Carotid UHF‑RFID Heartbeat Detection Master Dataset (Subjects 1–15)

This dataset contains raw S11 return-loss time series recorded from a passive UHF RFID tag placed on the skin over the carotid artery of 15 human subjects (ages 20–50), together with per-subject acquisition metadata and per-trial summary statistics. For each subject, S11 was sampled at 20 Hz over a continuous 30-second window (600 samples per trial), for five trials in each recorded body position. Subjects 1–5 were recorded sitting, standing, and lying; Subjects 6–15 sitting and standing only. Heart rate was recorded via pulse oximetry once at the onset of each trial and serves as the reference. S11 was measured directly as a one-port reflection measurement using a vector network analyzer connected to a UHF reader antenna; no RFID reader interrogation or backscatter link was involved. S11 is reported as return-loss magnitude in positive dB, so higher values indicate less reflected power and a better antenna–tissue impedance match. The data spans three acquisition conditions, which should be treated as separate strata and not pooled. Subjects 1–5 were recorded with the original acquisition chain. Subjects 6–15 were recorded after a noise-cancellation stage was added, and differ further in the frequency waveband over which the S11 minimum was tracked: 860–930 MHz for Subjects 6–10, narrowed to 890–920 MHz for Subjects 11–15. The cohort boundary in signal quality falls exactly on this change, but the acquisition records do not establish it as the cause. Important limitation. Subjects 6–10 contain no recoverable cardiac component at any rate. The dominant periodicity in those recordings is respiratory, at approximately 0.29 Hz (17.3 breaths/min). These recordings must not be used to benchmark, train, or validate heart-rate estimators. They are retained as a labelled negative control, useful for testing whether an estimator correctly abstains when no cardiac signal is present, and for studying respiratory contamination of carotid S11 recordings. Subjects 11–15 are the appropriate subset for heart-rate algorithm benchmarking; Subjects 1–5 carry a recoverable but erratic cardiac component. Subject 13 is pacemaker-dependent and provides a near-fixed-rate control rhythm. Subject 5 exhibits the highest heart rates in the cohort. Subject-level records are limited to age, sex, self-reported fitness rating, and the health notes documented in the accompanying descriptor; no identifying information is included. Intended for research use in physiological sensing, RFID and RF signal processing, and biomedical instrumentation. Users should consult the accompanying data descriptor before use.