Influence of User Mobility and Antenna Placement on System Loss in B2B Networks

In this paper, the influence of user mobility and on-body antenna placement on system loss in body-to-body communications in indoor and outdoor environments and different mobility scenarios is studied, based on system loss measurements at 2.45 GHz. The novelty of this work lies on the proposal of a classification model to characterise the effect of user mobility and path visibility on system loss, allowing to identify the best set of on-body antenna placements. To quantify the influence of visibility and mobility on the average system loss, a combined score is proposed, allowing to map system loss onto the degree of visibility and mobility that depends on the scenario being considered and on on-body antenna placements. Overall, a good agreement is observed between the proposed classification model and the average measured values of system loss, with the higher values of combined scores being associated with lower values of systems loss. For the cases under study, the average values of system loss are 61.6 dB for the cases of the antennas being positioned only on the front of the body and/or the head, and 64.5 dB if at least one of the antennas is placed on an arm.


I. INTRODUCTION
Wireless Body Area Networks, or simply Body Area Networks (BANs), are composed of a set of wireless devices attached to the human body [1], nowadays having a major role in many applications, such as the monitoring of vital signs, which is important for usage scenarios in medicine, sports, military, police, civil protection and even entertainment. The cooperative work between BANs, and in particular in Bodyto-Body (B2B) scenarios, has attracted a lot of attention, given the potential applications already envisaged for 5G and beyond. The characterisation of the channel in B2B scenarios, namely concerning path loss, is quite important for system design, since the location of the antennas on the body together with users' mobility will lead to quite diverse situations, concerning not only shadowing but also signal variability.
As mentioned in [2], B2B communications are extremely difficult to characterise, due to their specific characteristics.
The associate editor coordinating the review of this manuscript and approving it for publication was Wei Feng .
The radio channel is strongly influenced by several aspects: characteristics of the devices placed close to the body; characteristics, placement and orientation of antennas; specifications of radios and their frequencies; environment surrounding users associated with applications; and mobility of users, among many others. All these factors make the characterisation of B2B communications a complex process, leading to the need for models for a large variety of cases.
Several studies on B2B channel characterisation can be found in literature, namely based on measurements at the 2.4 GHz band. S.L. Cotton et al. [3] conducted measurements in a car parking outdoor environment, with antennas on the chest and the back of users in several mobility conditions, showing received power results, and these measurements were taken by S.L. Cotton [4] to analyse models for shadowing. R. Rosini et al. [5] performed indoor measurements in a room with users moving in different ways for several antennas' placements, leading to path loss and multipath fading models. A similar approach was taken by F. Mani and R. D'Errico [6], extracting path loss and shadowing and multipath fading models as well.
The measurements performed by S.J. Ambroziak et al. [7] were done at both indoor and outdoor environments, for a number of antennas placements larger than the previous ones, involving several mobility scenarios, leading to an analysis of system loss, in terms of average and standard deviation.
F.D. Cardoso et al. [8] addressed propagation inside circular metallic structures and propose a channel model assessed with measurements at 2.45 GHz in a passenger ferry discotheque with a circular shape. The measurement campaign focused on system loss measurements in the link between a transmitting antenna located on a human body in different locations and a receiving one positioned off-body. Different walking scenarios have been considered, allowing to perform the analysis of system loss for various conditions and on-body antenna placements. In [9], F.D. Cardoso et al. present a wideband characterisation of the propagation channel in circular metallic indoor environments, and an analytical model for the mean and the average delay spread, based on the results from [8].
In [10], M.E.H. El Azhari et al. used the scenario of a tunnel and users with antennas on the chest, extracting path loss, and shadowing and multipath fading parameters for several static positions of two users.
In general terms, these papers present path loss models based on the fitting of measurements, leading to an average decay rate with distance, and analyse the fitting of several statistical distributions (e.g., Log-Normal, Rayleigh, Rice and Nakagami).
Recently, F.D. Cardoso et al. [2] have analysed the B2B radio channel at 2.45 GHz, in indoor and outdoor environments and different mobility scenarios, for different on-body antenna configurations. For each environment, three mobility scenarios were measured and analysed, i.e., Approach, Departure and Parallel, corresponding to typical situations of day-to-day people's mobility. The on-body antenna configurations for both transmitter (Tx) and receiver (Rx) were right and left sides of the head, front side of the torso, front side of the waist and external sides of both arms at the wrist.
The novelty of the current paper lies on the extension of the analysis in [2], in order to identify the best set of on-body antenna placements by taking certain factors into account, such as the degree of visibility and mobility, as well as the stability of the propagation channel during the whole path.
The paper is organised as follows. The measurement environments setup and scenarios are briefly described in Section II. An analysis of measurement results in different scenarios is presented in Section III. The effect of user mobility and antennas is addressed in Section IV, together with the relationship between the measured value of system loss and visibility, and mobility conditions; also, a classification model for the choice of the on-body antenna location is proposed. A discussion on the choice of the best on-body antenna placement is presented in Section V. Conclusions are drawn in Section VI.

