Indoor Visible Light Localization Method Using TOA/RSS Hybrid Information

Yang Cao; Yuchao Dang; Xiaofeng Peng; Yue Li

doi:10.3788/CJL202148.0106005

Abstract

Objective Compared with wireless local area network(WLAN)/ultra wide band(UWB) and other electromagnetic wave wireless communication technologies, indoor visible light communication technology has the advantages of low cost and an advantageous edge. Research on the indoor localization method based on received signal strength(RSS), time of arrival(TOA), time difference of arrival(TDOA), angle of arrival(AOA), and other classical localization algorithms is the key to realizing localization technology. However, compared with traditional electromagnetic wave wireless communication technology, in the visible light communication environment, the localization methods based on most localization algorithms are not mature, and the improvement in the localization performance is often limited by the singular localization information. In this paper, TOA and RSS data are combined to reduce the effect of nonlinear errors on the localization accuracy in indoor visible light communication environments. Combined with the inertial sensing data at the receiving end, the robustness and low localization delay of the proposed localization method are guaranteed. Additionally, the localization accuracy of the system is further improved.

Methods Four sources are evenly distributed on the ceiling to simulate the indoor environment with a length and a width of 5m and a height of 3m. A channel model of indoor visible light communication is established, which has a direct line of the sight link and a multiple order reflection indirect line of the sight link. Then, the distribution of indoor received optical power is obtained, and the empirical formula of signal strength and distance is established using the mapping relationship between the received optical power and linear distance, between the source and receiver. Moreover, the time stamp record is used to measure the signal transmission time at the receiver. The particle filter based on an unscented Kalman filter is used to combine TOA and RSS data to improve the accuracy of distance estimation. Furthermore, the least-square method is used to estimate the localization coordinates. Finally, based on the inertial sensing data of the receiver, the movement trend is analyzed, and high-precision localization results are obtained.

Results and Discussions Based on the channel model, a localization simulation is conducted. A total of 625 test points are selected in the room, and the localization results are obtained by coordinate estimation. The indoor localization error fluctuates from 1.6 to 3.2cm, and the overall localization error is low lying with a small center and rising edge, exhibiting only a small fluctuation range. First, the simulation parameters are fixed, and the localization performance of the proposed localization method, RSS method based on trilateral localization, and traditional fingerprint localization method are compared. For the proposed localization method, the probability of the localization error at less than 3cm is 98.1% and the average and maximum localization errors are 2.02 and 3.39cm, respectively. For the traditional fingerprint localization method, the probability of the localization error at less than 3cm is 40.8% and the average and maximum localization errors are 3.11 and 6.12cm, respectively. For the RSS method based on trilateral localization, the probability of the localization error at less than 3cm is 1.6% and the average and maximum localization errors are 5.61 and 9.67cm, respectively. Second, the localization performance of 12-, 6-, and 3-W LEDs is compared. Under 12-W transmitting power, the probability of the localization error at less than 3cm is 98.1% and the average and maximum localization errors are 2.02 and 3.39cm, respectively. Under 6-W transmitting power, the probability of the localization error at less than 3cm is 91.2% and the average and maximum localization errors are 2.52 and 3.77cm, respectively. Under 3-W transmitting power, the probability of the localization error at less than 3cm is 42.4% and the average and maximum localization errors are 3.18 and 5.08cm, respectively. Finally, the localization time of the proposed localization method, RSS method based on trilateral localization, and traditional fingerprint localization method are compared. For this, 30 positioning processes of the three localization methods are selected. The RSS method based on trilateral localization exhibits the shortest localization time, while the localization time of the proposed localization method and fingerprint localization method fluctuates by approximately 1s.

Conclusions The simulation results are listed below:

1) Under fixed parameters, the maximum localization error of the proposed method is 44.61% less than that of the traditional fingerprint localization method, and the average localization error is reduced by 35.04%. Compared with the RSS method based on trilateral localization, the maximum and average localization errors of the proposed method are reduced by 64.94% and 63.99%, respectively. Therefore, the localization accuracy of the proposed localization algorithm is better than that of the other two localization methods.

2) The localization performance clearly decreases with a decrease in the LED transmitting power. However, no significant difference is observed between the performance of the proposed localization method at 3?W transmission power and that of fingerprint localization method at 12?W transmission power. Moreover, the proposed localization method shows better performance than the RSS method under 12?W transmission power, thus demonstrating the robustness of the proposed localization method.

3) The localization time of the proposed localization method is stable. Although it is nearly the same as that of the traditional fingerprint localization method, the overall trend is stable and it is only slightly longer than that of the RSS method based on trilateral localization. Thus, the proposed localization method can achieve an improved localization effect only by sacrificing a small amount of time resources.

In conclusion, the overall localization effect of the proposed method is good, and the localization error do not fluctuate significantly, thereby ensuring the robustness of the proposed localization method as well as low localization delay and good localization performance.