## Files in this item

FilesDescriptionFormat

application/pdf

Peter_Kairouz.pdf (2MB)
(no description provided)PDF

## Description

 Title: MIMO communications over multi-mode optical fibers Author(s): Kairouz, Peter Advisor(s): Singer, Andrew C.; Shanbhag, Naresh R. Department / Program: Electrical & Computer Eng Discipline: Electrical & Computer Engr Degree Granting Institution: University of Illinois at Urbana-Champaign Degree: M.S. Genre: Thesis Subject(s): Communication theory optical fiber communications multi-input multi-output (MIMO) communications multi-mode fibers (MMF) systems Orthogonal Frequency Division Multiplexing (OFDM) detection algorithms single carrier systems random matrix theory information theory Abstract: We consider multi-input multi-output (MIMO) communications over multi-mode fibers (MMFs). Current MMF standards, such as OM3 and OM4, use fibers with core radii of 50\,$\mu$m, allowing hundreds of modes to propagate. Unfortunately, due to physical and computational complexity limitations, we cannot couple and detect hundreds of data streams. In order to circumvent this issue, two solutions were presented in the literature. The first is to design new fibers with smaller radii so that they can support a desired number of modes. The second is to design multi-core fibers with a reasonable number of cores. However, both approaches are expensive as they necessitate the replacement of installed fibers. In our work, we consider input-output coupling schemes that allow the user to couple and extract a reasonable number of signals from a fiber with many modes. This approach is particularly attractive as it is scalable; i.e., the fibers do not have to be replaced every time the number of transmitters or receivers is increased (which is likely to happen in the near future). In addition, fibers with large radii can support higher peak powers, relative to fibers with small radii, while still operating in the linear regime. However, the only concern is that fibers with more modes suffer from increased mode-dependent losses (MDLs). Our work addresses this last concern. This thesis is divided into two parts. In the first part, we present a channel model that incorporates intermodal dispersion, chromatic dispersion, mode dependent losses, and mode coupling. We later extend this model to include the input and output couplers and provide an input-output coupling strategy that leads to an increase in the overall capacity. This strategy can be used whenever channel state information (CSI) is available at the transmitter and the designer has full control over the couplers. We show that the capacity of an $N_t \times N_t$ MIMO system over a fiber with $M \gg N_t$ modes can approach the capacity of an $N$-mode fiber with no loss. Moreover, we present a statistical input-output coupling model in order to quantify the loss in capacity when CSI is not available at the transmitter or there is no control over the input-output coupler. It turns out that the loss, relative to $N_t$-mode fibers, is minimal (less than 0.5 dB) for a wide range of signal-to-noise ratios (SNRs) and a reasonable range of MDLs. This means that there is no real need to replace the already installed fibers and that our strategy is indeed a better approach to solving the above problem. In the second part, we explore reduced complexity maximum likelihood sequence detection (MLSD) algorithms for single carrier MIMO systems. These algorithms can be used for optical as well as wireless communications. We show that a sphere decoding (SD)-like approach can be used to reduce the computational complexity of the vector Viterbi algorithm (VVA), an extension to the Viterbi algorithm for MIMO systems. Our combined SD-VVA approach is attractive because it provides substantial computational savings while solving an exact MIMO MLSD problem. Our results show a $50\%$ reduction in complex multiplications and real additions, relative to the full VVA, for a $2\times2$ MIMO system using $16$-QAM signal constellation and operating at an signal-to-noise ratio ($\SNR$) of $10$ dB. This figure is increased to $60\%$ when the $\SNR$ is increased to $15$ dB. We show that larger savings can be achieved for larger MIMO systems and higher order signal constellations. Finally, we show how our algorithm can be modified in order to further reduce the complexity of VVA while still achieving close to optimal performance. Issue Date: 2013-02-03 URI: http://hdl.handle.net/2142/42285 Rights Information: Copyright 2012 Peter Kairouz Date Available in IDEALS: 2013-02-03 Date Deposited: 2012-12
﻿