本论文开发出了一个全新的理论和算法框架用於无线网络的分布式功率控制。我们提出两种快速分布式功率控制算法，并对此作了深入的研究。 此种算法相当普适，比如适用于目前热门的LTE和认知无线电网络。 它在解的最优性以及收敛速度等方面击败了著名的高通公司的"荷载溢出型分布式功率控制算法" (收录于重要论文[HandeRanganChiangWu08] )以及"分布式加权比例型信干噪比均衡算法" (收录于重要论文[TanChiangSrikant 11)。 / 作为一个重要而富有挑战性的研究课题，通过分布式功率控制达至无线网络效用的最大化一直受到业界的普遍关注。 这方面的研究通常把问题表述为一个最优化问题，即在某些功率约束条件下，优化整体系统的效用函数。 (其中，系统的效用函数通常是各无线收发链路的信干噪比的增函数。 )此问题已经有了不错的集中式解决方案，但成本更低廉、更易于布置、更为实用的分布式解决方案则欠奉，尤其是经严格证明可行的分布式解决方案。 这是因为分布式算法一般只适用于相对简单或者有特殊结构的优化问题。 而无线设备之间的相互干扰和各自信号功率之间的复杂关系使得分布式求解极其困难。 在算法设计上，很小的疏漏就可能导致解决方案无效或者不收敛。 例如，尽管论文[HandeRanganChiangWu08] 和[TanChiangSrikant 11] 都声称各自的分布式算法提供了问题的最优解，但我们通过大量的仿真实验以及理论研究发现并非如此。 我们发现"荷载溢出型分布式功率控制算法"时常要么无法收敛，要么收敛得极其慢。而"分布式加权比例型信干噪比均衡算法"则经常在几次迭代之後就已经发散。 / 我们开发出了全新的分析和算法框架，并将其推广到适用于一般线性功率约束的情况。(前述论文的分析框架是基于某些非常特殊的线性功率约束。)在此基础上，我们逐一找出了前述算法中的错漏之处，并设计出我们的分布式梯度投影功率控制算法，以及与之相匹配的步长规则。 我们严格证明了该步长规则的有效性和算法的收敛性、最优性，并给出了算法复杂度的分析。(相较之下， [HandeRanganChiangWu08] 在算法收敛性证明上语焉不详，在其它方面则付之阙如;而[TanChiangSrikant 11] 的算法收敛性证明存在明显错误，在其它方面同样付之阙如。 )在某些情况下，我们的算法可以进一步提速并提升运行性能。 大量的仿真实验证实我们的算法在解的最优性和运行速度两方面都较前述算法优越。在某些情况下，我们算法的收敛速度上百倍快于前述算法。 / 总而言之，本论文成功解决了重要的效用优化问题并取得比前述论文更好的结果。它开发出全新的理论和算法框架，完全解决了步长规则和收敛性、最优性这些难题。展望未来，我们相信，本论文为快速功率控制在无线和移动解决方案中的应用打下了坚实的理论基础。 我们期待该理论框架能够提供更多問題的解決方案。 / This thesis develops a new theoretical and algorithmic framework for practical distributed power control in wireless networks. It proposes and investigates fast optimal distributed power control algorithms applicable to LTE as well as cognitive radio. The proposed algorithms beat the well-known Qualcomm's load-spillage distributed power control algorithm in [HandeRan-ganChiangWu08] and the distributed weighted proportional SINR algorithm in [TanChiangSrikant11] in terms of both the optimality of the solution and the convergence speed. / Wireless network utility maximization via distributed power control is a classical and challenging issue that has attracted much research attention. The problem is often formulated as a system utility optimization problem under some transmit power constraints, where the system utility function is typically an increasing function of link signal-to-interference-plus-noise-ratio (SINR). This problem is complicated by the fact that these wireless devices may interfere with each other. In particular, the wireless devices are affected by each other's transmit power, and the transmit powers and interferences experienced by the devices are interwoven in a complex manner. / Despite that, there have been good centralized algorithms for solving the problem. "Decentralized" solutions, on the other hand, are a different story. In practice, decentralized algorithms in which the devices interact with each other in a loosely coupled manner to improve the network utility, are easier to deploy than centralized algorithms. However, the design of workable (and provably workable in the mathematical sense) solution is very challenging. Small neglects can lead to solutions that are invalid or non-convergent. For example, although both papers [HandeRanganChiangWu08] and [TanChiangSrikant11] claim their distributed algorithms to be optimal, we discover some experimental evidence suggesting that certain parts of these algorithms are not quite right. Oftentimes, the former fails to converge or converges extremely slowly, while the latter could diverge in the first few iterations. / To fix these glitches and to broaden the scope of the problem, we develop a new analytical and algorithmic framework with a more general formulation. With this framework, we can identify the sources of the defects and shortcomings of prior algorithms. We further construct an optimal distributed (sub)gradient projection algorithm with provably valid step size rules. Rigorous convergence proof and complexity analysis for our algorithm are given (note: convergence proof