Spectrum-efficient And Energy-efficient Power Management In Wireless Network

Spectrum scarcity and energy conservation are two vital research topics in wireless communicatons. In the paper, we investigate the power management with high spectrum efficiency and energy efficiency for wireless networks.For the purpose of improving the spectrum efficiency, we consider the joint power control of primary user (PU) and secondary user (SU) under underlay mode in OFMD-based cognitive radio network. Applying the Stackelberg game theory, we model and analyze the hierarchical joint power allocation for both PU and SU. As the PU has priority over SU on spectrum usage, we formulate the PU as leader and SU as follower. The strategy for each user (PU and SU) is power allocation over sub-carrirer, and the utility is its maximal transmission rate. We consider two constraints:the total power constraint and the interference-to-signal ratio (ISR) constraint, in which the ratio between the accumulated interference and the received signal power at each PU should not exceed a certain threshold. According to these constraints, the feasible strategy set can be obtained.Theoretical analysis is performed on the forumulated Stackelberg game. Specificly, for the lower game of SUs given PU’s power allocaton, we study the existence of the Nash Equilibrium (NE), and give a sufficient condition for the uniqueness of the NE. Moreover, an iterative algorithm, which converges to the NE, is presented for the general channel condition, and the closed-form solutions for the NE are derived in the perfectly symmetric channel. Regarding the upper problem, when there is only one PU, the upper game can be written as an optimization problem that contains the lower non-cooperative sub-game. At first, based on the theoretical results, we propose two kinds of power allocation algorithms the one-PU case. Considering availability of the private infomation for the SUs at the PU, two scenarios are investigated. An analytical hierarchic power allocation method is proposed when the PU can acquire the additional information to anticipate SUs’reaction. The analytical algorithm has two steps:1) The PU optimizes its power allocation with considering the SUs’reaction to its action.2) The SUs allocate the power according to the NE of the sub-game given PU’s optimal power allocation. Furthermore, we design two distributed iterative algorithms for the general channel even when private information of the SUs is unavailable at the PU. The first iterative algorithm has a guaranteed convergence performance, and the second iterative algorithm employs asynchronous power update to improve time efficiency. Finally, we extend to the multi-PU and multi-SU scenario, and a distributed iterative algorithm is presented.Cooperative transmission in vehicular networks is studied by using coalitional game and pricing to imprive the effective spectrm efficiency. By cooperation among vehicles and cooperation between vehicle and road side unit (RSU), the data collision can be mitigated and the transmission accuracy and throughout can be improved. Then the effective spectrm efficiency can be enhanced. Vehicles cooperate to avoid interfering transmissions by scheduling the transmissions of the vehicles in each coalition. Second, a RSU can join some coalition to cooperate the transmissions of the vehicles in that coalition. Although there are costs in receiving and forwarding the vehicles’ signals, the RSUs could achieve revenues by charging the vehicles. We analyze the conditions that forming coalition is profitable for vehicles and RSUs (e.g., there is at least one vehicle in each coalition). Morover, we study the stablilty of coation, and a sufficient condtion for nonempty of the core is derived.Energy efficiency is an essential notation in green communicatons. Energy efficient wireless communication is important for energy conservation. In this paper, we consider the power allocation of the physical layer and the buffer delay of the upper application layer in energy harvesting green networks. The data are generated in the application layer, and placed in a buffer at the transmitter. The transmitter periodically removes some data from the buffer, and transmits the data to the receiver. Meanwhile, the transmitter allocates some power from the storage battery of the harvested energy, and the rest power will be supplied by the power grid. We investigate the delay-optimal power allocation under the constraint on the average grid power. The purpose is to maximize the power utility so as to improve energy efficiency. When the data arrival process and the harvested energy arrival process are the Markov processes and the channel process is a Markov chain, we formulate the problem as a constrained Markov decision process (MDP) problem, in which the state and action are defined. The state includes the queue state (i.e., the queue length in the buffer), the battery state (i.e., the stored energy in the battery), the channel state, the data arrival, and the harvested energy arrival. The action consists of the transmission rate and the power allocation from the battery. That is to say, the policy is two-dimensional. We prove that the optimal rate and optimal battery power allocation can be obtained successively for the infinite battery capacity. In addition, we derive some structural properties of the optimal rate. For finte capacity case, we verify the existence of the stationary policy. Using the Lagrangian methodology, the constrained MDP can be relaxed to an unconstrained problem (UP), which is an average cost MDP. We prove that the optimal solution of the UP with a certain Lagrangian multiplier is the optimal solution of the original constrained MDP. Meanwhile, the average cost MDP (i.e., UP) can be analyzed by converting to the discount cost MDP.We derive two necessary conditions for the optimal policy. Under certain conditions, the policy that serving nothing and allocating no energy from the battery is optimal. We also prove that serving everything combined with allocating the minimal of the total required power and total energy in the battery are optimal under other certain conditions. We analyze the relations between the transmission rate and the power allocation from the battery. We find that the transmission rate is dominant, and we propose a conjecture that the original problem can be reduced to a MDP problem with the policy to be the transmission rate only. In the end, on the basis of the theoretical investigation, we propose two specific deterministic policies:radical policy and conservative policy.In the last part of the paper, we extend the MDP framework to the charging scheduling of electric vehicles so as to improve the energy efficieny at the charging station. The charging station has multiple charging points, and it also is equipped with renewable energy generation devices. To garanttee relatively stable services, the charging station can also purchase grid power. There is a contrationt on average cost of grid power purchasement at the charge station. Once an EV arrives at the charging station, it waits in a queue before charging. In each period, the charging station chooses some EVs from the head of the queue for charging. Meanwhile, the station also determines how much energy is supplied from the storage battery (the rest of the required energy is supplied from the power grid). Under the uncertainty of the EV arrival, the renewable energy, the grid power price, and the charging energy of each EV, we study the mean delay optimal scheduling with the average cost constraint. Since the amount of charging energy (i.e., the number of energy blocks to charge) for each EV is random, the scheduling problem is very challenging. We propose queue mapping method to solve the difficulty. We map the EV queue to a charge demand queue. Furthermore, we prove that the minimization of the average EV queue length is equivalent to the minimization of the average charge demand queue length. We find that the minimization of the average charge demand queue length can be studied under the MDP framework when EV arrival, the renewable energy arrival, the grid power price variation are all Markovian. For conciseness, we give a special case of problemthe problem and investigate this relatively simplified problem to show the solving process. General cases can be analyzed through similar solving process.We consider the conventional power grid and smart grid, respectively. In conventional grid, a battery is required to store the renewable energy before its usage. The finite battery capacity will have great effects on the policy. In contrast, the renewable energy can be stored in the smart grid and no storage battery is needed. Under same policy, the queue length in smart grid is no longer than in conventional grid. It is derived that the optimal charging number of EVs and optimal renewable power allocation can be obtained successively in smart grid. Some structural properties of the optimal charging number of EVs are also obtained. In conventional grid, it is verified that there is a stationary dertministic policy that is optimal. The contraind MDP can be sloved by its Largrange relaxed unconstrained MDP and its corresponding discount MDP thereafter. We derive two necessary conditions of the optimal policy. Moreover, we discuss the reduction of the two-dimensional policy to be the number of charging demands only in both conventional and smart grid and guess that greedy policy for renewable energy is optimal when we fix the charging number of EVs in a period.