دانلود رایگان مقاله لاتین فرآیند رگرسیون گوسی از سایت الزویر
عنوان فارسی مقاله:
فرآیند رگرسیون گوسی برای محلی سازی بر اساس اثر انگشت
عنوان انگلیسی مقاله:
Gaussian Process Regression for Fingerprinting based Localization
سال انتشار : 2016
برای دانلود رایگان مقاله فرآیند رگرسیون گوسی اینجا کلیک نمایید.
مقدمه انگلیسی مقاله:
1. Introduction
The received signal strength (RSS) based mobile user localization method has recently attracted significant attention. Perhaps this happened because RSS measurements from the wi-fi access points in indoor scenarios provide a cost-effective positioning system. It does not require any additional hardware unlike time of arrival (TOA), time-difference of arrival (TDOA) and angle of arrival (AOA). The time based localization techniques are also limited by the fact that it requires highly precise synchronization. On the other hand, RSS based localization techniques suffer from the harsh wireless channel such as multipath fading, and nonhomogeneous environment. Hence, it is of sufficient interest to develop the robust fingerprint method which is relatively stable during different days with high localization accuracy. Several localization algorithms in literature are based on two steps procedures. In the first step, inter-nodal range is estimated with learning of radio propagation model. Subsequently, these estimated ranges are further utilized in positioning the user. In doing so, large range error propagates into positioning phase. In contrast, fingerprinting based localization methods provide higher accuracy at the expense of extensive training. It may be noted that training is required even for learning radio propagation model to some extent. In this paper, sparse RSS data are collected from the given area of interest. Subsequently, Gaussian process regression (GPR) is employed to build the posterior mean and variance at each of the locations. These predicted variances are further utilized for localization during test phase. The motivation for using Gaussian process in this work stems from the fact that it not only predicts the RSS mean but also infers the variance at each location. The main contributions of the paper are enumerated herein. 1. We present the textbook derivation of Cramér-Rao Lower Bound (CRLB) on estimation error of kernel function hyper-parameters in the context of GPR framework using basic CRLB theory. This helps us in obtaining the minimum variance of the unbiased estimator for given hyper-parameter. We also obtain the required number of snapshots for the good estimate of hyper-parameters using CRLB expression. 2. We show that the localization accuracy with fingerprint constructed using GPR is higher than the Horus fingerprinting approach. Further, localization accuracy is not significantly affected by the reduction in number of samples at each fingerprint location. Accuracy improvements of 10% and 30% are observed in two sites compared to the Horus fingerprinting approach. 3. We further illustrate that localization accuracy is relatively insensitive to the choice of different kernel functions such as Gaussian, Laplacian and Exponential. The performance of Laplacian and Exponential kernel functions is the same because the only difference lies with the length scale parameter. 4. There are plenty of insignificant APs like commuter phone wi- fi, vehicle wi-fi besides fixed APs, in crowded wireless environment, e.g, supermarket. To find out the set of strongest APs in the given area of interest, a criterion based on dimensionality reduction using principal component analysis is employed. The remainder of the paper is organized as follows: Section 2 overviews existing techniques for localization. Section 3 describes the Gaussian process regression for fingerprinting based localization. Performance evaluation is presented in Section 4. A brief conclusion is presented in Section 5. Cramér-Rao lower bound (CRLB) analysis for the kernel function parameters is discussed in Appendix.
برای دانلود رایگان مقاله فرآیند رگرسیون گوسی اینجا کلیک نمایید.
کلمات کلیدی:
Gaussian process - Wikipedia https://en.wikipedia.org/wiki/Gaussian_process In probability theory and statistics, a Gaussian process is a particular kind of statistical model ..... Gaussian process regression can be further extended to address learning tasks in both supervised (e.g. probabilistic classification) and ... Definition · Alternative definitions · Covariance functions · Brownian Motion as the ... [PDF]Gaussian Processes for Machine Learning www.gaussianprocess.org/gpml/chapters/RW2.pdf In this chapter we describe Gaussian process methods for regression problems; ... One can think of a Gaussian process as defining a distribution over functions,. [PDF]Introduction to Gaussian Process Regression www.inference.phy.cam.ac.uk/hmw26/papers/gp_intro.pdf by HM Wallach - 2005 - Cited by 1 - Related articles Jan 25, 2005 - Making predictions. Model selection: hyperparameters. Hanna M. Wallach hmw26@cam.ac.uk. Introduction to Gaussian Process Regression ... 1.7. Gaussian Processes — scikit-learn 0.18.1 documentation scikit-learn.org/stable/modules/gaussian_process.html Gaussian Processes (GP) are a generic supervised learning method designed to solve ... implements Gaussian processes (GP) for regression purposes. For this ... [PDF]Gaussian Processes: A Quick Introduction https://www.robots.ox.ac.uk/~mebden/reports/GPtutorial.pdf by M Ebden - 2015 - Cited by 4 - Related articles Gaussian process regression (GPR) is an even finer approach than this. ... Gaussian processes (GPs) extend multivariate Gaussian distributions to infinite dimen ... Gaussian Process Regression Models - MATLAB & Simulink https://www.mathworks.com › ... › Regression › Gaussian Process Regression Gaussian process regression (GPR) models are nonparametric kernel-based probabilistic models. Gaussian Processes for Dummies · - ohAI katbailey.github.io/post/gaussian-processes-for-dummies/ Aug 9, 2016 - That's when I began the journey I described in my last post, From both sides now: the math of linear regression. Gaussian Processes (GPs) are ... Gaussian Processes for regression: a tutorial https://paginas.fe.up.pt/~dee10008/papers/201201_report_ML_jmelo.pdf by J Melo - Cited by 2 - Related articles Gaussian Processes for regression: a tutorial. José Melo. Faculty of Engineering, University of Porto. FEUP - Department of Electrical and Computer Engineering. Gaussian Processes for Regression - NIPS Proceedings https://papers.nips.cc/paper/1048-gaussian-processes-for-regression.pdf by CKI Williams - Cited by 807 - Related articles Gaussian Processes for Regression. Christopher K. I. Williams. Neural Computing Research Group. Aston University. Birmingham B4 7ET, UK.
