Robust Multi-Objective Optimization for EE-SE Tradeoff in D2D Communications Underlaying Heterogeneous Networks

Abstract

In this paper, we concentrate on the robust multiobjective optimization (MOO) for the tradeoff between energy efficiency (EE) and spectral efficiency (SE) in device-to-device (D2D) communications underlaying heterogeneous networks (HetNets). Different from traditional resource optimization, we focus on finding robust Pareto optimal solutions for spectrum allocation and power coordination in D2D communications underlaying HetNets with the consideration of interference channel uncertainties. The problem is formulated as an uncertain MOO problem to maximize EE and SE of cellular users (CUs) simultaneously while guaranteeing the minimum rate requirements of both CUs and D2D pairs. With the aid of ε-constraint method and strict robustness, we propose a general framework to transform the uncertain MOO problem into a deterministic single-objective optimization problem. As exponential computational complexity is required to solve this highly non-convex problem, the power coordination and the spectrum allocation problems are solved separately, and an effective two-stage iterative algorithm is developed. Finally, simulation results validate that our proposed robust scheme converges fast and significantly outperforms the non-robust scheme in terms of the effective EE-SE tradeoff and the quality of service satisfying probability of D2D pairs.

Resource Allocation with a Rate Guarantee Constraint in Device-to-Device Underlaid Cellular Networks

“…In this paper, we assume that the CSI of all possible combinations of the links is available at the BS, which may require indispensable CSI feedback. However, this assumption has been widely used in previous literature for performance evaluation [7][8][9][11][12][13][14][15][16][17][18][19]. Figure 1b illustrates a channel assignment scheme, where each channel with a fixed channel bandwidth is pre-assigned to one CU before the channel assignment.…”

“…Therefore, interference management must be applied to effectively handle the interference, and to guarantee a certain degree of quality of service (QoS). In the literature, various power control and resource allocation schemes have been proposed to guarantee QoS requirements in terms of signal-to-interference-plus-noise ratio (SINR) or minimum required data rate [8][9][10][11][12][13][14][15][16][17]. In [8], a power control scheme was studied to maximize the sum of the achievable data rates of a CU-DU pair.…”

“…Joint power control and resource allocation schemes were proposed in [11][12][13][14][15][16][17]. A single DU is assigned to only one channel, and vice versa in [11,12], whereas multiple DUs are allowed to be assigned to the same channel in [13,14]. In addition, a DU is allowed to be assigned to multiple channels in [15][16][17].…”

“…In most previous sum rate or energy efficiency maximization schemes [8][9][10][11][12][13][14][15][16][17][18][19][20][21][22], even though SINR or minimum data rate per UE was considered as a constraint, the individual data rate may not be guaranteed with a fixed amount of channel bandwidth. When the received SINR is low due to high interference and maximum transmit power limitation, the fixed channel bandwidth may not be wide enough to guarantee the target rate.…”

Outage Constrained Robust Beamforming Design for SWIPT-Enabled Cooperative NOMA System

“…For instance, a cooperative SWIPT NOMA protocol was investigated in [6], in which near NOMA users act as relays to help far NOMA users with the energy from EH instead of draining their own batteries. In [7], considering a cooperative multiple-input single-output (MISO) SWIPT NOMA system, the authors maximized the data rate of strong user 1 by jointly optimizing the power splitting (PS) ratio and the beamforming precoding vectors while guaranteeing the quality of service (QoS) demand of the weak user 2 . However, all these studies assume that the BS has perfect knowledge of channel state information (CSI), which is quite difficult in reality due to channel estimation errors, quantization errors and feedback delay [8].…”