Edge Swarm

EdgeSwarm is a Web2.5 Decentralized Physical Infrastructure Network (DePIN) designed to execute distributed AI inference, web scraping, and tensor math across a global swarm of consumer devices. The network utilizes an off-chain mempool for high-speed task routing and settles consensus and $SWARM token payouts on-chain via the Base Sepolia blockchain.

📱 1. Android Node (Mobile Worker) The Android node operates as a highly optimized, headless background service capable of executing genuine offline Large Language Model (LLM) inference and secure web scraping.

Minimum Hardware Requirements To ensure network stability and prevent Out-Of-Memory (OOM) operating system crashes, mobile devices must meet the following specifications:

OS: Android 8.0 (Oreo) or newer (API Level 26+) RAM: 8GB minimum (12GB+ highly recommended) Processor: 64-bit architecture (ARM64-v8a) Storage: 3GB+ free internal storage (for the Gemma 1.5GB model file and swap space)

Key Features MediaPipe AI Engine: Runs Google's gemma-2b-it-cpu-int4.bin model entirely offline. Safe Chunking: Bypasses Android's aggressive memory management by streaming the 1.5GB model into internal storage via a 4MB chunked buffer. Web3j Integration: Autonomously generates ECDSA key pairs and cryptographically signs payloads (Task:ID|Score:100|Hash:HASH|HW:ID) before returning them to the enterprise server. Smart Routing: Identifies as PIX10_... (or equivalent device ID) to intercept mobile-friendly tasks like secure HTML web scraping.