Signal Propagation Measurements with Wireless Sensor Nodes

F E D E R Signal Propagation Measurements with Wireless Sensor Nodes Joaquim A. R. Azevedo, Filipe Edgar Santos University of Madeira Campus da Penteada 9000-390 Funchal Portugal July 2007 1. Introduction Several experimental measurements have been done to evaluate the RF signal propagation inside a laboratory and outdoors. The measurements were realized with a portable spectrum analyser with different antennas. It was used the Tmote Sky Sensor Mote in the measurements and Micaz. Most of the measurements were made to verify the performance of the sensor nodes in several environments. As it will be presented, the existence of obstacles and indoor reflections affect severally the link quality. This can influence the topology used for the wireless sensor network. Although both the Tomote sensor node and the Micaz have the same radio chip, the measurements demonstrated that the performances of both systems are different. 2. Radiation pattern of the sensor nodes The Tmote sensor node incorporates an internal inverted-f antenna, which is a wire monopole where the top section is folded down to be parallel with the ground plane [1]. The antenna gain is 3.1 dbi [2] and the radio operates at 2.4 GHz (12,5 cm of wavelength) In datasheet it is referred that the antenna may attain 50 meter indoor and 125 meter outdoor. The sensor node uses a Chipcon CC2420 radio for wireless communications and the maximum output power was set to 0 dbm [3]. The first graph of figure 2.1 depicts the antenna pattern, while the Tmote is mounted horizontally with antennas parallel section aligned to the 0 degree direction. The main null is 24 db below the maximum of the pattern. The second graph depicts the antenna pattern, while the Tmote is mounted vertically with antennas parallel section aligned to the 0 degree direction. The polarization is horizontal. As we can observe, the radiation pattern is not omnidirectional in any plane. Therefore, the received signal of a sensor node depends on the antenna orientation of the receiver. 1

Fig. 2.1 Radiation pattern of the Tmote for horizontal and vertical mounting. The Micaz sensor node incorporates an external monopole antenna of λ/4 and operates at 2.4 GHz [4]. The theoretic radiation pattern of a monopole is equal to the dipole of half a wavelength. However, when it is introduced in the sensor node the radiation pattern is changed. Figure 2.2 presents the measured radiation pattern of the Micaz in an anechoic chamber with the antenna in the vertical [5]. As we can observe, the pattern is far from circular. The main null is 9 db below the maximum of the pattern. 30 330 (dbm) -2 0 0-4 300-6 -8-10 60 270-12 -14 90 240 120 210 180 150 Fig. 2.2 Radiation Pattern of Micaz Mote. 2

3. Measurements of the patterns inside a laboratory The Tmote and Micaz sensor nodes were used to evaluate its performance in indoor and outdoor environments. Considering the theory, it was compared the measured results of signal propagation. The measurement of the received signal strength was realised using the portable spectrum analyser R&S FSH 3. Another way to measure the received signal is the RSSI (Received Signal Strength Indicator) parameter of the Tmote and Micaz sensor nodes. 3.1. Tmote sensor node In order to compare the results obtained with different antennas and to verify the contribution of reflections inside the laboratory, the Tmote sensor node was placed 5.4 meters apart from the measurement equipment (figure 3.1). A spectrum analyser was used to make the measurements. The sensor node and the spectrum analyser antennas were at positions in distance of 40 cm above the floor. In figure 3.1 it is represented the relative positions of the transmitter sensor node and reception antenna in the laboratory, with the antennas parallel section aligned to the 0 degree direction. Figure 3.2 shows the laboratory used for the experiments. receptor transmitter Figure 3.1 Position of the sensor node and of the spectrum analyser. Figure 3.2 Used laboratory. 3