II. MEASUREMENT SCENARIOS, VISIBILITY AND MOBILITY A. DESCRIPTION OF SCENARIOS
As detailed in [2], measurements were performed in indoor and outdoor environments. The indoor environment was a corridor in one of the buildings of the Gdańsk University of Technology and the outdoor one was in the square in front of the previous building. The measurements were carried out by avoiding the influence of other people or objects (e.g., cars) moving in the outdoor environment.
The measurement campaign has already been detailed in [7] and [11]. The measurements were made at 2.45 GHz with a variable sample period (with an average of 150 ms and a standard deviation of 50 ms).
Both users wore a patch antenna with the same characteristics. Six different on-body antenna placements were considered, Fig. 1: right and left sides of the head (HE_R/L), front side of the torso (TO_F), front side of the waist (WA_F) and external sides of the right and left arms, at the wrist (AB_R/L). The antenna has a gain of 6.6 dBi, a radiation efficiency of 80.2 % and input return loss of 12.35 dB. The half-power beamwidths (HPBWs) in the E-and H-planes are 85 • and 95 • , respectively. As referred in [2], due to the low radiation into the body, its presence does not lead to significant distortion of the given antenna parameters, Fig. 2 [12], [13].   Three walking scenarios were considered, Fig. 3: Approach (A), Departure (D) and Parallel (P). In all scenarios, the walk routes of both B1 and B2 users were parallel, separated by 1 m, and walking for 6 m at the same time, so their speed was approximately the same as well. In Scenario A, users started at 6 m from the start line and stopped at the end line, while in Scenario D the situation was reversed, and in Scenario P both users walked in parallel for 6 m. For each Tx-Rx antenna configuration, scenario and environment, measurements were repeated 10 times.
Not all combinations of Tx-Rx antenna's placement were considered, since some of them would be redundant, Table 1 showing the measured 21 antenna placement configurations. One refers to the placement of the antennas according to the antennas pair Tx-Rx, e.g., HE_R-TO_F stands for the Tx antenna placed on the right side of the head and the Rx one at the front of the torso.