and complexity analysis were missing in [HandeRanganChiangWu08] and incorrect in [TanChiangSrikant11]). In some scenarios, our algorithm can be further accelerated to yield even better performance. Extensive simulation experiments confirm that our algorithms always outperform the prior algorithms, in terms of both optimality and efficiency. Specifically, simulation demonstrates at least 100 times faster convergence than the prior algorithms under certain scenarios. / In summary, this thesis solves the important SINR-based utility maximization problem and achieves significantly better results than existing work. It develops a new theoretical an dalgorithmic framework which completely addresses the difficult convergence and step-size issues. Going forward, we believe the foundation established in this work will open doors to other fast distributed wireless and mobile solutions to problems beyond the power control problem addressed here. / Detailed summary in vernacular field only. / Detailed summary in vernacular field only. / Detailed summary in vernacular field only. / Detailed summary in vernacular field only. / Zhang, Jialiang. / Thesis (Ph.D.)--Chinese University of Hong Kong, 2013. / Includes bibliographical references (leaves 83-87). / Electronic reproduction. Hong Kong : Chinese University of Hong Kong, [2012] System requirements: Adobe Acrobat Reader. Available via World Wide Web. / Abstract also in Chinese. / Chapter 1 --- Introduction --- p.1 / Chapter 1.1 --- Overview --- p.1 / Chapter 1.2 --- Thesis Organization --- p.6 / Chapter 1.3 --- Notations --- p.7 / Chapter 2 --- System Model and Problem Formulation --- p.8 / Chapter 2.1 --- System Model --- p.8 / Chapter 2.2 --- Nonnegative Linear Power Constraints --- p.9 / Chapter 2.3 --- Network Utility --- p.10 / Chapter 2.4 --- Problem Formulation --- p.11 / Chapter 2.5 --- Characterization of T[subscript c] --- p.13 / Chapter 2.6 --- Multiple Constraints --- p.16 / Chapter 3 --- Nice Properties of SINR Constraints --- p.18 / Chapter 3.1 --- Convexity, Differentiability and Monotonicity --- p.19 / Chapter 3.2 --- Fast Distributed Gradient Computation --- p.20 / Chapter 3.2.1 --- Distributed SINR-Driven Single-Constrained Power Control --- p.21 / Chapter 3.2.2 --- Network Duality --- p.23 / Chapter 3.3 --- The Case of Multiple Constraints --- p.27 / Chapter 4 --- Network Utility Maximization in Log-SINR Domain --- p.32 / Chapter 4.1 --- Single Active Constraint and Ascent Directions --- p.34 / Chapter 4.2 --- Multiple Constraints and Subgradient Projection --- p.39 / Chapter 4.3 --- Unconstrained Equivalence and Complexity results of M = 1 --- p.46 / Chapter 4.4 --- Simulation Experiments --- p.52 / Chapter 4.4.1 --- Simulation Settings --- p.52 / Chapter 4.4.2 --- Negative results of algorithm 6 in [7] --- p.54 / Chapter 4.4.3 --- Negative results of Qualcomm’s load-spillage algorithm in [25] --- p.56 / Chapter 4.4.4 --- More results of our algorithms --- p.62 / Chapter 5 --- Related Work --- p.64 / Chapter 6 --- Conclusion --- p.68 / Chapter 7 --- Appendix --- p.72
| Identifer | oai:union.ndltd.org:cuhk.edu.hk/oai:cuhk-dr:cuhk_328223 |
| Date | January 2013 |
| Contributors | Zhang, Jialiang., Chinese University of Hong Kong Graduate School. Division of Information Engineering. |
| Source Sets | The Chinese University of Hong Kong |
| Language | English, Chinese |
| Detected Language | English |
| Type | Text, bibliography |
| Format | electronic resource, electronic resource, remote, 1 online resource (xvii, 87 leaves) : ill. (some col.) |
| Rights | Use of this resource is governed by the terms and conditions of the Creative Commons “Attribution-NonCommercial-NoDerivatives 4.0 International” License (http://creativecommons.org/licenses/by-nc-nd/4.0/) |
Page generated in 0.003 seconds