Three antennas were constructed for the frequency of interest to deal with the spectrum analyser: a dipole of half wavelengh dipole, with maximum gain 2.15 dbi and SWR=1.55 (figure 3.3a), a bi-quad antenna with 8.5 of gain and SWR=2.4 (figure 3.3b), and a Yagi antenna with 11.5 dbi of gain and SWR=1.35 (figure 3.3c). Due to the antenna SWR, the dipole antenna has less about 0.2 db in the received power, the bi-quad less 0.8 db and the Yagi less 0.09 db. In reference to the dipole, the Bi-quad has a gain of 6.35 dbd and the Yagi of 9.35 dbd. In order to confirm the antenna gains, it was made some measurements in the exterior to minimize the reflections. The received signal strength varies about ±1 db due to the outside reflections. The difference between the received signal obtained by the Bi-quad and the dipole was of 5.8 dbd and for the Yagi was of 9.3 dbd. The expected value for Bi-quad is (8.5-0.8)-(2.15-0.2)=5.75 dbd and for the Yagi is (11.5-0.09)- (2.15-0.2)=9.46 dbd. As we can notice the results coincide with the measured values very well. a) Dipole b) Bi-quad c) Yagi Figure 3.3 Used antennas. To the different orientations of the sensor node, figure 3.4 presents the results measured with the three antennas. The sensor node is in the horizontal mounting and, therefore, the received antennas were place horizontally to the ground. Comparing with the radiation pattern of figure 2.1 we can observe the influence of the reflections inside the laboratory. In fact, the nulls of the pattern are less pronounced in this picture. Moving the sensor node in a small distance (about half wavelength) towards two different directions, figure 3.5a) and 3.5b) show the corresponding measured values. In the radiation pattern of figure 2.1 (horizontal polarization) exits a maximum around 225 and another one (2 db below) around 135. The measures suggest a maximum of radiation around 135. 4

315 0-35,0-40,0-45,0,0 45 Dipole Bi-quad Yagi,0,0 270,0 90 225 135 180 Figure 3.4 Measured received signal strength for different directions of the sensor node. 270 315 0-35,0-40,0-45,0,0,0,0,0 45 90 Dipole Bi-quad Yagi 270 315 0-35,0-40,0-45,0,0,0,0,0 45 90 225 135 225 135 180 180 Figure 3.5 Dependence of the measured received signal strength with the position. The fluctuation of the received RF signal strength for each antenna is better perceived in figure 3.6 for three positions half a wavelength apart. The minimum values are more affected by laboratory reflections. In small distances the signal have varied several db. Considering the average of the measured values, the difference between the mean received power of the Bi-quad antenna values and the dipole antenna values is of 3.6 db (standard deviation of 0.8 db) and the difference between the mean received power of the Yagi antenna values and the dipole antenna is of 6.7 db (standard deviation of 2.3 db). Since the theoretical expected difference is 5.75 db and 9.46 db, respectively, the means are 2.15 db and 2.76 below these values. To understand these results, we must take into account that the dipole antenna is omnidirectional, whilst the bi-quad and the Yagi are directional. Therefore, the dipole can receive more energy from behind reflections than the other antennas. 5

a) Dipole -35,0-40,0 0 45 90 135 180 225 270 315-45,0 (dbm),0,0,0,0 b) Bi-quad -35,0-40,0 0 45 90 135 180 225 270 315 (dbm) -45,0,0,0,0,0 c) Yagi -35,0-40,0 0 45 90 135 180 225 270 315-45,0 (dbm),0,0,0,0 Figure 3.6 Variation of the received signal in small in nearby distances. Let us see when the received antennas are in the vertical instead of on horizontal and maintaining the sensor node in the horizontal polarization. Figure 3.7 presents the results for a rotation of the sensor node. The received signal strength gives, on average, a value of 10.5 db lower using the dipole (standard deviation of 3.8 db) and of 9.4 using the Yagi antenna (standard deviation of 4.9 db). Due to the reflections, the receiver signal strength varies reasonably. 6

-30 0 45 90 135 180 225 270 315-35 -40-45 Dipole - vertical Dipole - horizontal Yagi - vertical Yagi - horizontal Figure 3.7 Comparison of the received signal strength for vertical and horizontal polarizations. 3.2. Micaz sensor node Some measurements were also made with the Micaz sensor node, in order to evaluate the radiation pattern inside the laboratory. The distance to the reception equipment and the distance to the ground is similar to the Tmote sensor node measurements. The dipole antenna and Yagi antennas were used in the measurements with the spectrum analyser. Figure 3.8 shows the results for the antennas in the vertical position, taking into account the vertical polarization of the monopole antenna of Micaz. The zero degrees corresponds to the sensor side where is the monopole. Considering the average of the measured values, the difference between the mean received power of the Yagi antenna and the dipole antenna is of 9.5 db (standard deviation of 1.2 db). The theoretical expected value is of 9.46 dbd, which coincides with the measured values very well. Comparing the radiation pattern with figure 2.2, we can observe that with the reflections the pattern is more circular. 315 0 dbm 45 Dipole Yagi 270-75 -80 90 225 135 180 Figure 3.8 Measured received signal strength for different orientations of the sensor node. 7

