Autonomous power control for next-generation wireless computing
Abstract/Contents
- Abstract
- In wireless communication networks, transmit power control (TPC) can significantly improve network performance by reducing interference and improving energy efficiency. In autonomous TPC, each transmitter is responsible for regulating its own power without centralized coordination and without explicit control information exchanged between different links. Autonomous TPC is a particularly important area of research for wireless networks in which centralized control is difficult or impractical - examples of such networks include WiFi, sensor networks, and the emerging Internet of Things. In these networks, the importance of TPC is being magnified by the rapidly increasing density of heterogeneous wireless devices that have limited available energy and processing power. In this dissertation, we use a dynamic programming framework to design low-complexity, fully autonomous TPC algorithms which exhibit promising performance characteristics for both current and next-generation wireless networks. The dynamic programming framework we use is based on a fundamental tradeoff between transmit power and delay; for example, a link faced with high interference is allowed to queue up (delay) packets to be transmitted at a later time, when the interference may be lower. We develop two new classes of autonomous TPC algorithms, and we describe these algorithms' dynamics and show how their nonlinear functional forms can lead to a network behavior called ``induced'' time division multiple access (iTDMA), in which links automatically take turns utilizing the channel without any explicit cross-link coordination. As a result of this behavior, we demonstrate that our algorithms can provide substantial throughput improvements over previous schemes. Additionally, we discuss various problems that arise in implementing our algorithms, such as synchronization and scaling issues, and explain how they can be resolved. The most critical problem we describe is ``locking, '' in which links using our algorithms spontaneously synchronize into inefficient power evolution patterns which compromise network performance. We propose several randomization-based mitigation mechanisms to alleviate this behavior, and we demonstrate that these mechanisms considerably improve our algorithms' performance. Finally, we develop and analyze continuous-time analogs (fluid models) of the discrete-time versions of our algorithms in order to explore the iTDMA behavior further. This analysis provides new insights into the throughput improvements of our algorithms over canonical approaches, particularly in high-interference channels. It also represents a first step toward understanding the complex mathematical stability properties of our algorithms.
Description
| Type of resource | text |
|---|---|
| Form | electronic; electronic resource; remote |
| Extent | 1 online resource. |
| Publication date | 2014 |
| Issuance | monographic |
| Language | English |
Creators/Contributors
| Associated with | Mounzer, Jeffrey |
|---|---|
| Associated with | Stanford University, Department of Electrical Engineering. |
| Primary advisor | Bambos, Nicholas |
| Thesis advisor | Bambos, Nicholas |
| Thesis advisor | Osgood, Brad |
| Thesis advisor | Singh, Jatinder |
| Advisor | Osgood, Brad |
| Advisor | Singh, Jatinder |
Subjects
| Genre | Theses |
|---|
Bibliographic information
| Statement of responsibility | Jeffrey Mounzer. |
|---|---|
| Note | Submitted to the Department of Electrical Engineering. |
| Thesis | Thesis (Ph.D.)--Stanford University, 2014. |
| Location | electronic resource |
Access conditions
- Copyright
- © 2014 by Jeffrey Joseph Mounzer
- License
- This work is licensed under a Creative Commons Attribution Non Commercial 3.0 Unported license (CC BY-NC).
Also listed in
Loading usage metrics...