Titelaufnahme

Titel
User cooperation in wireless time-variant channels / von Zhinan Xu
VerfasserXu, Zhinan
Begutachter / BegutachterinZemen, Thomas
ErschienenWien, 2016
Umfangxv, 141 Seiten : Illustrationen, Diagramme
HochschulschriftTechnische Universität Wien, Univ., Dissertation, 2016
SpracheEnglisch
DokumenttypDissertation
Schlagwörter (EN)Interference Alignment / Limited Feedback / Vehicular channel modeling / GSCM / Cluster
URNurn:nbn:at:at-ubtuw:1-7619 Persistent Identifier (URN)
Zugriffsbeschränkung
 Das Werk ist frei verfügbar
Dateien
User cooperation in wireless time-variant channels [1.85 mb]
Links
Nachweis
Klassifikation
Zusammenfassung (Englisch)

For wireless communications, the spectrum limitation and user mobility are two major challenges to promise reliable communications with high data rates and low latency. In order to improve spectral efficiency, future mobile systems employ full reuse of wireless spectrum, which makes interference management become a funda- mental prerequisite to cope with the growing inter-cell interference. Future mobile standards, e.g. 5G, will support critical applications that require ultra reliability and low latency. Examples include traffic safety applications and autonomous driving. In this thesis, we propose and analyze interference management techniques for interference channels. We devise non-stationary vehicular channel modeling approaches and evaluate the cooperative communication performance. In the first part of the thesis, interference alignment (IA) in K-user interference channels is studied. Different interference management schemes for K-user interference channels are introduced and compared. Since global channel state inform tion (CSI) plays a central role to achieve IA as well as the optimal degrees of freedom (DoF), we propose a joint channel estimation, feedback and prediction framework for IA in time-variant channels. The proposed algorithm allows reduced-rank channel prediction. An upper bound for the rate loss caused by feedback quantization and channel prediction is derived. We develop a subspace dimension switching algorithm, which is able to find the subspace dimension associated with a higher rate. The scaling of the required number of bits is characterized in order to decouple the rate loss due to quantization from the transmit power. To relief the global CSI burden, we study opportunistic interference alignment (OIA) algorithms, which exploit channel randomness and multiuser diversity by user selection. We tackle the problem of feedback reduction for OIA using threshold-based feedback schemes. We investigate different threshold choices, user scaling laws and the achievability of DoF for real-valued feedback and 1-bit feedback, respectively. For OIA with real-valued feedback, the threshold and the corresponding feedback load to achieve the optimal DoF are analyzed. For OIA with 1-bit feedback, we find an optimal choice of the 1-bit quantizer to achieve the DoF d = 1. For DoF with d > 1, an asymptotic threshold choice is provided by solving an upper bound for the rate loss. In the second part of the thesis, we focus on channel modeling and perfor- mance analysis for vehicular communications. A geometry-based stochastic channel model (GSCM) for road intersections is developed. We use the proposed model to evaluate the communication performance in terms of frame error rate. The evaluation is performed at various transmitter/receiver locations and velocities with three different types of channel estimators. In order to overcome the low signal-to-noise ratio due to non-line-of-sight, we deploy a relay node at the intersection, which increases the transmission reliability significantly. In order to reduce the complexity of the GSCM, a cluster-based vehicular channel model is proposed. The cluster-based model achieves a low computational complexity suitable for a real-time implementation. We apply a joint cluster identification-and-tracking algorithm to the measurement data in delay-Doppler plane. We divide the cluster locations in the delay-Doppler plane into different characteristic regions and characterize the time-variant cluster parameters in each region. For a low-complexity implementation, we draw the cluster parameters randomly using pre-computed dis- tributions. The proposed model is validated with measurement data.