To make a comparison, the received signal strength indicator (RSSI) of the Micaz was also considered. The Micaz used to receive the signal and connected to the computer was placed in the same position of the dipole connected to the spectrum analyser. The results are represented in figure 3.9. 315 0 dbm 45 Stectrum Analyser RSSI -75 270-80 90 225 135 180 Figure 3.9 Comparison between the received signal strength obtained from the spectrum analyser and RSSI. 3.3 RSSI and received signal Most of the measurements considered in this work for the signal strength in the reception were obtained using the spectrum analyser. Another way to measure the received signal could be the RSSI (Received Signal Strength Indicator) parameter of the Tmote and Micaz sensor nodes. However, taking into account the datasheet of the radio component CC2420 used in these sensor nodes [3], there exists an accuracy error of ±6 db in the RSSI readings. Therefore, we expect that the readings from the spectrum analyser should be more accurate for reading the real signal on the received antenna position. Furthermore, the RSSI readings have a difference of ±3 db in linearity. Some comparisons were made inside the laboratory to evaluate the differences between the RSSI and direct measurements and between sensor nodes. Considering Tmote sensor nodes, it was showed that the transmitted power is almost the same for the various sensor nodes. To get this conclusion, several sensor nodes were considered as transmitters and the signal at reception was obtained using the spectrum analyser. When the RSSI parameter of the sensor nodes was used, it was found out variations between RSSI readings in the same position. For three Tmote sensor nodes were made measurements in four different locations. The three sensor nodes presented different RSSI readings for the same position. The values have deferred in ±4 db. For the previous positions, the measurements made with the spectrum analyser gave values that can be of 9 db lower then with RSSI readings. The spectrum analyser readings have, on average, a value of 4 db compared with the RSSI readings with a standard deviation of about 4 db. Comparing the measurements obtained by Tmote sensor nodes with Micaz sensor nodes it gave a mean difference between the RSSI Micaz readings of 14 db below the RSSI readings of Tmote. Using the spectrum analyser the difference in the received signal is about 11 db below for Micaz comparing with Tmote. 8

Other works have reported that the RSSI obtained from Micaz sensor node did not report the actual signal strength [6]. 3.4 Influence of the distance Another set of measurements was done to obtain the variation of the received signal strength with the distance to the Tmote sensor node. The maximum radiation direction of the node antenna was considered. The sensor node and the spectrum analyser antenna were at positions in distance of 80 cm above the floor. Figure 3.10 presents the measured signal using the dipole antenna for several distances from the sensor node, defined by the continuous line. The distance between measured points is 3 cm. The signal has a decaying behaviour in the distance to the sensor and a great variation due to the influence of the reflections in the walls, floor and ceiling. In distances of 3 cm the signal can change around 8 db. For greater distances, the signal can change 15 db in small distances. A function for the decaying of the signal can be obtained from the model of the path loss [7], d PL ( d) = PL ( d + n + X σ d 0 ) 10 log 10 (3.1) 0 where n is the path loss exponent and indicates the rate at which the signal attenuates with the distance (n=2 for free space). P L (d 0 ) is the path loss at a known reference distance d 0 which is in the far field of the transmitting antenna (typically 1 km for large urban mobile systems, 100 m for microcell systems, and 1 m for indoor systems) and X σ denotes a zero mean Gaussian random variable (in db) with standard deviation σ, and reflects the variation in average received power. From the measurements made we can obtain an estimative for the n parameter, using the average of results calculated from the formula, P n = ( d) P ( d L 10log 10 L d d 0 0 ) (3.2) where d 0 =1 m. The result is n=2.8 for the path loss exponent. This result is in consonance with those presented in literature. Using this value in (3.1) the result for the path loss is the represented by the dashed line of figure 3.10. The standard deviation for the difference between the measured results and this curve is 4.6 db. Based on equation (3.1) when the receiver measure a value P L (d), the estimated distance to the transmitter is PL ( d0 ) PL ( d ) d d 0 10 10n = (3.3) Since the sensivity of the sensor nodes is of 94 dbm, it will be reached at about 120 meters. This can be understood if there continues to exist a line of sight between the two antennas. However, even in this case, considering that the signal can fluctuates around ±10 db, the reception sensor node can lose the signal in about 50 meters. 9

