Dealing with crosstalk in electromagnetic field measurements of portable devices

Portable devices measuring radiofrequency electromagnetic fields (RF-EMF) are affected by crosstalk: signals originating in one frequency band that are unintentionally registered in another. If this is not corrected, total exposure to RF-EMF is biased, particularly affecting closely spaced frequency bands such as GSM 1800 downlink (1,805-1,880 MHz), DECT (1,880-1,900 MHz), and UMTS uplink (1,920-1,980 MHz). This study presents an approach to detect and correct crosstalk in RF-EMF measurements, taking into account the real-life setting in which crosstalk is intermittently present, depending on the exact frequency of the signal. Personal measurements from 115 volunteers from Zurich canton, Switzerland were analyzed. Crosstalk-affected observations were identified by correlation analysis, and replaced by the median value of the unaffected observations, measured during the same activity. DECT is frequently a victim of crosstalk, and an average of 43% of observations was corrected, resulting in an average exposure reduction of 38%. GSM 1800 downlink and UMTS uplink were less often corrected (6.9% and 8.9%), resulting in minor reductions in exposure (7.1% and 0.92%). The contribution of DECT to total RF-EMF exposure is typically already low (3.2%), but is further reduced after correction (3.0%). Crosstalk corrections reduced the total exposure by 1.0% on average. Some individuals had a larger reduction of up to 16%. The code developed to make the corrections is provided for free as an R function which is easily applied to any time series of EMF measurements. Bioelectromagnetics. 39:529-538, 2018. © 2018 Wiley Periodicals, Inc.