Research On Energy-efficient Resource Allocation In Next Generation Wireless Networks

With the large-scale commercialization for 4G,5G has become a research hotspot in both academia and industry since 2014.The two major perfor-mance goals,"larger capacity" and "lower latency",are permeated through-out the previous generations of mobile communication systems.However,they are endowed with new connotations in the era of 5G.As continuous de-velopment of the next generation wireless networks,latency/energy-sensitive and computation-intensive mobile applications(e.g.,augmented/virtual reali-ty,high-definition video,online gaming and massive low-power Internet-of-Things applications),will become an essential part of the future society.These demands in 5G require higher transmission rate,lower latency,higher ener-gy efficiency and smarter network management.In addition to the traditional transmission peak rate,user mobility and spectral efficiency,energy and com-putation efficiency have gradually become significant indicators for the design and optimization of 5G as well.Aiming at energy efficiency,this thesis stud-ies the flow and joint optimization of communication,computation and energy resources in 5G networks,so as to improve the efficiency of the network.Specif-ically,the main research content and contributions are given as follows.First,consider the joint optimization of communication-and-energy re-sources.1)A projected gradient based game theoretic approach for multi-user pow-er control:Consider the joint optimization of spectrum and energy resources.Unlike most existing studies which simplify the problem by considering on-ly a single-licensed user or channel,we investigate a more realistic scenario where multi-licensed users share multi-channels with multi-unlicensed users.We formulate the power control problem as a non-cooperative game with cou-pled constraints,where the Pareto optimality and achievable total throughput can be obtained by a Nash equilibrium(NE)solution.To achieve NE of the game,we first propose a projected gradient based dynamic model whose equi-librium points are equivalent to the NE of the original game,and then derive a centralized algorithm to solve the problem.Simulation results demonstrate that the convergence and effectiveness of our proposed algorithm,as well as the robustness of our proposed solution as the network size increases.2)A spectrum access policy with energy harvesting:Consider a single-user spectrum sharing network system with ambient-energy harvesting.Specifical-ly,in order to seek the optimal sensing and access policy,a hybrid spectrum sharing strategy combined with the opportunistic spectrum access(OSA)and underlay scheme is proposed for improving the achievable throughput of un-licensed users.Moreover,we model the optimal sensing and access policy as a finite-horizon Markov decision process(MDP)taking into account both the peak interference power constraint at licensed users and the energy causality constraint at unlicensed users.Then,an iterative optimization algorithm based on Bellman Equation is proposed,numerical results demonstrate that the pro-posed hybrid access scheme achieves better performance than that of the OSA system with different network parameters.Next,consider the joint optimization of communication-computation-and-energy resources.3)Stochastic control of computation offloading to a helper with dynamical-ly loaded CPU:We design energy-efficient control policies in a computation of-floading system with a random channel and a helper with a dynamically loaded CPU(due to the primary service).Specifically,the policy adopted by the helper aims at determining the sizes of offloaded and locally-computed data for a giv-en task in different slots such that the total energy consumption for transmission and local CPU is minimized under a task-deadline constraint.As the result,the polices endow an offloading user robustness against channel-and-helper ran-domness besides balancing offloading and local computing.By modeling the channel and helper-CPU as Markov chains,the problem of offloading control is converted into an MDP.Though dynamic programming(DP)for numerical-ly solving the problem does not yield the optimal policies in closed form,we leverage the procedure to quantify the optimal policy structure and apply the result to design optimal or sub-optimal policies.For different cases ranging from zero to large buffers,the low-complexity of the policies overcomes the"curse-of-dimensionality" in DP arising from joint consideration of channel,helper CPU and buffer states.4)Stochastic computation offloading based on constant local-CPU fre-quency:On the basis of the system model studied above,we consider an exten-sion that the local-CPU frequency is constant.Similar to the case of adjustable local-CPU,the optimal computation offloading control problem that minimizes the user-energy consumption is modeled as a finite-horizon MDP.As this prob-lem introduces two new constraints,which make the closed-form optimal poli-cies difficult to obtain.Therefore,we propose a heuristic sub-optimal computa-tion offloading control policy based on relaxation-and-truncation.Simulation results shows the close-to-optimal performance of the proposed policy.Finally,we briefly summarize the research contents of this thesis,and the future promising research directions are prospected as well.