The Study Of Mobile Location Algorithms For NLOS (Non-Line-of-Sight) Environment

The location estimation of a moibile station (MS) in the celluar mobile communication network has received considerable attention since the US Federal Communication Commission (FCC) passed the E-911 emergency call rule in 1996. In the following regulation of E-911 loation requirements, the accuracy requirement is 100m for 67% of time and 300m for 95 of time for network-based location systems. Getting the accurate MS location in the cellular network can not only satisfy the requirements of emergency rescue, but also provide many others location-based services, such as location-sensitive billing, fleet and traffic management, cellular network optimization and design, location-based messaging, etc. Location estimation methods for mobile stations (MS) in the cellular system are based on signal characteristic parameter measurements, such as signal strength, time of arrival (TOA), time diffrence of arrival (TDOA) and angle of arrival (AOA), between the mibile station and base station (BS). In these mobile location techniques, the wireless location algorithms based on the time of arrival or the time difference of arrival have gained deeply inveastigations because of it's better location accuracy and simpilicity of actualization. This paper is expanded on the wiless location algorithms based on time-related measurements. In the cellular mobile communication networks, when the line-of-sight (LOS) propagation path between mobile station and base station is blocked, the radio signals will propagate through non-line-of-sight paths by reflections, dispersions and diffractions. Then there will be a positive additional time delay in the TOA measurement, which is called NLOS error. The NLOS error is only related with the radio propagation environments, which can't be eliminated by increasing the TOA/TDOA measurements accuracy of the systems receivers. For the loacation systems based on TOA/TDOA measurements, the NLOS error, which causes large location errors, is the decisive issue of location accuracy. This has led to the development of algorithms that focus on identifying and mitigating the NLOS error. Based on the above analysis, this thesis aims at the research of mobile location algorithms in the NLOS propagation environment. 1. After analysis of the TOA measurements in NLOS environment, the time measumens model of the actual cellular network is established. TOA measurements are corrupted by standard measmrement noise and NLOS error. The standard measmrement noise caused by measuring equipments is always modeled as gaussian variable with zero mean and small variance. The large NLOS error for the different channel environments can be modeled as the frequently used models for delay profiles which are exponential, uniform or delta random variable. 2. Based on the detailed analysis of previous research of mitigating NLOS errors, there are two kinds of location algorithms: (1) NLOS errors are firstly eliminated from the TOA measurements, then, the smoothed data is approximative with the LOS measurements. After that, the mobile position is estimated by the LOS location algorithms. (2) By increasing the robustness of the algorithms for NLOS error, or reducing the weighting factor of the base station affected by NLOS error in the algorithms, location error could be decreased. In practice, the weighting factor of each BS is hard to get, so the first kind of NLOS location algorithms and some LOS position algorithms are investigated in great deal. 3. In NLOS enviroments the existing basic location algorithms are simulated and analyzed. Many relative parameters are examined in these simulations, such as the cell size, the number of base stations taking part in the location service, equipment measurement errors, the different distribution and possibility of NLOS errors, and the MS velocity, etc. The simulation results show that as following: (1) The twice WLS algorithm has accurate location precision in good channel environments. Location errors of this algorithm increase greatly with larger NLOS errors. (2) When MS is quiescent or moveing slowly, WYLIE algorithm using polynomial smoothing and LOS reconstruction can accurately estimate MS location. But, during observation, if the LOS/NLOS propagation environments change as the result of rapid moving, the result of polynomial smoothing and LOS reconstruction will be worse which adversely affects the location precision. Based on above analysis, this paper uses the hypothesis test of sample standard deviation in WYLIE algorithm to identify the LOS/NLOS propagation environments and use the twice WLS algorithm as the location estimator after NLOS mitigation.