Search

This overview paper discusses the framework of sequential fractional programming for energy efficiency maximization in future 5G networks. One of the main features of future systems will be the presence of severe multi-user interference and the need of improved energy efficiency compared to present systems. However, present approaches to energy efficiency maximization, which are based on the theory of fractional programming, result in an exponential complexity in interference-limited networks. In this context, the work shows how to extend available fractional programming approaches to obtain radio resource allocations enjoying strong optimality properties, while at the same time requiring an affordable complexity to be computed. The resulting framework is termed sequential fractional programming, and several examples of its applications to leading 5G candidate technologies are discussed in detail. [ABSTRACT FROM AUTHOR]

We study the optimal transmission strategy of a multiple-inputsingle-output wireless communication link. The receiver has perfectchannel state information while the transmitter hasonly long-term channel state information in terms of the channelcovariance matrix. It was recently shown that the optimal eigenvectors of the transmitcovariance matrix correspond with the eigenvalues of the channelcovariance matrix. However, the optimal eigenvalues are difficult tocompute. We study the properties of these optimal capacity achieving eigenvalues, and present a necessary and sufficient condition for theoptimal eigenvalues of the transmit covariance matrix. Furthermore, we develop a necessary and sufficient condition forachieving capacity when transmitting in all directions. We compare thecapacity gain of an optimal diversity system with a system which works with beamforming, and we derive an upperbound. We answer the main questions regarding the system design using the developed results. Additionally, we show inwhich way the multiplexing gain can be computed in case the channel covariancematrix is given. We compute the maximum number of required paralleldata streams, and we define a multiplexing function inorder to obtain a measure for the available multiplexinggain. Furthermore, we show that the capacity gain is small considering theadditional complexity at the receiver. We illustrate allresults by numerical simulations.