Showing papers by "Mats Bengtsson published in 2021"
06 Jun 2021
TL;DR: In this paper, the authors proposed to apply deep unfolding to the weighted minimum mean square error (WMMSE) algorithm to provide a locally optimum solution to the otherwise NP-hard weighted sum rate maximization beamforming problem.
Abstract: The weighted minimum mean square error (WMMSE) algorithm was proposed to provide a locally optimum solution to the otherwise NP-hard weighted sum rate maximization beamforming problem, but it can still be prohibitively complex for real-time implementation. With the success of deep unfolding in trading off complexity and performance, we propose to apply deep unfolding to the WMMSE algorithm. With respect to traditional end-to-end learning, deep unfolding incorporates expert knowledge, with the benefits of immediate and well-grounded architecture selection, fewer trainable parameters, and better explainability. However, the classical formulation of the WMMSE algorithm given by Shi et al. is not amenable for deep unfolding due to matrix inversions, eigendecompositions, and bisection searches. Therefore, we present an alternative formulation that circumvents these operations. By means of simulations, we show that the deep unfolded WMMSE algorithm performs on par with the original WMMSE algorithm, at a lower computational load.
18 citations
TL;DR: Two mask-compliant spectral precoding schemes are proposed, which mitigate the resulting TxEVM seen at the receiver by capitalizing on the immanent degrees-of-freedom in (massive) MIMO systems and consequently improve the system-wide throughput.
Abstract: Although spectral precoding is a propitious technique to suppress out-of-band emissions, it has a detrimental impact on the system-wide throughput performance, notably, in high data-rate multiple-input multiple-output (MIMO) systems with orthogonal frequency division multiplexing (OFDM), because of (spatially-coloured) transmit error vector magnitude (TxEVM) emanating from spectral precoding. The first contribution of this paper is to propose two mask-compliant spectral precoding schemes, which mitigate the resulting TxEVM seen at the receiver by capitalizing on the immanent degrees-of-freedom in (massive) MIMO systems and consequently improve the system-wide throughput. Our second contribution is an introduction to a new and simple three-operator consensus alternating direction method of multipliers (ADMM) algorithm, referred to as TOP-ADMM, which decomposes a large-scale problem into easy-to-solve subproblems. We employ the proposed TOP-ADMM-based algorithm to solve the spectral precoding problems, which offer computational efficiency. Our third contribution presents substantial numerical results by using an NR release 15 compliant simulator. In case of perfect channel knowledge at the transmitter, the proposed methods render similar block error rate and throughput performance as without spectral precoding yet meeting out-of-band emission (OOBE) requirements at the transmitter. Further, no loss on the OOBE performance with a graceful degradation on the throughput is observed under channel uncertainty.
5 citations
TL;DR: A novel spectral precoding approach which constrains the EVM while complying with the mask requirements is proposed, and both proposed schemes outperform previously developed schemes in terms of important performance indicators such as block error rate and system-wide throughput.
Abstract: Spectral precoding is a promising technique to suppress out-of-band emissions and comply with leakage constraints over adjacent frequency channels and with mask requirements on the unwanted emissions. However, spectral precoding may distort the original data vector, which is formally expressed as the error vector magnitude (EVM) between the precoded and original data vectors. Notably, EVM has a deleterious impact on the performance of multiple-input multiple-output orthogonal frequency division multiplexing-based systems. In this paper we propose a novel spectral precoding approach which constrains the EVM while complying with the mask requirements. We first formulate and solve the EVM-unconstrained mask-compliant spectral precoding problem, which serves as a springboard to the design of two EVM-constrained spectral precoding schemes. The first scheme takes into account a wideband EVM-constraint which limits the average in-band distortion. The second scheme takes into account frequency-selective EVM-constraints, and consequently, limits the signal distortion at the subcarrier level. Numerical examples illustrate that both proposed schemes outperform previously developed schemes in terms of important performance indicators such as block error rate and system-wide throughput while complying with spectral mask and EVM constraints.
4 citations
06 Sep 2021
TL;DR: In this paper, an intelligent reflecting surface (IRS) is exploited for simultaneous wireless information and power transfer in a SWIPT scenario without the need for cabled links or channel state information.
Abstract: In this paper, intelligent reflecting surfaces (IRSs) are exploited for simultaneous wireless information and power transfer. In the considered setup, the IRS, illuminated by a power transmitter, provides power to integrated sensors through energy harvesting while simultaneously transmitting information through differential permutation-based coding. The problem of allocating the IRS elements to either information or power transmission is studied, highlighting the tradeoff between the system throughput and the harvested power. The conducted performance analysis emphasizes the suitability of using IRSs in a SWIPT scenario without the need for cabled links or channel state information.
Posted Content•
TL;DR: In this article, the tradeoffs between complexity and reliability for decoding large linear block codes were studied and artificial neural networks were used to predict the required order of an ordered statistics-based decoder to reduce the average complexity and hence the latency of the decoder.
Abstract: In this paper, we study the tradeoffs between complexity and reliability for decoding large linear block codes. We show that using artificial neural networks to predict the required order of an ordered statistics based decoder helps in reducing the average complexity and hence the latency of the decoder. We numerically validate the approach through Monte Carlo simulations.