Research On The Identification For Signal Propagation And Robust Positioning Algorithm In WSN

For Wireless Sensor Network (WSN), it becomes key technologies that the identification of NLOS signals propagation and robust localization of unknown nodes. In indoor environment, the power of signals will be weakened by reflection and diffraction of some obstacles. Thus, it produces the NLOS error. For this condition, this thesis has a research on parameter characteristics of signal propagation and proposes two algorithms based on modified expectation maximization (EM) and propagation delay (PD) estimation, respectively. Combining with TDOA positioning algorithm based on Taylor’s series expansion (TSE), the result achieves better performance in NLOS identification and node localization.Focusing at the parameter characteristics of propagating signals, a new application based on modified EM algorithm is applied in NLOS identifications. Because the measurements of signal contain both LOS and NLOS conditions, both approximately following the Gaussian distribution, it could be concluded as Gaussian mixture model (GMM) as a whole for research. Based on GMM, it determines the number of model branches to identify the state of signal propagation. The performance of the positioning accuracy under the NLOS situation is shown to be improved from the result of simulation.A modified method to identify the state of signals based on PD estimation of measurements is proposed in this thesis. The signal PD is acquired by using the theory phase-only-correlator (POC). Then, it is multiplied by the propagation speed of signal to get the estimation of distance measurements, using which, compared with the ones of RSSI positioning model, to identify the strong NLOS measurement. Next, the parameters of distribution is estimated by using EM algorithm and alternating optimization approach to identify the weak NLOS from the remaining measurements after identifying the serious ones. From the simulation, it tests and verifies the higher detection rate and translates into the accuracy of positioning primly.For robust positioning of unknown nodes, a modified TDOA localization algorithm based on Taylor’s Series Expansion (TSE) is proposed. By using selected residual-weighting (SRwgh) method to reducing calculation, it decreases energy consumption in communicating between WSN nodes. It makes the value of estimation with SRwgh as the initial one of iteration to overcome the shortcomings of TSE. When it satisfies the discriminant conditions, the procedure will stop. The proposed algorithm not only improves the positioning accuracy, but also reduces the huge amount of computation.