Datasets of Indoor Wireless Channel Measurements for Machine Learning Applications

This is a simple dataset of raw IQ measurements on a point-to-point wireless indoor channel, captured in a static laboratory environment. A sequence of random symbols (either QPSK symbols or random Gaussian symbols) are passed through a raised-root-cosine filter and modulated to different frequencies (433 MHz, 708 MHz, 2450 MHz). These measurements are carried out for varying signal-to-noise conditions (approximately 0 dB, 10 dB, 20 dB, estimated pre-measurement). These datasets may be used, for example, for data-driven channel modeling with state-of-the-art AI and machine learning algorithms (e.g., generative adversarial networks), for validating conventional channel models against real measurements, or for investigating the properties of real wireless channels. A detailed description is provided in the file README.pdf