The selection of an adequate exposure assessment approach is imperative for the quality of epidemiological studies. The use of personal exposimeters turned out to be a reasonable approach to determine exposure profiles, however, certain limitations regarding the absolute values delivered by the devices have to be considered. Apart from the limited dynamic range, it has to be taken into account that these devices give only an approximation of the exposure due to the influence of the body of the person carrying the exposimeter, the receiver characteristics of the exposimeter, as well as the dependence of the measured value on frequency band, channel, slot configuration, and communication traffic. In this study, the relationship between the field strength measured close to the human body at the location of the exposimeter and the exposure, that is, the field strength at the location of the human body without the human body present, is investigated by numerical means using the Visible Human model as an anatomical phantom. Two different scenarios were chosen: (1) For FM, GSM, and UMTS an urban outdoor scenario was examined that included a transmitting antenna mounted on the roof of one of four buildings at a street crossing, (2) For WLAN an indoor scenario was investigated. For GSM the average degree of underestimation by the exposimeter (relation of the average field levels at the location of the exposimeter to the field level averaged over the volume of the human body without the body present) was 0.76, and for UMTS 0.87; for FM no underestimation was found, the ratio was 1. In the case of WLAN the degree of underestimation was more pronounced, the ratio was 0.64. This study clearly suggests that a careful evaluation of correction factors for different scenarios is needed prior to the definition of the study protocol. It has to be noted that the reference scenario used in this study does not allow for final conclusions on general correction factors.

Based on numerical computations using commercially available finite difference time domain code and a state-of-the art anatomical model of a 5-year old child, the influence of skin conductivity on the induced electric field strength inside the tissue for homogeneous front-to-back magnetic field exposure and homogeneous vertical electric field exposure was computed. Both ungrounded as well as grounded conditions of the body model were considered. For electric field strengths induced inside CNS tissue the impact of skin conductivity was found to be less than 15%. However, the results demonstrated that the use of skin conductivity values as obtainable from the most widely used data base of dielectric tissue properties and recommended by safety standards are not suitable for exposure assessment with respect to peripheral nerve tissue according to the ICNIRP 2010 guidelines in which the use of the induced electric field strengths inside the skin is suggested as a conservative surrogate for peripheral nerve exposure. This is due to the fact that the skin conductivity values derived from these data bases refer to the stratum corneum, the uppermost layer of the skin, which does not contain any nerve or receptor cells to be protected from stimulation effects. Using these skin conductivity values which are approximately a factor 250-500 lower than skin conductivity values used in studies on which the ICNIRP 2010 guidelines are based on, may lead to overestimations of the induced electric field strengths inside the skin by substantially more than a factor of 10. However, reliable conductivity data of deeper skin layers where nerve and preceptor cells are located is very limited. It is therefore recommended to include appropriate background information in the ICNIRP guidelines and the dielectric tissue property databases, and to put some emphasis on a detailed layer-specific characterization of skin conductivity in near future.

A new head exposure system for double blinded human provocation studies, which requires EEG recording during exposure with GSM900- and UMTS-like signals has been developed and dosimetrically evaluated. The system uses planar patch antennas fixed at 65 mm distance from the subject's head by a special headset, which provides minimum impairment of the test subjects and ensures an almost constant position of the antennas with respect to the head, even in case of head movements. Compared to exposure concepts operating small antennas in close proximity to the head, the concept of planar antennas at a certain distance from the head produces a much more homogeneous SAR distribution in the temporal and parietal lobe of the brain. At the same time the resulting uncertainty of exposure due to variations in head size, variations of the dielectric properties of tissues and unavoidable small changes of the antenna's position with respect to the head, is reduced to the order of approximately 3 dB, which is a significant improvement to comparable head exposure systems reported in literature in the past. To avoid electromagnetic interference on the EEG recording caused by the incident RF-field an appropriate double-shielded filter circuit has been developed. Furthermore, the effect of the presence of the sintered Ag/AgCl EEG electrodes and electrode wires on the SAR distribution inside the head has been investigated and was found to be minimal if the electrode wires are arranged orthogonal to the incident electric field vector. EEG electrode arrangement parallel to the incident field vector, however, might cause drastic changes in the SAR distribution inside the head.

The relation between the incident electromagnetic field strength and both the whole-body and the local specific absorption rate (SAR) was investigated for typical heterogeneous exposure scenarios for frequencies relevant for mobile communication. The results were compared to results from plane wave exposure. Heterogeneous exposure arises from multiple path propagation of the electromagnetic waves to the location of interest. It is shown that plane wave exposure does not represent worst-case exposure conditions. When the electric field strength arising at plane wave exposure is compared to the electric field strength averaged over the volume of the human body occurring during multipath exposure, 12% of all heterogeneous cases examined represent worse exposure conditions than plane wave exposure for whole-body exposure at 946 MHz, 15% at 1840 MHz, and 22% at 2140 MHz. The deviation between plane wave and heterogeneous whole-body SAR ranges from -54% to 54%. For partial-body SAR averaged over 10 g of tissue, a range from -93% to 209% was found when comparing multiple wave exposure to single incoming plane waves. The investigations performed using the Visible Human as phantom showed that the basic restrictions are met as long as the reference levels are not exceeded. However, this must not be necessarily the case when different phantoms are used to perform similar investigations because recent studies demonstrated that reference levels might not be conservative when phantoms of children are used. Therefore, the results of this work indicate the need to extend the investigations to numerical simulations with additional human phantoms representing parts of the human population having different anatomy and morphology compared to the phantom used within the frame of this project. This also applies to phantoms of children.