B. VISIBILITY AND PATH STABILITY
The antennas placement has quite an impact on signal behaviour, since the several configurations can lead to a variety of cases regarding the visibility between Tx and Rx antennas during the whole displacement, and, in addition, whether they are within the HPBW of each other. Fig. 4 shows the schematic of antennas' visibility for all cases.
One can see that the configuration (HE_R-HE_L, Scenario P) corresponds to a ''pure'' Line-of-Sight (LoS) (with the antennas ''seeing'' each other in the direction of maximum gain) during the whole displacement, while in (TO_F-TO_F, Scenario A), although there is indeed LoS between the two antennas during the entire path, they do not ''see'' each other within HPBW when they are close to the end line. Also, cases exist where Non-LoS (NLoS) is very clear, such as (TO_F-TO_F, Scenario D), others where a mixture of LoS and Quasi-LoS (QLoS) exists during the path, such as (AB_R-AB_L, Scenario P) due to the lack of synchronism between arms during mobility, as well as Obstructed-LoS (OLoS).
As previously mentioned, and illustrated in Fig. 5, the visibility between antennas depends on the relative position of both Tx and Rx. For the given geometry, the project distance between Tx and Rx antennas along the paths, d, at which the antennas start to see each other out of their HPBW can be evaluated as where: • W : Tx/Rx path separation; • 2ϕ: HPBW of the antennas. Hence, the percentage of distance (or time, assuming that the Tx and Rx move at constant speed), relative to the whole path length, 2L, for which the antennas are within the HPBW of each other, d b , is given by Considering the scenarios' geometry in Fig. 3, for an angle of ϕ = 42.5 • one gets d b = 91%; from now on, one refers to it as the long visibility path. Regarding the other case, which in this situation occurs only 9.0% of the time, one refers to it as VOLUME 10, 2022 the short visibility one. From the system loss viewpoint, long and short visibility paths correspond to the situations in which the antennas are ''seeing'' each other within the HPBW or not, respectively, hence, being associated with lower or higher values of system loss.
It should be noted that the definition of long and short visibility paths does not apply to scenario P, since both Tx and Rx keep the same relative position during the whole path. For the values of L and W being considered and antennas with ϕ > 45.4 • , which is the most usual situation in this type of communication systems, one gets d b > 80%. Thus, to classify the degree of visibility between Tx and Rx antennas one considers the geometry of the long visibility path (i.e., 91% of the whole path for Scenarios A and D).
The cases shown in Fig. 6 illustrate the different situations in terms of visibility, as a function of the antennas' alignment and body obstruction, where one considers the link to be within the HPBW (inHPBW) or outside it (outHPBW), in the latter one still having two cases: inside the outwards hemisphere (outHem), i.e., the one containing the main lobe where the antenna radiates outside the body; inside the inwards hemisphere (inHem), i.e., the one within which the antenna radiates into the body, which creates additional obstruction to the link. The relative positions of Tx and Rx antennas are classified into six categories, according to link between them: LoS, QLoS, OLoS and NLoS, the intermediate two being further differentiated between a strong ''s'' and a weak ''w'' signal strength. Table 2 shows the result of this classification, where background colours are used to enable an easier reading of the information. There is a symmetry between Tx and Rx in all these cases.
According to this classification, the visibility conditions for the given scenarios and antennas placement are indicated in Table 3.
It must be noted that the use of highly directive antennas is not addressed, since they are not appropriate to be used for BAN applications, like the ones being studied here, due to the level of mobility of the body itself, specially arms and legs, but also head and torso, thus, leading to a mix of LoS and NLoS situations due to the misalignment caused by the body movement. Still, the proposed approach can be used but different conditions will be observed for the long and short visibility paths, e.g., for scenario A, as illustrated in Fig. 4, the long visibility path will be mainly corresponding to a situation of outHPBW as opposite to inHPBW, hence, higher values of system loss will be observed. In any case each situation must be analysed and proper tables with the visibility  conditions must be derived, depending on the antenna characteristics and scenario under consideration.

C. MOBILITY
System loss depends not only on the visibility conditions but also on the mobility of different parts of the body, e.g., head, torso/waist and arms. For example, while TO_F, WA_F and HE_R/L can be considered relatively static to the body, AB_R/L are quite dynamic.
To characterise mobility as a function of antenna's position, three categories were considered: High (H), Medium (M) and Low (L), as defined in Table 4. In short, the links between antennas on the head, torso or waist can be considered with a Low mobility, while those between arms are of High mobility, mixed ones being considered Medium.

A. DATA PROCESSING
The measurements registered the received power, from which the system loss (the relationship between the power supplied to the input terminal of the Tx antenna and the one available at the output terminal of the Rx antenna, [14]) was easily obtained, the average and the standard deviation being then calculated by using Matlab R .
For illustration, an example of the comparison between measured system loss and theoretical one (obtained with the free-space path-loss model, with an exponent of 1.7, together with the influence of the antennas' radiation pattern) is illustrated in Fig. 7. One can observe that a good agreement is obtained both for indoor and outdoor environments. Also, the influence of visibility can be clearly identified, with the short visibility path (the lower values, being observed for 9% of the path) corresponding to distances up to 1 m. For larger distances, the long visibility path, here classified as a wQLoS situation, the system loss increases with distance, also because the HPBW of the antennas get more misaligned.   An example of the influence of mobility is presented in Fig. 8, where two different situations for Scenario P are illustrated: Low (HE_R-HE_L) and High (AB_R-AB_L).
The periodic movement of the arms is clearly identified in the High mobility scenario, where the period is approximately 2 s, which is in good agreement with usual walking conditions. The average system loss is not very distinct between the two mobility cases, i.e., 36.6 dB and 37.2 dB for indoor and outdoor environments in Low, and 38.8 dB and 38.7 dB for indoor and outdoor in High, respectively. However, as expected, the standard deviation is more sensitive to mobility, i.e., 2.3 dB and 2.8 dB for indoor and outdoor environments in Low, and 6.1 dB and 2.7 dB for indoor and outdoor in High, respectively; the higher influence of mobility indoors can be assigned to the presence of stronger reflections.

