Uniform TitleResource allocation in coordinated and un-coordinated wireless systems with greedy or non-greedy users
NameSingh, Jasvinder (author), Rose, Christopher (chair), Yates, Roy (internal member), Mandayam, Narayan (internal member), Spasojevic, Predrag (internal member), Balachandran, Krishna (outside member), Rutgers University, Graduate School - New Brunswick,
SubjectElectrical and Computer Engineering,
Wireless communication systems,
Code division multiple access,
Adaptive signal processing--Algorithms
DescriptionIn this thesis, we investigate wireless resource optimization problems arising in the context of unlicensed bands. The first half of this thesis assumes a multiple access channel communication model (many transmitters talking to a single receiver), while the second half assumes an interference channel (a collection of multiple interfering transmitter-receiver pairs). The problems considered can further be classified based on the level of coordination available among the devices, and based on the optimization objectives of the devices. The devices are either greedy (they choose their actions to maximize their own utility), or are non-greedy (they choose their actions to maximize/satisfy a common social objective). We investigate four resource optimization scenarios that encompass the possibilities of presence or absence of coordination infrastructure as well as the cases of greedy and non-greedy devices. As an example of the greedy uncoordinated scenario, we consider application of interference avoidance algorithms in generalized CDMA systems.
We introduce variants of standard interference avoidance procedures which produce more easily tracked incremental codewords, and study the response of the system to abrupt changes in the background interference as might be encountered in a practical system. Next we consider a sensor network scenario where multiple sensors are transmitting correlated symbols to a common receiver, with an objective of minimizing the total mean square error (TMSE) in the symbol estimates at the receiver. The source-channel separation theorem for the point to point case does not hold for this problem and the optimal communication scheme is unknown. We propose a CDMA based transmission scheme that exploits the correlation between the sensors' symbols and facilitates statistical cooperation among the sensors through the choice of their codewords. We give an analytical characterization of the TMSE-minimizing codeword set for this scheme, and compare its performance both with a separation-based scheme, and an information-theoretic upper bound.
The second half of this thesis considers scheduling problems for interfering links where each link employs an ON-OFF modulation scheme in each time slot. First we consider the case of non-greedy interfering links and come up with a distributed scheduling solution that requires no coordination among the links. We prove the convergence of our distributed scheduling algorithm and compare its
performance with centralized scheduling. After this, we consider the case of greedy interfering links that coordinate with each other through a mediating authority called the "spectrum server". Each link reports the set of links that interfere with it, based on which the spectrum server constructs the interference graph and finds an optimal schedule for the links (maximizing a certain global objective). Since the links are greedy, they will choose their reports to maximize their individual utilities. We therefore investigate the following natural question: "Under what conditions is it realistic to assume that the links will send truthful reports?" and present some preliminary results.
NoteIncludes bibliographical references (p. 84-88).
CollectionGraduate School - New Brunswick Electronic Theses and Dissertations
Organization NameRutgers, The State University of New Jersey
RightsThe author owns the copyright to this work.