Wireless Positioning Algorithms And Their Performance Analysis Under NLOS Environment

In addition to provide security service to mobile subscribers, the location of mobile stations can be also applied to network performance enhancement, more effective resource management, system capacity adjustment, and other various value-added services. The mobile positioning normally consists of two steps, the first step provides estimates of location parameters according to different location methods, and the second step determines the MS position using appropriate positioning algorithms, based on the available location parameters. In modern wireless communication environment, especially in urban environment, due to the influence of several adverse factors, such as Non Line of Sight (NLOS) propagation and Geometric Dilution of Precision (GDOP), the performance of all existing position methods is inevitably influenced, and the position estimation of the mobile terminals may be significantly biased as well. Thus, it is difficult for current radio location systems to meet the E-911 location requirement under the NLOS propagation environment. It is therefore the purpose of this thesis to investigate and propose accurate positioning algorithms (the second step) and their performance evaluation, in order to mitigate the NLOS errors.Based on the analysis of NLOS influence to the location algorithms, NLOS error mitigation algorithms in location estimation are investigated. The first algorithm proposed in this thesis is a hybrid location method based Chan and Taylor algorithms, where the Chan algorithm is a closed form non-iterative maximum likelihood (ML) algorithm under the line-of-sight (LOS) environment, but its performance degrades significantly under NLOS environment, and the Taylor algorithm is a Taylor series expansion method depending heavily on the initial values in order to guarantee convergence. The proposed hybrid location method first calculates the initial MS position using Chan algorithm, then calculates the intermediate MS position according Taylor algorithm using the obtained initial results as the its initial values. The final MS position is then the linear combination of the intermediate results based on certain carefully chosen weights. It is shown that the hybrid location method can significantly improve the location accuracy under bad channel environments, and can also overcome the divergence problem of Taylor algorithm.The second algorithm proposed is the Geometry-constrained NLOS mitigation algorithm based on ML-detection. The algorithm first computes the mobile position using arbitrary 3 TOA measurements according to their geometric relation, and then selects the most possible result from all computed results as the final MS estimated position using the ML algorithm. The proposed GML algorithm can achieve optimal estimate while preserving the computational efficiency of the ML detection algorithm and without requesting initial MS positions. It can restrain the NLOS errors effectively without knowing the exact conditional p.d.f.'s of the TOA measurements, resulting excellent positioning performance. The algorithm is also not sensitive to the standard deviation errors.Due to the influence of multipath propagation and power control, the location accuracy of signal strength based techniques is limited. However, because it is easy to implement, cost effective and not necessary to modify the base station and mobile station, it is still widely used in GSM and other location systems. This thesis addresses the problem of improving the location accuracy of GSM positioning system by employing the existing GSM data. Hence, there is no need for any additional hardware, network overlay or infrastructure change. A data fusion model based on Geometric Dilution of Precision (GDOP) weighting for GSM mobile position estimation is proposed. By applying a set of practical GSM system measurement data, it is shown that the location accuracy is improved by 8% compared with the conventional CI+TA when the location error is within 300m. If the measurement data is processed beforehand, the location error will be less than 225 meters in 67% of the time, better than other positioning algorithms. The thesis also investigates the performance of standard positioning algorithms based on received signal strength difference (RSSD) and derives the corresponding CRLB bound, together with the numerical results.Next, the single reflection ring of scatterer model is discussed, which is. suitable for analyzing many kinds of location algorithms. By analyzing the time delay characteristic of this model, a new TOA reconstruction method based on the ring of scatterer model and its improvement is proposed. The proposed method can significantly reduce the NLOS error and improve the location accuracy compared with the conventional method.Considering that the future wireless communications will ensure seamless service provisioning across a multitude of wireless networks, from private to public, from indoor to wide area, and be able to provide an optimum delivery via the most appropriate network available, mobile terminals can communicate with each other cooperatively using high-speed wireless links. It is hoped that the information between MSs can also be used cooperatively to enhance the location accuracy. In this thesis, the cooperative location of wireless sensor network and cellular-controlled short-range communication system is investigated. Firstly, aiming at the enhanced positioning accuracy of wireless sensor network node location, an improved DV-DISTANCE positioning algorithm, which uses a differential error correction scheme for reducing cumulative distance error and node location error accumulated over the multiple hops, is proposed. Then, a cooperative high accuracy positioning algorithm for 4G is proposed together with the simulation analysis.Finally, a general wireless location platform is discussed. As various wireless networks will co-exist in the future, a general platform built upon the existing independent mobile communication infrastructure is necessary. In this thesis, a new general value-added service platform for wireless location, suitable for specialized location service providers, is proposed to overcome the limitations of current individual location system with simplistic service and small coverage. The system architecture of the general platform is presented, and a concept of location middleware layer based on GSM and 3GPP Parlay API standard and other systems, is proposed, together with the possible location services which can be provided.