B. ANALYSIS OF MEASUREMENT DATA
This section addresses the dependence of system loss on the different antenna placements and scenarios; for conciseness, detailed results are presented only for indoors. A detailed analysis of the measurement results is presented in [2]. The figures show the average values in columns and the range of plus/minus a standard deviation in bars.

1) APPROACH SCENARIO
Scenario A is the one with the highest visibility between antennas, results being shown in Fig.9, in ascending order for the average, according to the visibility classification.
As expected, the lowest averages (below 54 dB) are observed for LoS cases (TO and WA cases), followed by sQLoS (a mixture of all cases), the highest ones (above  66 dB) occurring for wQLoS (just AB and HE cases). The highest standard deviations (above 9 dB) occur for wQLoS (AB_L-AB_L and HE_L-HE_L), showing the importance of both body mobility (not only the mobility of a person him/herself, but also of a persons' arms and legs) and antenna ''mutual'' visibility on system loss; the lowest standard deviations (around 5 dB) occur for sQLoS (WA_F-AB_R and AB_R-TO_F).
As detailed in [2], no significant distinction is observed between indoor and outdoor environments, but there are trends worthwhile exploring. The indoor average system loss is in the range [51.4, 67.0] dB while the outdoor one is within [49.1, 66.7] dB, global averages being 59.8 dB and 58.0 dB, respectively. Since this is the scenario with the highest visibility between antennas, one can expect that the surrounding environment will not have too much impact, hence, this small difference of 1.8 dB between the global averages. As far as standard deviations are concerned, the ranges are [5.2, 9.3] dB and [5.1, 13.4] dB, for indoors and outdoors, respectively, with corresponding global averages of 7.0 dB and 7.2 dB, confirming the previous statement on the impact of the surroundings.

2) DEPARTURE SCENARIO
Scenario D is the one where there are no LoS situations during the whole path, but just a few cases at its beginning, results being shown in Fig. 10. The lowest average system losses occur for wQLoS (just AB and HE cases), where HE_R-HE_R and HE_R-HE_L are the lowest, most probably due to reflections on nearby walls), while the highest correspond to NLoS (just TO and WA cases), the other cases in wQLoS being in the middle (again, a mixture of all cases).
Lower standard deviations occur for the cases where the average system loss is higher, and higher ones are associated with wQLoS paths with the highest mobility (e.g., AB_R-AB_R). These results reinforce the idea presented for Scenario A on the importance of both body mobility and antenna ''mutual'' visibility on system loss in B2B communications, confirming that system design in application scenarios should look into these aspects.
Average system losses range within [64.0, 79.9] dB for indoors and within [66.1, 85.1] dB for outdoors, with corresponding global averages of 71.5 dB and 75.9 dB. The higher values, compared with Scenario A ones (at least 10 dB), are clearly due to the lack of LoS. The global difference between indoors and outdoors, 4.4 dB, reflects the importance of the surrounding environments, i.e., of reflections in nearby objects, since in indoors there are many more reflections on walls, hence, having a major contribution for NLoS situations.
Standard deviations range within [4.7, 13.9] dB and [4.7, 12.2] dB for indoors and outdoors, respectively, with corresponding global averages of 8.0 dB and 6.6 dB. Again, these results show the importance of reflections on system loss, with indoors presenting a non-negligible higher value.

