I’m just published the first formal specification of PeerTap — a censorship-resistant, anonymous, decentralized Layer 2 overlay network. PeerTap creates cryptographically isolated Ethernet broadcast domains (CryptoVLANs) over libp2p, supporting encrypted unicast, broadcast, and multicast via uniform packet forwarding.
Key features:
- Raw Ethernet encapsulation via TAP/TUN
- MLS (TreeKEM) group encryption inside VLANs
- Anonymous VFT-based routing (no IPs, no metadata)
- Trunk-level E2EE, stateless relay mesh
- Spanning tree for efficient broadcast/multicast
- Native support for Linux, Windows, macOS; mobile via TUN
Why it matters
Most peer-to-peer libraries like libp2p and waku are tightly coupled to specific applications. As a result, each app builds its own isolated p2p island, with small domain of global connectivity and weak resilience to mesh topology dynamic. The fewer users within such a segment, the less stable and useful it becomes.
PeerTap takes a different approach: it provides a common p2p network substrate for all applications — including those that don’t use any p2p libraries themselves. Apps interact through cryptographically isolated VLANs over a shared global infrastructure. This incentivizes all users — regardless of app — to join the same encrypted mesh, benefiting from shared connectivity, relays, and anonymity.
PeerTap aims to become the Internet of comprehensive decentralized, privacy-preserving overlay networks. Feedback, critique, and contributors are welcome.