anilcosaran / passerby

Peer-to-peer location services for the decentralized web -- discover what's around you, no central authority required

passerby

🧭 Table of contents

• What is Passerby?
• Research emulator
• Design goals
• The asterisk
• Technology overview

🌎 What is Passerby?

Passerby is a peer-to-peer protocol for location-aware resource discovery. Using Passerby, you can find nearby people or resources — or help people find you — without requiring a coordinating central authority.

It's Byzantine fault tolerant^* and it supports a full node on mobile devices.

You might use it to build decentralized hyperlocal applications — like food delivery, ride hailing, or online dating. More broadly, though, our goal is to create the location layer for the decentralized web — that is, a persistent, decentralized, and scalable mechanism for coordinating with peers based on geographic proximity.

🔨 Research emulator

The gif above was captured in Passerby Park — a graphical research emulator for studying the behaviors of Passerby networks.

✏️ Design goals

Low latency, high frequency updates

Passerby should support decentralized applications in which peer locations change rapidly, like ride hailing.

Mutual offline discovery

If Bob discovers Alice while Alice is experiencing a temporary loss of connectivity, Alice must hear about it when she reconnects — even if at that point, Bob is far away.

Radical portability and zero dependencies

This reference implementation is designed to work in a variety of disparate JavaScript runtimes, including Node.js and Hermes (React Native). Passerby relies only on libsodium for cryptographic primitives.

❗ The asterisk

Passerby is currently in development. It is not production-grade software. Security vulnerabilities are likely to exist, both at the level of protocol design and concrete implementation.

💾 Technology overview

Our research topics include distributed data structures, space filling curves, and applied cryptography. If that sounds interesting to you, consider becoming a contributor.

Thorough documentation is forthcoming. Until then, here's a guide to our source layout:

Module	Description
consensus	Byzantine consensus
core	math, crypto, logging, cross platform compatibility, elementary data structures
dht	distributed hash table
pht	distributed trie
protocol	protocol logic
psm	passerby state machine
repman	dynamic replica management
transport	transport layer abstraction (reliable UDP or local network simulation)
whoami	STUN-based NAT traversal and self-identification