multi-modal underwater communication dataset

Underwater Drone Network Telemetry Dataset (URRC–ODI, 2021–2025)

This dataset contains real-world, high-frequency underwater communication and robotics telemetry collected between January 2021 and September 2025 from the Underwater Robotics Research Centre (URRC) at the Norwegian Marine Technology Institute (NTNU OceanLab, Trondheim, Norway). The dataset is jointly curated under the Ocean Data Initiative (ODI) to support scientific research in underwater networking, marine robotics, and resilient acoustic communication systems.

Data were captured from autonomous underwater vehicles (AUVs), relay nodes, and coastal base-station links operating in structured missions, dynamic patrol routines, and long-distance acoustic communication trials. Sampling was performed at a 15-minute frequency, producing more than 1.5 million multi-modal records stored in a single CSV file.

Dataset Contents

Each record contains synchronized measurements across six major categories:

1. Communication & Acoustic Signal Parameters

Signal strength (dB), SNR (dB), channel attenuation, bandwidth (kHz)

Propagation delay (ms), frame/bit error rates

Transmission and received power (W)

Modulation type, frequency drift, packet transmission rate

2. Network Flow & Topology Metrics

Source/destination node identifiers

Hop count, path stability index

Packet size, retransmission counts

Throughput (kbps), jitter (ms), end-to-end delay (ms)

Network density and routing protocol type

3. Mobility & Localization Attributes

Drone depth (m), drone speed (m/s)

Base-station distance (m), position error (m)

Movement type (hover/patrol/relay)

Heading angle (degrees), communication range (m)

4. Environmental & Oceanographic Conditions

Water temperature (°C), salinity (ppt), turbidity (NTU)

Current velocity (m/s), ambient noise level (dB)

Hydrostatic pressure (kPa), pH level, light intensity (lux)

5. System Health & Energy Indicators

Battery level (%), transmission energy (J)

CPU and memory utilization (%)

Sensor health score, transmission queue length

6. Security & Behavioral Indicators

Packet drop ratio, unexpected beaconing frequency

Invalid packet rate (%), network entropy

Node-degree changes, latency spikes

Duplicate ACK ratio, control-message ratio

Ground-Truth Labels

Each timestamped record includes two labels:

Attack_Type — 8-class categorical label

Normal

Packet Loss

Routing Manipulation

Energy Exhaustion

Jamming / Interference

Replay Attack

Localization Error

Delay Spike

Anomaly_Label — Binary label (0 = normal, 1 = anomaly)

File Format

underwater_drone_network_dataset.csv

UTF-8 encoded, comma-separated

Includes Timestamp column and 65+ feature fields

Ethics & Availability

The dataset contains no personal or sensitive human data. All measurements originate from mechanical and environmental sensors.This dataset is publicly released under the Ocean Data Initiative (ODI) guidelines to promote reproducible marine robotics research.