3) PARALLEL SCENARIO
Scenario P is the more varied one, ranging from strong LoS links to NLoS ones, results being presented in Fig. 11.
As expected, the configuration with the lowest average system loss is the one with a clear LoS and very low relative mobility, i.e., HE_R-HE_L, with minor impact from the surroundings (35.6 dB), also corresponding to the lowest standard deviations (2.0 dB). The effect of relative mobility can be clearly identified by comparing the previous situation (HE_R-HE_L) with AB_R-AB_L, where the arms are moving in a non-synchronised way between the two bodies, the average being 47.2 dB and the standard deviation 6.6 dB.
The cases with the highest averages are AB_L-HE_R and WA_F-AB-R, corresponding again to NLoS and wOLoS. The highest standard deviations (higher than 6.1 dB) are found for LoS and sQLoS visibility cases (AB_R-AB_L and AB_R-TO_F), respectively, supporting previous statements.
When comparing these results with the outdoor ones, it is observed that the range of variation for the average system loss is [35.6, 69.3] dB for indoors and [35.8, 72.8] dB for outdoors, with corresponding global averages of 56.5 dB and 56.4 dB, hence, the difference being negligible. These averages are slightly lower than in the Scenario A case (around 2 dB), due to the existence of clear LoS cases in Scenario P. The negligible global difference between indoors and outdoors shows that the clear LoS cases minimise the importance

IV. SYSTEM LOSS DEPENDENCE OF VISIBILITY AND MOBILITY
In order to study the effect of user mobility and path visibility on system loss a classification model is proposed, allowing to characterise the different situations that can be found, and aiming at providing inputs for deriving a method for the choice of the best antennas location for a given usage scenario, to minimise the global system loss.

A. ANALYSIS OF VISIBILITY AND MOBILITY
While visibility accounts for the higher or lower degree of alignment of the antennas together with the influence of body obstruction, mobility is classified according to the degree of relative movement of the body, e.g., arms, rather than the whole movement of a person him/herself.
To properly investigate the relation between the measured system losses and the observed visibility and mobility conditions, one has listed the values of system loss by increasing order, as illustrated in Table 5 for the case of Scenario A outdoors (all results can be found in [2]). Lower values are usually associated with higher visibility and low mobility conditions, e.g., WA_F-TO_F and TO_F-TO_F. Still, there are some exceptions resulting from the specific conditions of each scenario and the influence of the surrounding environment, both indoors and outdoors.
To assess the relationship between visibility and system losses, one has calculated averages and standard deviations for each scenario, by averaging all possible Tx and Rx antenna locations in each scenario, for both indoors and outdoors. The relationship between average system loss and visibility is presented in Fig. 12. It should be noted that not all visibility conditions are observed in all scenarios, as shown in Table 3. The general trend of increasing system loss when ranging from LoS to NLoS is observed, as is the difference among the three scenarios.
On the other hand, mobility is closely related to the standard deviation of system losses, as depicted in Fig. 13. For each scenario, the standard deviation increases with the degree of mobility, as expected.
In order to better understand the influence of visibility and mobility in indoor and outdoor environments, assuming that in a real situation users move in a mixture of environments and scenarios, in which visibility and mobility is not under control, one has calculated averages and standard deviation of system loss in both indoors and outdoors, for all three scenarios, Fig. 14 and Fig. 15.
System losses increase for weaker visibility conditions, both indoors and outdoors. Averages are higher indoors for higher visibility (LoS and sQLoS), but then tend to be opposite (higher for outdoors) when visibility becomes lower (the extreme being in NLoS). This can be explained by the influence of signal reflections in the environment, which are more significant indoors due to the presence of nearby walls and ceiling, namely in low visibility conditions. Standard deviations observe a clear relationship with the degree of mobility, namely outdoors, i.e., increasing standard deviation with increasing mobility, as expected.
On the other hand, it is important to assess the influence of mobility on the average system loss, to understand if it should be considered as a parameter for quantifying link quality. Fig. 16 shows the average system losses in indoors and outdoors as a function of mobility, for all scenarios.   For Low and Medium mobilities, average system losses do increase with mobility, but for High mobility this is no longer the case for outdoors. This behaviour is explained by the influence of some specific configurations in Scenario P, in which, besides the high mobility being observed, Tx and Rx antennas move close to each other during the whole path, i.e., for the four configurations corresponding to the antennas being positioned on the arms and/or the head they are always within the HPBW of each other. Fig. 17 shows the average system losses in indoors and outdoors as a function of mobility, for Scenarios A and D. Now,   the average system losses increase for increasing mobility in all cases, showing that the impact of mobility on system loss depends on the relative movement between the two bodies.

