Application of Kalman filter in GPS

Abstract

We assume that we have a moving user who is needed to keep track of the position by GPS. The velocity of user is unchanged. Therefore, there will have an acceleration error in x, y, z axis (the velocity is changed). GPS receivers are installed in users to update their positions continuously. However, measurement errors and transmission errors affect to the estimation of user’s positions. Measurement position and real position are a little bit different. To reduce the difference, we propose to use Linear Kalman Filter (LKF) and Extended Kalman Filter (EKF) to keep track of moving users. We used C and Matlab programming language to write the codes and simulate them. We obtain the great results. By using Linear Kalman Filter (LKF) and Extended Kalman Filter (EKF), the user’s position is closer to the real position.

Besides, we realize that GPS signal power is very weak so that it cannot go through obstacles. As the results, GPS is useless in indoor environment. Moreover, when indoor users can receive the GPS signal, it’s still useless. The reason is that the GPS errors are rather high (even several hundred meters) for free users. In other words, it depends on the weather and the policy of US Defense Department. In conclusion, GPS is useless in indoor environment. Therefore, people are still lost in huge terminal or supermarket, etc. Indeed, we need to find out a solution to solve this problem. To meet the locating in indoor environment, we establish new locating system based on GPS model. It was called GPS-indoor. We assume that there’s a moving user in indoor environment with constant velocity. Thus, the acceleration noise in x, y, z directions make the velocity changes. Indoor-GPS system uses RFID equipment to guarantee the receivers can always receive the signals from 4 points in space like 4 satellites in GPS. The difference here is Indoor GPS uses RSSI to determine the distance while GPS uses TOA. The moving users update their positions continuously. However, the measurement errors and transmission errors affect so much to the results. The measurement position and true position are different. To reduce the difference, we propose to use Extended Kalman Filter (EKF) to keep track of moving users. We also use Matlab programming language to write the code and simulate it. From the stimulation results, it shows that the error between measurement position and real position is improved. In addition, we developed Kalman Filter algorithm furthermore on hardware based on TMS320C6713 DSP card. To do that, we wrote Kalman Filter algorithm in C language. Then we used CCS (Code Composer Studio) to compile, build, and run the codes on DSP card. Next step, we tried to show the output from DSP card. By using MATLAB and Link for Code Composer Studio, MathWorks tools, CCS IDE and RTDX work together to help us test and analyze the processing algorithms in MATLAB workspace.