-20-30 -40 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5 5.5 Distance (m) Figure 3.10 Variation of the received signal with the distance to the sensor node. For a comparison, figure 3.11 presents some positions of the sensor node and measurements realized with dipole and bi-quad antennas. Once again, the mean received power difference between results of the bi-quad antenna and of the dipole is of 3.6 db. -20 0,25 0,5 0,75 1 1,25 1,5 1,75 2 2,25 2,5 2,75 3 3,25 3,5 3,75 4 4,25 4,5 4,75 5-25 -30-35 -40-45 Dipole Bi-quad Distance (m) Figure 3.11 Comparison of the variation of the received signal with the distance to the sensor node. 3.5. Influence of the height to the ground The variation of the received signal strength with the height of the transmitter antenna was analysed, considering the receptor dipole antenna at 93 cm above the floor and 5.4 m from the transmitter. Then the Tmote sensor node was varied from 2 cm till 236 cm and the results were registered, and represented in figure 3.12. Once again, it is clear the effect of the reflections. We can also observe a great fluctuation of the signal. This is due to the radiation pattern variation with the distance to the ground for small heights and also the influence of the ceiling for higher distance to the ground. In fact, at the antenna positions of the experiment, the theoretical nulls in the pattern are about 0.37 cm apart in the height variable for a perfect conducting ground. To confirm the ground influence, for a reflection in a perfect ground we have, βh1h F = 2 sin d 2 (3.4) 10

where β=2π/λ, λ is the wavelength, h 1 is the height of the transmitter antenna, h 1 is the height of the reception antenna, and d is the distance between antennas. For comparison, the factor defined by the previous expression is depicted in figure 3.12 by the dashed line, where the maximums were moved to 45 db to permit the comparison. -40-45 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 1.1 1.2 1.3 1.4 1.5 1.6 1.7 1.8 1.9 2 2.1 2.2 2.3 2.4 Height (m) Figure 3.12 Variation of the received signal with the height of the sensor node. Another set of measurements was made with the receptor antenna at 100 cm above the floor and also at a position 5.4 m from the transmitter. The Tmote sensor node was varied from 10 cm to 170 cm in height. Figure 3.13 depicts the results. From the theory, the null in the pattern obtained by varying the height is about 0.35 cm apart, which can be verified by the figure. 10 20 30 40 50 60 70 80 90 100 110 120 130 140 150 160 170-40 -45 Height (cm) Figure 3.13 Variation of the received signal with the height of the sensor node for a different position of the received antenna. Varying both antennas (reception and emission) related to the floor from 2 cm till 26.5 cm, the result is the one represented in figure 3.14. The dashed line corresponds to the tendency line of the measures. At 25 cm the average of the received signal is about 10 db higher than 2 cm. 11

-40,0 0 5 10 15 20 25 30-45,0,0,0,0,0,0 Height (cm) Figure 3.14 Variation of the received signal with the height of the sensor node and received antenna. 4. Measurements on the distance in a corridor For greater indoor distances, a set of measurements was made in a corridor with 2.4 m width and about 40 m long. Figure 4.1 shows the measured received signal using the dipole antenna for several distances from the Tmote sensor node (continuous line). The distance between measured points is 30 cm. The sensor node is located 11.15 m from the beginning of the corridor. The sensor node and the measurement antenna were 40 cm above the floor. The conclusions for indoor propagation presented previously can be verified for the signal fluctuation but the signal has a lower decaying compared with the obtained inside the laboratory. Considering the reference at 1.5 m, the application of the path loss model gives a value for the path loss exponent of n=1.9. This value is near the free space propagation. We should taking into account that the corridor may have some waveguide characteristics. Substituting this parameter in (3.1), the result is represented by the dashed line of figure 4.1. The standard deviation for the difference between the measured results and this curve is 4.7 db. -20 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29-30 -40-80 Distance (m) Figure 4.1 Variation of the received signal with the distance to the sensor node. 12

