided by one or more gateway nodes that have a backhaul link (e.g., satellite dish, existing DSL, or a fiber connection from a nearby town). These gateways are owned by the cooperative and share bandwidth among members via a fair queuing algorithm.

Membership is managed by a lightweight cooperative smart contract (on a simple blockchain or a distributed ledger like Hyperledger Fabric) that records node ownership and bandwidth shares. However, the critical point is that the network works even without the ledger—it's only for accounting. Each node broadcasts its cooperative membership token to neighbors, and the authentication portal allows access only to nodes with valid tokens.

The system boots from a SD card flashed with the Canopy firmware. On first boot, the node runs a wizard: scanning for existing meshes, configuring radio channels (auto-negotiated to minimize interference), and setting a password for local management. From then on, it's plug-and-play.

• Regulatory spectrum issues: Wi-Fi in the 2.4/5 GHz bands is mostly unlicensed, but outdoor point-to-point links may require registration in some countries.
• Technical skill gap: Rural communities may not have the technical expertise to assemble and maintain nodes; we underestimated the training burden.
• Cooperative governance complexity: Forming a legal cooperative and sustaining democratic participation is hard; the board could become inactive.
• Physical security: Nodes on rooftops or poles could be stolen or vandalized.
• Funding interruption: Reliance on grants makes the project fragile; membership fees alone may not cover maintenance.
• Spectrum interference: In dense mesh environments, self-interference can degrade performance; aggressive frequency hopping may be needed.

• Should we use a separate ISM band (e.g., 915 MHz) for a control channel to bootstrap the mesh?
• What is the optimal membership-to-bandwidth ratio? We assumed 1 node per household, but bandwidth demand varies wildly.
• How do we handle nodes that go offline permanently? Should the cooperative have the right to de-authorize? legal implications.
• Is a blockchain-based token necessary, or is a simple database sufficient for a small cooperative?
• How do we ensure the firmware remains updatable without breaking existing mesh links?

The core team of four met weekly for six months, built a handful of prototypes, and even started a pilot in a single village. But a key grant application was rejected, and then one member relocated, another had a child, and the remaining two couldn't sustain the momentum. The hardware and plans sit in a storage locker, waiting for someone to pick them up again.

The project's name came from the idea of a forest canopy—each tree (node) connected to its neighbors, forming a protective layer overhead. All design files and code are on GitHub under a Creative Commons license, but they've had few contributions. The most complete piece is the assembly guide, which is sadly only in English. If you're interested in reviving this, start by joining the Matrix room (still active, barely) and reading the 'handbook' in the repo. The spectrum analysis we did for the 5 GHz band is still valid.