Deep Learning-Based Power Control Scheme With Partial Channel Information in Overlay Device-to-Device Communication Systems

Abstract

In the overlay device-to-device (D2D) communication systems, transmit power control is critical to better manage interference, so that the sum rate is maximized. Such power control for sumrate optimization is NP-hard, which is typically tackled by iterative algorithms such as weighted minimum mean square error (WMMSE) method. However, the iterative power control schemes inherently incur high complexity and excessive latency. To overcome the limitations, we propose a deep learning-based power control scheme with reduced complexity and latency, where partial and outdated channel state information (CSI) is considered. Using a deep neural network (DNN)-based approach, we formulate an optimization problem to maximize the spectral efficiency under the constraints of user fairness and energy efficiency, where the DNN-based method is based on unsupervised learning with no label data generation process. In addition, a CSI reporting method based on the channel-to-interference power ratio is proposed for partial CSI feedback, which considerably reduces the feedback overhead. Through simulations, we show the results of the spectral efficiency, energy efficiency, and fairness performance for various topographical sizes and channel correlation coefficients. Also, it is shown that the proposed scheme achieves better spectral efficiency and energy efficiency than the WMMSE scheme even when it uses a small amount of CSI feedback.INDEX TERMS Deep neural network, transmit power control, spectral efficiency, energy efficiency, index of fairness, partial channel state information.

A Self-Regulating Power-Control Scheme Using Reinforcement Learning for D2D Communication Networks

“…On the other hand, new power-control schemes based on deep learning for D2D networks have been proposed to overcome the limitations of the conventional schemes such as optimization of threshold values, computational complexity, or signaling overhead [11][12][13][14][15][16]. Deep reinforcement learning (DRL)-based power control schemes for D2D communications underlying cellular networks were investigated [11][12][13].…”

“…Although underlay D2D communications can significantly enhance overall spectral efficiency, the quality of cellular communications cannot be tightly guaranteed because of the cross-interference caused by D2D communications. Thus, deep-learning-based power control schemes for overlay D2D communication systems were proposed in [14][15][16]. Cellular and D2D users utilize different radio resources that are orthogonal to each other in order to guarantee the quality of cellular communications by avoiding the cross-interference.…”

“…However, this proposed scheme does not outperform the conventional algorithm based on fractional programming, and it requires global channel state information, although it is outdated [14]. A deep-learning-based power control scheme using partial and outdated channel state information was proposed in [15]. This proposed scheme achieved better spectral efficiency and energy efficiency with reduced complexity and latency compared to the iterative conventional power allocation scheme.…”

“…Despite intensive investigations on the power control problem in D2D communication networks, the closed-form solutions of general power control problems to maximize the sum-rate of D2D communication networks in which multiple D2D links share the same radio resource are not available, as they are typically NP-hard. As an alternative, new power-control schemes have prepared to overcome the limitations of conventional schemes using deep learning have been proposed [11][12][13][14][15][16]. However, they unfortunately do not allow each D2D user to autonomously determine its transmission power level because cellular base stations (BSs) play a key role in coordinating the transmission power levels of cellular and D2D users or each D2D pair needs to collect not only local information that can be obtained directly by the transmitter or receiver of a D2D pair but also non-local information that can be obtained from neighboring D2D pairs, thereby causing extra signaling overhead.…”

A Study on the Impact of Integrating Reinforcement Learning for Channel Prediction and Power Allocation Scheme in MISO-NOMA System

“…To get over limitations raised when standard iterative power control techniques are utilized, such as high complexity and unnecessary latency, the work in [ 13 ] introduced a deep learning framework to manage these issues. In the presented structure, the outdated and partial CSI is exploited, and a Deep Neural Network (DNN) framework is created to construct an optimization problem to boost the spectral efficiency in device-to-device communication systems.…”