Excluding Beacons From Position Calculation

Excluding unreliable beacons from the position calculation is a very simple approach, that comes with minimal costs considering the power consumption and the software complexity. It was considered to exclude up to two beacons based on:

1. the highest standard deviation 2. the weakest signal strength

Excluding Beacons Based on the Standard Deviation

It was decided to try to exclude data based on the standard deviation, because the fluctuation of measured RSSI significantly increases if the beacon is in NLOS or in an unstable environment.

The position was estimated in all the 10 measuring points marked in Fig. 5.3. The results for the control points only can be found in Appendix C. The position was calculated at each point 99 times.

Fig. 6.2 illustrates that excluding one or two beacons based on the standard deviation provides higher accuracy within MOD1. The results are not very convincing within MOD3. This could be caused by very unstable environment, where more than data from two beacons was significantly violated. The accuracy is still increased in the lower part of the plot. It can be said that excluding beacons based on the standard deviation of RSSIs can bring higher accuracy, but this increase is not very significant. The position estimation accuracy is worse in Fig. 6.2 than in Fig. C.1, but the overall assumption is the same

CDF of two models with up to two beacons excluded (all points)

MOD1: all beacons

Figure 6.2.: CDF of MOD1 and MOD3 with up to two beacons excluded from position calculation based on the highest SD of RSSI. Measurements in all the 10 points.

Improvement in the position estimation accuracy, while excluding beacons based on

6. Results

the standard deviation compared to using all beacons is summarized in Table 6.1. The measurements from all the ten points were used in this comparison. The negative value means the exclusion brought worse accuracy than using all the six beacons. The results in Table 6.1 shows that excluding beacons based on the highest standard deviation is not reliable approach. The mean error was improved only in MOD1 by 2%, while excluding two beacons with the highest standard deviation.

µ 50% 95%

(%) (%) (%) MOD1: -1 -7 30 -86

MOD1: -2 4 2 -10

MOD3: -1 -2 -26 40 MOD3: -2 -1 -2 -8

Table 6.1.: The effect of excluding beacons from position calculation based on the highest SD relative to using all the beacons

Excluding Beacons Based on the Signal Strength

Similar measurements as in Section 6.1.2 were repeated with up to two beacons with the weakest signal excluded.

Fig. 6.3 depicts CDF of these measurements in all the 10 points. CDF of only the control points only can be found in Fig. C.2. The accuracy in position estimation is significantly improved in both MOD1 and MOD3 by excluding one or two beacons with the weakest signal. In Fig. C.2 the accuracy of MOD3 is improved up to 60% of all the cases. The drop in the rest of the characteristics can be explained by unstable environment, where the data from more than two beacons was violated.

The overall accuracy in Fig. 6.3 is worse than in Fig. C.2, but it supports the previously described assumption.

0 1 2 3 4 5 6 7 8 9 10

CDF of two models, up to two beacons excluded (all points)

MOD1: all beacons

Figure 6.3.: CDF of MOD1 and MOD3 with up to two beacons with the weakest signal excluded from the position calculation. Measurements in all the 10 points.

The comparison of the achieved results relative to usage of all the beacons is summa-rized in Table 6.2. The measurements from all the 10 points were used in this comparison.

The accuracy is significantly improved for all the configurations. The results are much better while excluding two beacons in MOD1. The number of the excluded beacons does not seem to be as important in MOD3.

Excluding beacons based on their signal strength proved to be reliable approach to improve positioning accuracy.

Table 6.2.: The effect of excluding the beacons with the weakest signal from the position calculation relative to using all the beacons

6. Results

Comparison of the Two Exclusion Approaches

Fig. 6.2 and Fig. 6.3 show that the position estimation accuracy can be significantly improved by excluding some data from the position calculation. Excluding data based on the signal strength proved to be a better approach than excluding the data based on the standard deviation. It could be caused by a too short filtering window, which is not able to capture a stronger signal fluctuation.

A comparison for the measurements in only the control points can be found in Ap-pendix C.

The three best results from the two approaches for MOD1 are compared in Fig. 6.4.

The accuracy of all the measurements is lower than considering only the five control points. A stable improvement of the accuracy was not observed, while excluding the data based on the standard deviation. The accuracy improved by 2.40 m in 50% of all the cases, while excluding two beacons based on the weakest signal strength.

0 1 2 3 4 5 6 7 8 9 10

Comparison of the three approaches to position calculation (all points)

MOD1: all beacons

Figure 6.4.: Comparison of the three different approaches to the position estimation.

Measurements in all the 10 points.

These results show that the beacons with a stronger signal have a higher reliability and that it is beneficial to exclude some of the beacons from the position calculation if there is more of them available.