Figure 4.2 depicts another set of measurements in the same place but with antennas one meter above the floor. The mean difference between the two results is not significantly. -25-35 -45-75 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 Distance (m) 40 cm 100 cm Figure 4.2 Comparison of the variation of the received signal with the distance to the sensor node for receiving antennas at 40 cm and 1 m. In this test it is intended to compare the performance between of Tmote and Micaz sensor nodes. Considering the Micaz at positions 1 meter above the ground, in the referred corridor and for the same positions in distance, figure 4.3 shows the results by the continuous line. The mean of results obtained from Micaz sensor node is 13 db lower than the mean of results determined from Tmote sensor node. The dashed lines correspond to the curves obtained using (3.1) with the path loss calculated before (n=1.9). The difference between the two curves is 14 db below for the Micaz. Once again, the results demonstrated a difference of about 13 db between the signals of the two types of sensor nodes. -20-30 -40-80 -90 0 2 4 6 8 10 12 14 16 18 20 22 Distance (m) Tmote Micaz Figure 4.3 Comparison of the variation of the received signal with the distance to the sensor node for Micaz and Tmote. 13

5. Existence of obstacles in the propagation path All the tests made till know evolved line of sight between the transmitter and receptor. However, in a sensor network it is expected communication between two sensor nodes even when they cannot see each other. The existence of obstacles in the propagation path affects drastically the communication link. As a simple example, for the positions of figure 3.1 it was used a Tmote as transmitter and the spectrum analyser was placed 5.4 m apart to measure the received signal. The antennas were at 40 m above the ground. For several positions between the two antennas it was placed a metal plate with 1000 400 1 mm in the transversal section. Without the obstacle, the received signal is 49 dbm. With the obstacles, figure 5.1 shows the results for several distances of the plate to the receptor antenna. In central positions of the plate the received signal is less affected than for positions around the antennas. Although without line of sight, a lot of signal reaches the receptor antenna due to the reflections. When the plate approximates to the transmitter or receptor, more reflections are cancelled. The lowest peak is around 16 db below the unobstructed propagation value. 0 0,5 1 1,5 2 2,5 3 3,5 4 4,5 5 5,5-45 -75 Distance (m) Figure 5.1 Variation of the received signal due to a metal plate between the sensor node and the receptor. Another set of measurements was made to evaluate the effect of walls between the transmitter and receptor. Figure 5.2 shows the test bed. The used laboratories have 8.8 m by 5.7 m and contain office equipment, which will cause fading in the reception. The referred situations 1, 2 and 3 correspond to different reception positions for a Tmote sensor node placed near the wall on the horizontal plane, as it is illustrated in the figure. The sensor node is 1.2 m above the ground and the receiver is 0.75 m from the ground. The situation 1.1 corresponds to the position of the sensor node referred in situation 1 but with measurements made in another laboratory. Figure 5.3 depicts the results. The higher curves correspond to the measurements inside the laboratory were it is the sensor node. The mean difference between the results of situation 1.1 and situation 1 is 21 db (standard deviation of 7 db). For situation 1, the mean difference related to the free space propagation is 10 db (standard deviation of 4.5 db). For situation 1.1, where a wall exits between the transmitter and the receptor, the mean difference related to the free space propagation is 23 db (standard deviation of 2.5 db). Therefore, the difference of results is 13 db. 14

Situation 1 Situation 2 Situation 3 3 2 5 7 3 2 5 7 3 2 5 7 1 4 1 4 1 4 6 6 6 Situation 1.1 3 2 5 1 4 7 6 Figure 5.2 Positions of the receptor and sensor node. 1 2 3 4 5 6 7-75 -80-85 -90 Situation 1 Situation 2 Situation 3 Situation 1.1-95 Figure 5.3 Results for the positions of figure 5.2. Let us consider a different orientation of the Tmote antenna. In situation 4 (figure 5.4) the sensor node is placed vertically on the wall but with the antenna in the horizontal. Situation 5 has the sensor node placed on the wall with the antenna in the vertical. Situations 1.4 and 1.5 are similar to these ones but the measurements are made in another laboratory. The results are represented in figure 5.5. The mean difference between the results of situation 1.4 and situation 4 is 17 db (standard deviation of 6 db) and the mean difference between the results of situation 1.5 and situation 5 is 20 db (standard deviation of 4 db). For situation 4, the mean difference related to the free space propagation is 7 db (standard deviation of 5 db) and for situation 5 the mean difference is 2.5 db (standard deviation of 5 db). For situation 1.4 the mean difference related to the free space propagation is 16.5 db 15

