Skip to content

Getting Started

What IONIS Predicts

IONIS (Ionospheric Neural Inference System) predicts HF radio signal-to-noise ratio (SNR) for any transmitter-receiver path given geographic coordinates, frequency, time of day, and solar conditions.

The model learns from 13B+ real-world amateur radio observations — WSPR beacons, Reverse Beacon Network spots, and contest QSOs — combined with solar flux and geomagnetic indices. Unlike traditional tools such as VOACAP, IONIS captures empirical propagation patterns directly from observed data.

Key metrics (V20 Production):

Metric IONIS V20 VOACAP
Pearson correlation +0.4879 +0.0218
PSK Reporter recall 84.14% —
Training data 31M signatures Physics model

Installation

IONIS packages are available from the COPR repository for Rocky Linux 9 / RHEL 9 / Fedora.

Enable the COPR Repository

sudo dnf copr enable ki7mt/ionis-ai

Install Packages

# Core schemas and configuration
sudo dnf install ionis-core

# Pipeline apps (ingesters, downloaders, validators)
sudo dnf install ionis-apps

# CUDA embedding engine (requires NVIDIA GPU)
sudo dnf install ionis-cuda

Verify Installation

# Load environment variables
source ionis-env

# Check ClickHouse connectivity
clickhouse-client --query "SELECT 1"

# List installed DDL files
ls /usr/share/ionis-core/ddl/

Quick Start

After installing the RPMs and setting up ClickHouse:

# 1. Source the environment
source ionis-env

# 2. Apply database schemas
for f in /usr/share/ionis-core/ddl/*.sql; do
    clickhouse-client --multiquery < "$f"
done

# 3. Verify tables exist
clickhouse-client --query "SHOW TABLES FROM wspr"
clickhouse-client --query "SHOW TABLES FROM solar"

# 4. Check table counts (after data is loaded)
db-validate --all

What's Next