Godkända
Frekvensoffsetsestimering för multipelanvändarsystem baserat på en låg dimensionell approximering av likelihoodfunktionen
Juan Carlos Bucheli Garcia (2014)
Start
2015-11-02
Presentation
2016-05-13 10:15
Plats:
E:2349
Avslutat:
2016-05-20
Examensrapport:
Sammanfattning
The project is based on the industry requirement of decreasing fabrication cost for wireless enabled devices and reducing the effect of non-ideal oscillators over the performance of the so mentioned devices. This is done through estimation of frequency offset from its nominal value. The reason for the deviation of the frequency as previously stated is the use of low price oscillators and also environmental factors as temperature and deterioration over time. In order to do so, based on the current work of Associate Professor Fredrik Rusek on maximum likelihood estimation for frequency offset through a low-dimensional estimator, an extension to multiple user systems is to be evaluated. Some of the issues to be considered in the project are the complexity of the estimator when the number of user deviates from one. Also, finding the number of users supported by a considerably low-complexity estimator in case it does not grow too large with the number of terminals. The objective of the project is to evaluate, propose and simulate a complete estimator for multiuser OFDM systems complying with the mentioned requirements based on previous work (e.g. F. Rusek, “On the (Low) Dimensionality of the log-Likelihood Function for Frequency Offset Estimation”). In order to do so, the main tasks where divided into the preliminary calendar corresponding to spring 2016 and starting from November 2015 as shown below: Part 1 (2015-11-2/11-16): getting familiar with the background and the system. Gather information and relevant knowledge about basis approximations (Two weeks). Part 2 (2015-11-16/11-30): Seek low-dimensional approximations to the log-likelihood function. Investigate different methods and techniques (Two weeks). Part 3 (2016-01-14/02-12): From the approximation generated from part 2, find an estimator with low-complexity to the frequency offset estimation problem (One month). Part 4 (2016-02-15/03-15): Build simulation chain and test performance under different settings. Investigate the number of users that can be simultaneously supported (One month). Part 5 (2016-03-16/03-21): Based on the output from part 4, return to part 2 and 3 and cross-examine the solution. See if something can be improved upon (One week). Part 6 (2016-03-22/04-21): Document. This will also be done continuously throughout the thesis (One month).
Handledare: Ove Edfors (EIT)
Examinator: Fredrik Rusek (EIT)