B. THE WEIGHTED SCORE MODEL
To quantify the influence of visibility and mobility on the average system loss, one proposes the combined score S MV , where: • w M ,V : weights for mobility (M ) and visibility (V ) scores; • S M : score for mobility, being the same for all three scenarios (it only depends on the antenna positioning on the body); • S V : score for visibility, obtained as the average scores among the three scenarios, where • p A,D,P : contribution parameters of Scenarios A, D and P for the mobility pattern in a given scenario; • S VA,VD,VP : score for visibility in each of the Scenarios A, D and P. The values of S M and S VA,VD,VP are defined as indicated in Table 6. The rationale, following the previous analysis of results, is to give a high score to a case with lower system loss, taking the analysis according to the data in Table 3.
In order to apply this model, one has to establish values for the weights and the contribution parameters, which strongly depend on the actual application scenario to be taken. Some example scenarios can be considered: • in some sports, such as running in track and field, the athletes tend to be in a situation similar to Scenario P, with a high mobility; • in other sports, such as basketball, the athletes tend to be in a very dynamic mixture of all scenarios (A, D and P), with a very high mobility; • in military training, Scenario P tends to be the closer one, with a mobility that can vary from low to high; • people walking in a street can be approximated by a mixture of all scenarios (A, D and P), with a mobility that can vary from low to high, and even being almost static (people stopping to talk or see a window shop); • in an emergency situation, one can find in Scenarios A and D a good description of some cases, mobility ranging from low to high. From this list, one can see that the values of w M ,V and p A,D,P can really change from one application scenario to another.
The rationale to propose such a simple model is because in real scenarios, like the ones described before, there are so many uncertainties involved in the characterisation of mobility and visibility conditions that no significant added value is introduced by considering a more complex model that, thus, will decrease its significance, in the sense that it will get harder to understand and adapt to different situations.
In what follows, to highlight the impact of the several individual scores, one considered a uniform situation (but the analysis can obviously be done for other values), i.e., w M = w V = 1/2 and p A = p D = p P = 1/3. The values obtained for S MV are presented in Table 7.
The impact of visibility and mobility is very clear, as three distinct regions can be seen in the table, highlighted with different colours: the green region has the highest values of S MV , between 0.6 and 0.8, corresponding to a high visibility and low mobility, in the cases where both antennas are placed    These results show that antenna placements leading to lower system losses are those on the head or on the front of the body, no significant difference being observed between the left and right sides of the body. When averaging these results by considering both left and right sides of the body, and also waist and torso in the front of the body, one obtains the results presented in Table 8, where the values of the average system loss are also shown.
Overall, a good agreement is observed between the theoretical approach for the influence of mobility and visibility and the average of the measured system loss. It should be noted that higher scores, leading to better results, are expected when antennas are placed on the front of the body (waist or torso) or on the head with no distinction among them; still regarding the average system loss, a slightly better situation is obtained with both antennas being positioned on the head (around 3 dB), which may be explained by the existence of a more predominant LoS case.
The relative low value of system loss obtained for the case in which both antennas are placed on the arms (64.0 dB) is due to the fact that in scenario P there is a strong link among the antennas during the whole path for AB_R-AB_L, having a strong influence on the obtained result. If the average is taken only by considering Scenarios A and D, one gets 66.8 dB (roughly 2 dB higher than the ''yellow'' locations and reaching almost 7 dB compared to the best ''green'' one) which is in good agreement with the low S MV score.

V. DISCUSSION ON THE CHOICE OF THE BEST ON-BODY ANTENNA PLACEMENT
In the real-life, people move in a diversity of scenarios that are not easily to characterise. One can mention situations such as people sitting together for a few hours to enjoy a concert in an auditorium or a similar one where people stand up, moving around and dancing, hence, having completely different patterns of movement and body mobility. People can be just walking in a street or on a square crowded of people, or they can be walking together watching the sea. All these situations result in a combination of mobility and visibility scenarios that are not easy to characterise.
In this paper, one proposes to model different visibility situations by considering only three different basic scenarios, i.e., Approach, Departure and Parallel. The modelling of real-life scenarios can be done by taking different values for the weights and the contribution parameters, as mentioned before. Rather than aiming at characterising all situations that can be found in real-life, the goal of this paper is to provide a model to identify the best on-body antenna locations for a given scenario. For the studied scenarios, it is found that a good agreement is observed between the proposed scoring strategy and the average system loss. Still, as already mentioned, further work is needed to properly characterise different real-life situations as a mixture of a finite set of environments.
Globally, it is observed that the best antennas' location is on the head and on the front of the body (waist or torso), but from the scenarios geometry one can expect that similar results are obtained with the antennas on the back.

