Impulse Radio Ultra Wideband Indoor Ranging And Localization Research Based On Low-resolution Quantization

Recently, the demand for accurate indoor localization is blooming as the increasing of indoor activity. However, the indoor localization technology is hard to realize due to the complex indoor environment, such as obstacles, dense multipath and the huge path loss. Furthermore, the low complexity and low power consumption requirement makes the indoor localization more difficult. Compared with the other candidate choic-es, the Impulse Radio Ultra-wideband (IR-UWB) technology has the advantages in pen-etrability, time resolution, multipath distinguishing, interference management, and the feasibility, thus becomes the ideal indoor localization technology. But these merits are bottlenecked by the huge bandwidth of UWB signal. According to the Nyquist theorem, the large bandwidth signal requires the high performance analog-to-digital convertor, which is expensive and power-hungry. As far as I know, the low complexity imple-mentation of IR-UWB localization system remains the main problem. In this thesis, I propose a low-resolution quantization method, losing the resolution of signal amplitude instead of time resolution, which is more critical in high accuracy localization. Based on this tradeoff, I propose several time-of-arrival (TOA) estimation and localization approaches, achieve the best balance of complexity and accuracy.Our research focuses on four aspects:low-resolution quantization IR-UWB sig-nal receive, TOA estimation algorithm, localization method, the realization of ranging system. I conclude the main contribution as follow:In the signal receive aspect; I theoretically analyze the influence of low-resolution quantization using Cramer-Rao bound and compared it with the full-resolution digital receiver and energy detection cases. At first, I introduce the quantization efficiency to describe the influence of quantization on TOA estimation and prove that the2-bit quantization can achieve88%efficiency of full-resolution quantization. Subsequently, the multipath overlapping coefficient is proposed to evaluate the impact of the over-lapping between the direct path and the following multipath. I prove that multipath overlapping is only determined by multipath arrival rate and pulse width. Then, when a priori knowledge of multipath parameters is given, the TOA estimation performance is promoted. Finally, I give the system design advice according to the above analysis:1. decreasing the quantization resolution while increasing the sampling rate to use a more narrow impulse, thus increase the TOA estimation performance by enlarging the equivalent bandwidth and diminishing the multipath overlapping;2. Utilizing multi-ple impulse to form one symbol so as to using the sub-optimum quantizer with more feasibility;3. The10ps sampling time jitter has little impact on system performance.In the TOA estimation method discussion aspect, I design two low complexity al-gorithms while achieving the satisfactory TOA estimation precision. The fist algorithm is stated as follow:Firstly, I model the TOA estimation problem as a two likelihood hypothesis test by determining whether the signal exists. Then locally detection ap-proach is proposed to detect signal in the low signal-to-noise (SNR) area. Subsequently the likelihood test statistics is formed using the quantized signal. By analyzing the sta-tistical feature of test statistics and using the generalized Neyman-Pearson criterion, I decide the TOA between the signal region and the noise region if a test statistic crosses a threshold. The second algorithm is based on the first one. Because the first algorithm only focuses on locally detection in low SNR situation, I proposed a generalized likeli-hood hypothesis test that utilizes the estimated parameter as if it was correct. At first the sub-optimum quantization is adopted to estimate the signal, and then I use the estimat-ed signal to update the quantization parameter to obtain the generalized test statistics and the decision threshold. Note that I use the deflection ratio criterion to analyze the estimator’s performance and the quantization parameters.In the localization research; I firstly derive the performance bound, the Cramer-Rao bound, of several positioning method, such as TOA, Time difference of arrival (TDOA), and angle of arrival (AOA) under the low-resolution quantization. Then a Taylor-series method is adopted to solve the least square positioning problem. Note that our approach is based on the real quantized signal rather than the traditional ranging error based analysis. Finally, I consider the no-line-of-sight (NLOS) ranging error offset method by utilizing the geometry of base stations.In the realization aspect; I design a monobit quantization based TOA ranging sys-tem and test it in different environment. This ranging equipment uses the high speed comparator to receive the signal, the parallelized digital signal processing part output the ranging results. The round time of fly (RTOF) of two nodes is measured so the nodes only need the perfect synchronization. Instead, coarse timing synchronization is only required because the time division dual (TDD) protocol is used. This two-node system support15.625Mbps communication rate and15cm ranging precision. The practical en-vironment test demonstrates the superiority of this ranging system. Our method keeps a satisfactory accuracy even in NLOS and dense multipath cases while remaining the low implementation complexity.