Overview

NearWave is a command-line tool for transferring data between devices using sound. It encodes arbitrary input into audio signals — audible or ultrasonic — and decodes them on the receiving end. No network, no pairing, no Bluetooth. Just a speaker and a microphone.

The system handles compression, optional encryption, chunked encoding, and error correction. It is designed to work reliably in real-world acoustic environments with ambient noise and hardware variation.

Core Idea

Sound is a viable transport layer for short-range data transfer. Every device with a speaker and a microphone already has the hardware. NearWave treats audio as a physical-layer protocol — structured, compressed, and optionally encrypted — rather than a novelty.

No network infrastructure required
No pairing or handshake over Bluetooth or Wi-Fi
Works across any combination of devices that can produce and capture sound
Operates in environments where radio-based communication is restricted or unavailable

Capabilities

Audible and ultrasonic transmission — configurable frequency ranges depending on environment and hardware
Compression — supports zstd and brotli to minimize transmission time
Optional encryption — AES-256-GCM for payloads that require confidentiality
Chunked encoding — splits data into independently decodable frames with per-chunk CRC validation
Error correction — Reed-Solomon codes for forward error correction over noisy channels
Adaptive gain — adjusts output amplitude based on ambient noise floor measurements

How It Works

Send Path

input → compress → encrypt → chunk → encode (FSK/OFDM) → audio output

Receive Path

audio input → demodulate → reassemble chunks → decrypt → decompress → output

Each chunk carries its own header with sequence number, CRC, and encoding metadata. The receiver reconstructs the stream independently per chunk, allowing partial recovery if frames are lost.

CLI Interface

Send a file:

nearwave send --file notes.txt --profile reliable

Receive and write to disk:

nearwave receive --output received.txt

Send with encryption:

nearwave send --file payload.bin --encrypt --key $NEARWAVE_KEY

Benchmark a profile:

nearwave bench --profile fast --size 4096

Performance

Benchmarking covers compression ratio, encoding latency, and end-to-end throughput across profiles and payload sizes.

Metric	Reliable	Fast	Ultrasonic
Throughput	~120 bps	~450 bps	~200 bps
Compression (zstd, 4KB text)	62% reduction	62% reduction	62% reduction
Encoding latency (4KB)	~34s	~9s	~20s
Error tolerance	High	Low	Medium

Compression is applied before encoding, so gains compound directly into reduced transmission time. Brotli achieves higher ratios on text payloads but adds ~15% more CPU time compared to zstd.

Profiles

Reliable — lower frequency range, wider tone spacing, stronger error correction. Best for noisy environments or longer distances. Slowest throughput.

Fast — higher symbol rate, narrower tone spacing, minimal redundancy. Best for quiet, close-range transfers where speed matters.

Ultrasonic — operates above ~18 kHz. Inaudible on most hardware. Requires devices with high-frequency speaker and microphone response. Throughput sits between reliable and fast. Useful when audible noise is unacceptable.

Use Cases

Offline data sharing — transfer text, config snippets, or small files between machines without any network
Air-gapped communication — move data across security boundaries where network connectivity is intentionally absent
Device-to-device transfer — quick exchange between a laptop and a phone, or between two terminals in the same room
Restricted environments — contexts where Wi-Fi and Bluetooth are disabled or monitored

Philosophy

NearWave is a protocol and a system, not just a tool. The CLI is one interface to a layered architecture — compression, encryption, encoding, modulation — each designed to be independently testable and replaceable. The goal is a principled approach to acoustic data transfer that treats sound with the same rigor as any other transport medium.

Status

In development. Core send/receive pipeline is functional. Compression and encryption layers are integrated. Ultrasonic profile is experimental.

Notes

Active benchmarking across hardware combinations is ongoing. Improvements to adaptive gain control and multi-channel encoding are in progress. Profile tuning for specific environments (office, outdoor, industrial) is planned.