Modular Parameter Optimization in Layered Communication Systems

• Applicant: Professor Dr.-Ing. Wolfgang K. Utschick
• Subject Area: Electronic Semiconductors, Components and Circuits, Integrated Systems, Sensor Technology, Theoretical Electrical Engineering
• Term: from 2008 to 2012
• Project identifier: Deutsche Forschungsgemeinschaft (DFG) - Project number 62502067

• Final Report Year: 2012

Final Report Abstract

In a layered communication system, the functionality needed for transmission of data is divided into a set of subfunctions, or layers. In general, each layer has a number of adjustable parameters, and each parameter has an impact on the overall system performance and also, either directly or indirectly, on the usage of physical resources. In this research project, the problem of determining a parameter setup that maximizes system performance under given resource constraints was considered. In particular, we had been focusing on modular strategies for solving the parameter optimization problem. Starting with a proposal for a new local optimization technique, we finally provided a general framework for utility maximization in convex and nonconvex wireless systems. Similar to standard decomposition approaches a two-layer model was used, consisting of a utility model and a physical layer model, introducing the concept of a rate region as a generic interface between both layers. Based on a monotonicity assumption, utility was maximized on the Pareto boundary of the rate region. It was further shown that the Pareto set of a proper rate region is essentially a topological manifold. Consequently, three methods for solving the respective utility maximization problem were provided. Since the parameterization of the physical layer based on a statistical description turned out to be pretty incompatible with the targeted framework, we decided to give up this direction during the rest of the project. However, we have been successfull in two other case studies where we were able to demonstrate the potentials of the proposed framework. Each case study resulted from the combination of a concrete physical layer model with a set of utility models. The solution of the physical layer subproblems was specific for each physical layer model. Moreover, the physical layer subproblems could again be formulated as utility maximization problems. As a result, the solution to the original utility maximization problem was obtained by a combination of the different methods presented in this work. As a follow-up result of this project we were able to introduce a new monotonic optimization framework for coordinated beamforming in multi-cell networks. Coordinated beamforming adresses an essential issue in wireless multi-cell networks where intercell interference is the major limiting factor. In essence, the proposed framework represents a powerful tool for computing benchmarks for certain scenarios and utility functions under jointly optimal beamforming and scheduling. As the major outcome of this research project we consider the introduction and analysis of a local and global optimization framework for modular parameter optimization in layered communication systems. As far as global optimization techniques are concerned the economical usability is rather limited due to numerical complexity issues of the respective algorithmic solutions. Nevertheless, the findings of this research project provided theoretical limits of communication scenarios which have been unknown beforehand and which might be helpful to understand the remaining gap between practically and theoretically achievable system performances. Additionally, the derived results on global optimality of coordinated beamforming in wireless cellular communication systems provide important hints about potential transmission strategies for communications in interference limited networks.

Publications

• Utility maximization strategies in the multi-user MIMO downlink. In: Proceedings of the 1st International Workshop on Cross layer Design, Jinan, P.R. China, September 2007
  J. Brehmer and W. Utschick
  
• Nonconcave utility maximization in the OFDMA downlink. In: Proceedings of WiOpt’08 - International Symposium on Modeling and Optimization in Mobile, Ad Hoc, and Wireless Networks, Berlin, Germany, April 2008
  J. Brehmer and W. Utschick
  
• Time-sharing solutions in MIMO broadcast channel utility maximization. In: Proceedings of the International ITG Workshop on Smart Antennas, Darmstadt, Germany, February 2008
  J. Brehmer, Q. Bai, and W. Utschick
  
• Nonconcave utility maximization in the MIMO broadcast channel. EURASIP Journal on Advances in Signal Processing, Special Issue for Cross- Layer Design for the Physical, MAC, and Link Layer in Wireless Systems, 2009
  J. Brehmer and W. Utschick
  
• On proportional fairness in nonconvex wireless systems. In: Proceedings of the International ITG Workshop on Smart Antennas, Berlin, Germany, February 2009
  J. Brehmer and W. Utschick
  
• Utility maximization in the multi-user MISO downlink with linear precoding. In: Proceedings of the International Conference on Communications, Dresden, Germany, June 2009
  J. Brehmer and W. Utschick
  
• Optimal interference management in multi-antenna, multicell systems. In: Proceedings of the International Zürich Seminar on Communications, Zürich, Switzerland, March 2010
  W. Utschick and J. Brehmer
  
• Computing upper bounds for coordinated multipoint transmission. In: Proceedings of the International ITG Workshop on Smart Antennas, Aachen, Germany, February 2011
  W. Utschick and A. Brack
  
• Utility Maximization in Nonconvex Wireless Systems. Series on Foundations in Signal Processing, Communications and Networking, Vol. 5, Springer 2012
  J. Brehmer