(standard deviation of 3.5 db) and for situation 1.5 the mean difference is 15.5 db (standard deviation of 1.5 db). Therefore, the difference of results is 9.5 db for the first case and 13 db for the second one. Situation 4 Situation 5 3 2 5 7 3 2 5 7 1 4 1 4 6 6 Figure 5.4 Point positions for the receptor and sensor node. 1 2 3 4 5 6 7-40 Situation 4-80 -90 Situation 5 Situation 1.4 Situation 1.5-100 Figure 5.5 Received signal with and without a wall between sensor node and measurement equipment. The previous results suggest attenuations introduced by the walls around 12 db. However, the received signal strength depends not only of the signal across the wall but also the signal diffracted around the windows and doors. 6. Outdoors experiments One of the objectives of the Foresmac project is to work in the forest environment. In this sense, after the work realised indoors let us make some measurements outdoors to get more parameters for the sensors deployment. The idea is to obtain an environment with characteristics near the free space propagation. For these experiments it was necessary to create measurement facilities appropriated for the objectives of the work. The flat roof of the University was used in order to minimise the reflections. The transmitter and the receptor antennas were placed 5 m above the ground and the distance between antennas was varied from 1 m to 8 m (figure 6.1). 16

Fig 6.1 Outdoors measurement facilities. The sensor nodes were placed in a hood connecting rod of 5 m long. To control the direction of the antenna a small motor controlled by radio was used (figure 6.2). To connect the reception antenna to the spectrum analyser it was necessary to get a coaxial cable with reduced attenuation loss. The cable length has 10 m long. The usual employed cable of 50 Ω, the RG58, has attenuation of 1.06 db/m for 2.4 GHz. With less attenuation it was utilised the coaxial cable RG213/U with 0.5 db/m for 2.4 GHz (5 db in 10 m). Tests realized with a signal generator have showed that the attenuation introduced by this cable was 4.7 db, a value that was considered in the measurements. Fig. 6.2 System to control the sensor node antenna orientation. 3.1. Tmote sensor node The signal of a Tmote sensor node was measured in the exterior using the dipole and Yagi antennas for several distances. Both systems were placed at 5 m above the ground. The sensor node has horizontal polarization and the maximum radiation was used. In reference to the 17

radiation pattern of the Tmote it was measured a difference of 15 db between the minimum and the maximum radiation. The continuous lines of figure 6.3 represent the received signal strength. Applying the path loss model, from (3.1) we obtain a value for the path loss exponent about n=2.1 for both antennas. The path loss parameter of the Yagi antenna was calculated using the distance from the sensor node until the end of the antenna and not to the excited element (difference of 25 cm). If the distance is considered till the excitation element of the Yagi, the path loss exponent would be n=2.4, which is not an expected result for the free space conditions and did not fit the measurements. As it was verified, the obtained results suggest a propagation factor near the free space conditions. Using the measurements, and taking into account that the dipole gain is around 1,9 dbi, The measured mean of the Tmote gain is 1,4 dbi (standard deviation of 0.6 db). If the free space propagation loss is represented including the dipole gain and Tmote gain the result is the one represented by the dashed lines of figure 6.3. The standard deviation for the difference between the measured results and these curves is 0.6 db with a maximum difference around ±1 db for the given distances. For comparison, the indoor measurements gave a standard deviation of 4.6 db and a maximum difference around ±10 db. The mean difference between the Yagi and dipole results is of 8.8 dbd with standard deviation of 0.8 db (excited element in the same position). The theoretical result is of 9.46 dbd. The importance of the Yagi antenna is to extend the limit of the spectrum analyser measurements in 9 db when compared with the dipole antenna. From figure 6.3 we can also observe that the fluctuation around the tendency curve increases for higher distances from the sensor, reflecting the influence of the ground. -20-25 -30-35 -40-45 1 2 3 4 5 6 7 8 Measured w ith dipole Free space+gains Measured w ith Yagi Distance (m) Figure 6.3 Variation of the received signal with the distance with horizontal polarization for the Tmote. The Tmote sensor node was positioned with the antenna in the vertical. From the measurements, the antenna gain for this polarization is around -6.4 dbi. Figure 6.4 shows the received signal strength for several distances between from the transmitter and the free space propagation including the antenna gains. Once again, the signal follows the free space 18

