Support Vector Machine Regression and Artificial Neural Network for Channel Estimation of LTE Downlink in High-Mobility Environments
Journal Title: Transactions on Machine Learning and Artificial Intelligence - Year 2016, Vol 4, Issue 4
Abstract
In this paper we apply and assess the performance of support vector machine regression (SVR) and artificial neural network (ANN) channel estimation algorithms to the reference signal structure standardized for LTE Downlink system. SVR and ANN where applied to estimate real channel environment such as vehicular A channel defined by the International Telecommunications Union (ITU) in the presence of nonlinear impulsive noise. The proposed algorithms use the information provided by the received reference symbols to estimate the total frequency response of the time variant multipath fading channel in two phases. In the first phase, each method learns to adapt to the channel variations, and in the second phase it predicts all the channel frequency responses. Finally, in order to evaluate the capabilities of the designed channel estimators, we provide performance of SVR and ANN, which is compared with traditional Least Squares (LS) and Decision Feedback (DF). The simulation results show that SVR has a better accuracy than other estimation techniques.
Authors and Affiliations
Anis Charrada, Abdelaziz Samet
Keywords
Related Articles
- EP ID EP277211
- DOI 10.14738/tmlai.44.2145
- Views 43
- Downloads 0
Disassembly Modeling of an of End-Of-Life (EOL) Mechanical Damper for Recycling
Today’s rapidly developing technologies and product designs have enabled manufacturers to deliver new products to consumers at a dramatic rate. This has in turn resulted in shorter lifespan for products, because, more of...
A Coordinate Transformation Method based on the Random Variable for Showing the Optical Properties
A coordinate transformation method of square projection circular with random various has been proposed for the optical properties of the cloaking material. The square produced randomly is projected to the circle and all...
Smart Classroom Environment Via IoT in Basic and Secondary Education
The rapid evolution of information and communication technologies (ICT) has created a new paradigm of the Internet, known as the Internet of Things (IoT), which represents a group of connected objects Using wireless link...
SIMARECH 3: A New Application for the Governance of Scientific Research
In the international context of research assessment, many countries appear to move critically towards a quantitative evaluation policy, which would eventually complement or replace the traditional peer review system. How...
An Unsupervised Neural Network Method for Age Group Estimation using Facial Features
Age estimation from facial features is an important research subject in the field of face recognition. It is an active research area that can be used in wide range of applications such as surveillance and security, telec...