• Introduction
• Quick Start
  • Installation
  • Dependencies
  • Writing Data
  • Reading Data
• Limitations
• How To
  • Store I/Q data into H5 groups
  • Name I/Q datasets
  • Name channels
  • Supported data types and shapes
  • Add metadata
• License
• Contact

This is an HDF5 read-write Python library for the data format in Recommendation ITU-R SM.2117-0. ITU-R SM.2117-0 is a data format definition for exchanging stored I/Q data with the intention of spectrum monitoring. Key Bridge Wireless supports the ITU in creating a library for the data format as a contribution to IEEE 1900.8 Working Group.

The HDF5 is a generic and flexible file format for storing data and metadata. The ITU-R SM.211-0 describes a way to store IQ data using the HDF5 format. Below is an overview of the format.

Check the content of a given HDF5 file using a desktop application called Argos. Below is an example of a file opened with Argos that follows the specs in the ITU-R SM.211-0.

To get started, make sure to install Python version 2.8.2 on your machine. Use the following scrips:

pip install itusm2117

Direct dependiencies are the following.

• numpy is used for array manipulation.
• h5py is for reading and writing HDF5 files.
• cerberus is for metadata validation and normalization.

import numpy as np
from itusm2117 import write_iq_dataset

# The data: a list or np.array with complex values.
# Other data types and shapes are possible, check the docs.
iq = [1+1j, 2+2j, 3+3j, 4+4j]

# The only mandatory metadata of the Recomendation that
# doens't have a default or fixed value is the Sampling Frequency
# in Hertz.
sampling_frequency = 125e6

# Method for writing the data. First argument is the filename.
# By default, it will store the iq data on a dataset named
# "Dataset_0" and channel named "Channel_0".
write_iq_dataset("my_iq_data.h5", iq, sampling_frequency)

from itusm2117 import read_iq_dataset

# Method for read the data. First argument is the filename and 
# the second is the dataset name.
metadata, recordings, channels = read_iq_dataset("my_iq_data.h5", "Dataset_0")
# metadata: A dict with the key-value pairs of metadata from "Dataset_0".
# recordings: np.array of complex64 with the IQ data. First dimension is the channel.
# channels: List with the channels names (can be only one).

Currently, this library does not support the following specs on the ITU-R SM.2117-0:

• Multisector datasets are used for IQ recordings with metadata changing over time
• Bitfield datasets are used for marking each sample of a recording with specific flags.
• Integet datasets. All datasets are stored as float32.

Check the following information for the different datasets and channel.

• Store I/Q data into H5 groups
• Name I/Q datasets
• Name channels
• Supported data types and shapes
• Add metadata

MIT

Contact us through our GitHub account for comments, questions, or pull requests.