propagation curve and the influence of the ground is more obvious for higher distances. Comparing with the results of figure 6.3, the received signal has a mean difference of 7.5 db with 0.8 db of standard deviation. Thus, the sensor node has a better reception signal for the horizontal position in an environment with minimal reflections. -40 1 2 3 4 5 6 7 8-45 Measured w ith dipole Free space+gains Distance (m) Figure 6.4 Variation of the received signal with the distance with vertical polarization for the Tmote. 3.2. Micaz sensor node As realized for the Tmote, the received signal strength from a Micaz sensor node was measured outside using the dipole and Yagi antennas for several distances from the transmitter. Figure 6.4 presents the results through the continuous lines. Applying the path loss model to the measurements of the dipole antenna, the path loss exponent is n=2.0. The standard deviation for the difference between the measurements and these curves is 0.8 db with a maximum difference around ±1.5 db for the considered distances. The mean difference between the Yagi and dipole results is of 9.1 dbd with standard deviation of 1.4 db (the theoretical value is 9.46 dbd). Comparing with the Tmote sensor node, Micaz has a mean received signal that is 13 db below the signal received of Tmote. This result was also obtained in previous tests. From the measurements, the suggested gain of Micaz is -11.8 dbi (standard deviation of 0.8 db). The gain difference between Tmote and Micaz is 13.1 db. Other tests to minimize the reflection on the ground gave similar conclusions for the Tmote antenna gain. 19

-30-35 -40-45 -75 1 2 3 4 5 6 7 8 Measured w ith dipole Free space+gains Measured w ith Yagi Distance (m) Figure 6.5 Variation of the received signal with the distance with horizontal polarization for the Micaz. If the Micaz sensor node is positioned with the antenna in the vertical, the received signal strength obtained is the one of figure 6.6. Comparing with the horizontal polarization, the received signal has almost the same amplitude. -45 1 2 3 4 5 6 7 8 Vertical polarization Horizontal polarization -75 Distance (m) Figure 6.6 Received signal of Micaz for vertical and horizontal polarizations. 3.3. Mica2 sensor node The Tmote and Micaz sensor nodes operate at 2.4 GHz whilst the Mica2 operate at 900 MHz band. This mote has a monopole antenna and Micaz. To have an approximation for the gain some measurements was realized in outdoor. To use the spectrum analyser, a half wavelength dipole was constructed. For a SWR of 1.35 and antenna gain is about 2.1 db. The figure 6.7 20

shows the received signal for vertical and horizontal polarizations. From the measurements the Mica2 gain is -8 dbi for vertical polarization. The horizontal polarization suffered more the environment influence. Received Signal Strenght (dbm) -30-35 -40-45 1 2 3 4 5 6 7 8 9 Vertical polarization Free space+gains Horizontal polarization Distance (m) Figure 6.7 Received signal of Mica2 for vertical and horizontal polarizations. 3.4. Influence of the height to the ground The variation of the received signal strength with the height to the ground was analysed, considering the Tmote at 5.4 m from the dipole. The values were obtained moving both antennas from 0.2 cm to 4.75 m, in steps of 5 cm. The results are presented in figure 6.8. As we can observe, the received signal has a great variation due to the ground reflection. For higher distances from the ground, the signal can varies around the average of ±2 db. 0,2 0,4 0,6 0,8 1,0 1,2 1,4 1,6 1,8 2,0 2,2 2,4 2,6 2,8 3,0 3,2 3,4 3,6 3,8 4,0 4,2 4,4 4,6-44 -46-48 -52-54 -56-58 -62 Eight (m) Figure 6.8 Variation of the received signal with the height of the sensor node. 21