VI. CONCLUSION
The main objective of this work, addressing body-to-body communications, was to characterise the influence of Tx-Rx antennas placement on the body and body mobility itself on system loss, since they are directly related to link quality, hence, to the data rate that can be obtained.
The analysis of the radio channel, in indoor and outdoor environments and different mobility scenarios, for different on-body antenna configurations, is based on system loss measurements at 2.45 GHz. For each environment, three mobility scenarios were measured, i.e., Approach, Departure and Parallel, corresponding to typical situations of day-to-day people's mobility. The on-body antenna configurations for both Tx and Rx were right and left sides of the head (HE_R/L), front side of the torso (TO_F), front side of the waist (WA_F) and external sides of both arms at the wrist (AB_R/L). The indoor environment was a corridor in a building, while the outdoor one was a part of the square just in front of the previous building.
A classification model to characterise the effect of user mobility and path visibility on system loss is proposed, allowing to characterise the different situations that can be found for the different scenarios being studied and any given position of Tx and Rx antennas.
To quantify the influence of visibility and mobility on the average system loss, a combined score is defined allowing to map the measured value of system loss to the degree of visibility and mobility that depends on the scenario being considered and on the on-body antenna placements. From the obtained results, a good agreement is observed between the proposed classification model and the average measured system loss, with the higher values of combined scores being associated with lower values of systems loss. For the cases under study, average values of system loss are 61.6 dB for the cases of the antennas being positioned only on the front of the body and/or the head, and 64.5 dB if at least one of the antennas is placed on an arm. This work will be further developed in order to have recommendations for the location of antennas and a path loss model for BAN design, accounting for specific real-life use cases beyond the ones being addressed here.
[13] M. Mackowiak  SŁAWOMIR J. AMBROZIAK (Senior Member, IEEE) was born in Poland, in 1982. He received the M.Sc., Ph.D., and D.Sc. degrees in radio communication from the Gdańsk University of Technology (GUT), Poland, in 2008, 2013, and 2020, respectively. Since 2008, he has been with the Department of Radiocommunication Systems and Networks, GUT, from 2008 to 2013, as a Research Assistant, from 2013 to 2020, as an Assistant Professor, and since 2020, as an Associate Professor. He is the author or coauthor of many publications, including book, book chapters, articles, reports, and papers presented during international and domestic conferences. He participated and still participates in several projects related to special application of wireless techniques and three COST Actions (IC1004, CA15104, and CA20120). Currently, his main scope of research is radio channel modeling in body area networks. His research interests include wireless communication and radio wave propagation. He has acted as a Consultant for the Portuguese telecommunications operators and regulator, besides other public and private entities, and has been in the Board of Directors of a Telecommunications Company. He has participated in 32 projects within European frameworks, having coordinated six and taken leadership responsibilities at various levels in many others, besides national ones. He has supervised over 220 M.Sc./Ph.D. students, having edited six books, contribute to European strategic documents, and authored over 500 papers in international and national journals and conferences, for which served also as a reviewer, an editor, and a board member. Internationally, he was a part of 37 Ph.D. juries, 74 research projects and institutions evaluation committees for funding agencies in 12 countries, and the European Commission and COST. He has been the Chairman of Conference of the Technical Program Committee and of the Steering Committee of 25 major conferences, besides other several duties. He was a National Delegate to the COST Domain Committee on ICT. He has launched and served as the Chairman for the IEEE Communications Society Portugal Chapter, besides being involved in several other duties in this society at the global level. He is an Honorary Professor of the Gdańsk University of Technology, Poland. He was a recipient of the 2021 EurAAP Propagation Award ''for leadership in the field of propagation for wireless and mobile communications.'' VOLUME 10, 2022