Multi-Objective Decision Based Resource Reconfiguration Algorithm For Two-Way Transmission Of Heterogeneous Networks

The heterogeneity of wireless networks is a trend of the development of wireless commu-nication technologies. It is thus a key problem of heterogeneous networks to design reason-able resource optimization algorithms in order to maximize resource utilization and provide QoS(Quality of Service) guarantee as much as possible. This thesis focuses on the average delay and total transmit power minimization problem for heterogeneous networks. Two algo-rithms are proposed, which are Lyapunov optimization based power allocation algorithm for heterogeneous coordinated multi-point (CoMP) transmission systems and sub-channel shar-ing and power control algorithm for Device-to-Device (D2D) communications underlaying cellular networks.The main contents of this thesis are described as follows.(1) Research background for heterogeneous networks is introduced and available liter-atures concerning resource management for heterogeneous networks are reviewed. Sub-sequently, necessary theoretical basis for the minimization of delay and transmit power is presented as a preparation for the following studies.(2) A power allocation algorithm aiming to minimize the average delay and total transmit power is proposed for delay-sensitive heterogeneous CoMP system. The dynamic multi-objective problem is first transformed into a single objective problem through the main ob-jective method. And then taking advantage of the Lyapunov optimization theory, power is allocated to each user by minimizing the upper bound of drift-plus-penalty expression by us-ing Lagrangian dual decomposition method in each time slot. Simulation results prove that the proposed algorithm has better performance of delay and transmit power compared with greedy strategy and proportional fairness algorithm. The influence of the penalty factor on delay and power performance is also analyzed.(3) A delay and transmit power minimization based sub-channel sharing and power con-trol scheme for D2D communications underlaying cellular networks is proposed. The lin-ear weighted method is used to transform the formulated dynamic multi-objective problem, which can be further modeled as a Markov decision problem. The post-decision states are introduced and the value function of post-decision states is approximated based on linear model. