Description
TitleOptimization techniques for configuration of advanced wireless networks
Date Created2011
Other Date2011-01 (degree)
Extentviii, 93 p. : ill.
DescriptionThis thesis is a collection of research topics that pertain to the optimal configuration of wireless networks. We start with techniques for the configuration of indoor testbeds that are used to assess novel protocols and application concepts under controlled and repeatable conditions. We also examine the wireless networks in which nodes can exploit stochastic energy harvesting, wherein the network configurations are described by energy-aware packet transmission policies. Finally, we consider an enhanced achievability scheme for the non-degraded Gaussian fading broadcast channel with channel state known only at the receivers. In this system, a network configuration is given by the assignment of transmitted signal components to the receivers. A fundamental issue common to synergistic research via testbed emulation is the replication of communication links of specified quality. It has been shown that the system-level mapping from the real world to the testbed can be established by the equivalence of average link gains. In this thesis, we investigate link gain matrix mapping methodologies for two different applications: hierarchical networks with a fixed access point (AP), and mesh networks. For the AP networks, we seek to minimize the root-mean-square mapping difference between the testbed and the real-world. For the mesh networks, we mimic the effects of log-normal shadow fading in the real world by inducing a Gaussian distribution for the distance-dependent part of dB link gain differences. Assuming free-space path loss on the indoor testbed, we present results for a variety of indoor and outdoor real-world scenarios, and we discuss the generality of our method. To acquire reasonably accurate link gain matrix for large-scale distributed wireless networks without exhaustive measurements, we propose a novel method for gain matrix estimation, which classifies the network links into distinctive categories according to their obstruction geometry and model them separately. We also develop a sampling method for selecting, in a structured way, the links to be measured in each category. The results show that all gain matrix elements can be predicted with good accuracy by measuring only a small fraction of the entire set. For this study, we invoke a ray-tracing tool in obtaining the actual gains to avoid the need for field measurements. Out of the demand for clean and cheap fuel, the renewable energy harvested from the environment has received ever-increasing attention. However, power management stands as a crucial issue for such a resource due to the uncertainty of stochastic replenishment. In this thesis, we introduce a Markov chain model to characterize different modes of energy replenishment. Depending on the energy status of a battery and the reward for successfully transmitting a message, we prove the existence of an optimal transmission policy that can maximize the average reward rate. Compared with the unconditional transmit-all policy, which transmits every message as long as the energy storage is positive, the proposed strategy is shown to achieve significant gains in the average reward rate. Finally, we propose an improved achievability strategy by applying reverse stripping (RS) and soft-decision decoding (SDD) towards the binary expansion superposition (BES) signaling proposed by Tse and Yates. Multi-stage reverse stripping at each receiver permits partial cancelation of the less-significant-bit (LSB) interference, which facilitates a decomposition of the fading channel into parallel effective channels and a simplified design for capacity-approaching codes. We observe that for the degraded AWGN broadcast channel and for the non-degraded intermittent AWGN channel, the inner bound achieved by BES signaling with RS-SDD can come within one bit of the capacity boundary for moderate to high SNR levels.
NotePh.D.
NoteIncludes bibliographical references
NoteIncludes vita
Noteby Jing Lei
Genretheses, ETD doctoral
Languageeng
CollectionGraduate School - New Brunswick Electronic Theses and Dissertations
Organization NameRutgers, The State University of New Jersey
RightsThe author owns the copyright to this work.