Return to search

Multiple-Input Multiple-Output Detection Algorithms for Generalized Frequency Division Multiplexing

Since its invention, cellular communication has dramatically transformed personal lifes and the evolution of mobile networks is still ongoing. Evergrowing demand for higher data rates has driven development of 3G and 4G systems, but foreseen 5G requirements also address diverse characteristics such as low latency or massive connectivity. It is speculated that the 4G plain cyclic prefix (CP)-orthogonal frequency division multiplexing (OFDM) cannot sufficiently fulfill all requirements and hence alternative waveforms have been in-vestigated, where generalized frequency division multiplexing (GFDM) is one popular option. An important aspect for any modern wireless communication system is the application of multi-antenna, i.e. MIMO techiques, as MIMO can deliver gains in terms of capacity, reliability and connectivity. Due to its channel-independent orthogonality, CP-OFDM straightforwardly supports broadband MIMO techniques, as the resulting inter-antenna interference (IAI) can readily be resolved. In this regard, CP-OFDM is unique among multicarrier waveforms. Other waveforms suffer from additional inter-carrier interference (ICI), inter-symbol interference (ISI) or both. This possibly 3-dimensional interference renders an optimal MIMO detection much more complex. In this thesis, weinvestigate how GFDM can support an efficient multiple-input multiple-output (MIMO) operation given its 3-dimensional interference structure. To this end, we first connect the mathematical theory of time-frequency analysis (TFA) with multicarrier waveforms in general, leading to theoretical insights into GFDM. Second, we show that the detection problem can be seen as a detection problem on a large, banded linear model under Gaussian noise. Basing on this observation, we propose methods for applying both space-time code (STC) and spatial multiplexing techniques to GFDM. Subsequently, we propose methods to decode the transmitted signals and numerically and theoretically analyze their performance in terms of complexiy and achieved frame error rate (FER). After showing that GFDM modulation and linear demodulation is a direct application of Gabor expansion and transform, we apply results from TFA to explain singularities of the modulation matrix and derive low-complexity expressions for receiver filters. We derive two linear detection algorithms for STC encoded GFDM signals and we show that their performance is equal to OFDM. In the case of spatial multiplexing, we derive both non-iterative and iterative detection algorithms which base on successive interference cancellation (SIC) and minimum mean squared error (MMSE)-parallel interference cancellation (PIC) detection, respectively. By analyzing the error propagation of the SIC algorithm, we explain its significantly inferior performance compared to OFDM. Using feedback information from the channel decoder, we can eventually show that near-optimal GFDM detection can outperform an optimal OFDM detector by up to 3dB for high SNR regions. We conclude that GFDM, given the obtained results, is not a general-purpose replacement for CP-OFDM, due to higher complexity and varying performance. Instead, we can propose GFDM for scenarios with strong frequency-selectivity and stringent spectral and FER requirements.

Identiferoai:union.ndltd.org:DRESDEN/oai:qucosa:de:qucosa:30965
Date14 September 2018
CreatorsMatthé, Maximilian
ContributorsFettweis, Gerhard, Bölcskei, Helmut, Technische Universität Dresden
Source SetsHochschulschriftenserver (HSSS) der SLUB Dresden
LanguageEnglish
Detected LanguageEnglish
Typedoc-type:doctoralThesis, info:eu-repo/semantics/doctoralThesis, doc-type:Text
Rightsinfo:eu-repo/semantics/openAccess

Page generated in 0.0017 seconds