On Performance Trade-offs In Green Wireless Communications Networks

The explosive growth in mobile traffic amounts and the Internet-of-Everything access re-quirements have being brought unprecedented development opportunities to wireless com-munication networks. Unfortunately, the concomitant disaster is the huge energy consump-tion (EC) and greenhouse emission, both of which are skyrocketing at a surprising speed. On the one hand, the enormous cost bill resulted from EC has become the biggest obstacles to the continuous development of mobile communication networks. On the other hand, the restricted battery capacity and the extremely limited battery duration are the fatal shackles to the ultimate traffic experience of users. As a powerful and promising solution, energy-efficient communications are the only choice. For this reason, green communications gain favors in both industry and academia upon proposition and are becoming an inevitable trend for future wireless network design.Aiming at resource-saving and eco-friendly design, green communications attempt to re-duce EC and improve energy efficiency (EE) from component, link, and network levels by adopting techniques or methods such as effective power amplifier development, flexible net-work deployment, energy-efficient resource management, intelligent supply-demand match, and new-type network architectures. Existing researches have shown that the improvement in EE usually degrades the performance of other technical metrics such as spectral efficiency (SE), throughput, delay, and fairness. Hence, to provide theoretical and technical supports for green operations such as performance balance, network deployment, and resource man-agement, lots of researches have focused on exploring the trade-offs between/among perfor-mance metrics such as EE, SE, throughput, delay, and fairness, revealing their coupling or restrictive relationships, and further devising on-demand control algorithms.Studies on performance trade-offs between/among different metrics constitute a series of key scientific problems in the researches of green communications. This dissertation devotes to investigating the three typical among them, namely performance trade-offs between EE and SE, between EE and delay, and between EE and fairness. Specifically, the main contents and contributions are summarized as follows.1. We quantitatively reveal the fundamental performance trade-off between EE and delay. We take orthogonal wireless networks as the system scenario, and formulate the problem as a stochastic optimization model, which optimizes the system EE subject to network stability and the average and peak transmit power constraints. By adopting fractional programming theory and Lyapunov optimization technique, a general and effective algorithm, referred to as the conTRollable Algorithm for Delay-EE tradeOFF (TRADEOFF), is proposed to solve the problem. The theoretical analysis shows that the developed formulation and algorithm can strike a flexible balance between EE and delay. Most importantly, we quantitatively derive the EE-delay trade-off as [O (1/V), O (V)] with V being a control parameter. Sim-ulation results validate the theoretical analysis on the EE-delay trade-off, as well as show the adaptation of the TRADEOFF to random and time-varying traffic arrivals and wireless channels.2. We quantitatively reveal the fundamental performance trade-off between throughput and delay with guaranteed EE. We adopt two admission control schemes, referred to as the first-out and first-in schemes, to explore the maximum network throughput. We then formulate it as two stochastic optimization problems, respectively aiming at throughput maximization (in the first-out scheme) and dropping rate minimization (in the first-in scheme) subject to re-quirement of EE (RoE), stability, admission control, and transmit power. To solve the prob-lems, the EE-Guaranteed algorithm for throUghput-delAy tRaDeoff (eGuard), respectively called eGuard-I and eGuard-II in the first-out and first-in schemes, is devised. Moreover, with guaranteed RoE, we theoretically show that the eGuard (I and II) can not only push the throughput arbitrarily close to the optimal with trade-offs in delay, but also quantitatively control the throughput-delay performance on demand. Simulation results consolidate the theoretical analysis and particularly show the pros and cons of the two schemes.3. We qualitatively reveal the fundamental performance trade-off between EE and fair-ness. We take Orthogonal Frequency Division Multiple Access (OFDMA) networks as the system scenario, and investigate the max-min EE-optimal problem (MEP) to ensure fairness among links in terms of EE. For this aim, we maximize the EE of the worst-case link sub-ject to the rate requirements, transmit power, and subcarrier assignment constraints. Due to the nonsmooth, nonconvex, and mixed combinatorial features of the formulation, we focus on the low-complexity suboptimal algorithms design. Leveraging the generalized fraction-al programming theory and Lagrangian dual decomposition, we first propose an effectively iterative algorithm to solve the problem. To further reduce the computational cost, we also devise algorithms by separating subcarrier assignment and power allocation, where we first assign subcarriers through an equal power distribution. We then fast and optimally solve the resultant power allocation problem by exploiting its special structure, which only requires tackling at most two simpler subproblems. Simulation results exhibit the convergence of the algorithms, verify the guaranteed EE fairness among links, as well as provide extensive comparisons between the MEP and the existing classical algorithms,namely the network EE-optimal problems (NEPs), the rate adaptive, the margin adaptive, and the utility adap-tive. Moreover, simulations reveal that there is a new trade-off between the network EE and fairness, just as the extensively explored trade-off between the system sum-rate and fairness.