Considering a typical permeability (ε r =5) and conductivity (σ=5 01-3 material to calculate the coefficient of reflection on the ground, F = 1+ cos( ψ ) Γ e h 2βh1h2 j d S/m) of the brick sen( ψ ) ε r j Γ h = sen( ψ ) + ε r j 2π β = λ h1 + h2 ψ = arctan d σ ωε σ ωε 0 0 cos cos 2 2 ( ψ ) ( ψ ) (3.5) with λ the wavelength, and h 1 and h 2 the height of emitter and receiver antennas, respectively, the theoretical curve for the received signal strength is shown in figure 6.9 (continuous line). The cos(ψ) term in F represents the radiation pattern of the reception antenna. For this calculation it was taken into account the radiation pattern of the dipole antenna and the previous results. From the curve, we can observe other influences in the received signal, such as the influence of the measure system. -42-44 -46-48 -52-54 -56-58 -62-64 0.5 1 1.5 2 2.5 3 3.5 4 4.5 Height (m) Figure 6.9 Comparison between the theoretical variation and the received signal with the height. 3.5. Different polarizations for the Tmote It was analysed the radiation of the Tmote sensor node for different polarizations and orientations. Figure 6.10 shows the three main orientations of the sensor node. 22

0 0 0 Horizontal mounting and horizontal pola rization Vertical mounting and horizontal pola rization Vertical mounting and vertical pola rization Figure 6.10 Different positions of the sensor node. The measures were made with the Tmote at 5 m above the ground. The dipole antenna used with the spectrum analyser was 4 m of distance and at the same height of the sensor node. Figure 6.11 presents the results. It was noticed some influence of the reflections in the lower values of the pattern. As we can see, the best results are obtained with the sensor node in horizontal mounting and horizontal polarization. As it was observed indoor, the maximum radiation is around 135. An approximation for the outdoor radiation pattern is presented in figure 6.11, for the horizontal mounting and horizontal polarization. The antennas were placed 5.4 m apart and 5 m above the ground. We can compare this graph with the one of figure 2.1. The surrounding environment is noticed in the results. 0 45 90 135 180 225 270 315-75 Vertical mounting and horizontal polarization Vertical polarization Horizontal mounting and polarization Figure 6.11 Results for the different orientations of the Tmote sensor node. 23

300 0 340 350 330 320 310 10 20 30 40 50 60 290 280-75 -80-85 70 80 270-90 90 260 100 250 110 240 120 230 220 210 200 190 180 140 150 160 170 130 7. References [1] Moteiv Corporation, 2006. Moteiv. "Tmote Sky: Ultra Low Power IEEE 802.15.4 Compliant Wireless Sensor Module." 2006. Available from http://www.moteiv.com/products/docs/tmote-sky-datasheet.pdf. [2] Raman, B., Chebrolu, K., Madabhushi, N., Go, D. Y., Valiveti, P. K.k and Jain, D., Implications of link range and (In)stability on sensor network architecture, Proceedings of the 12th annual international conference on Mobile computing and networking, Los Angeles, CA, USA, pp. 65-72, 2006. [3] CC2420 Datasheet, Chipcon. Available from http://www.chipcon.com/files/cc2420_data_sheet_1_3.pdf. [4] Crossbow Technology Inc. "MICAz Wireless Measurement System." 2005. Available from http://www.xbow.com/products/product_pdf_files/wireless_pdf/micaz_datasheet.pdf. [5] Tan, E. B., Lim, J. G., Seah, W. K., and Rao, S. V., On the Practical Issues in Hop Localization of Sensors in a Multihop Network, Vehicular Technology Conference, VTC 2006-Spring. IEEE63rd, pp. 358-362, 2006. [6] Scott, T., Wu, K, and Hoffman, D., Radio propagation patterns in wireless sensor networks: new experimental results, Proceeding of the 2006 International Conference on Communications and Mobile Computing, Vancouver, Canada, pp. 857-862, July 2006. [7] Andersen, J. B., Rappaport, T. S., Yoshida, S., Propagation Measurements and Models for Wireless Communications Channels, IEEE Communications Magazine, vol. 33, pp. 42-49, 1995. 24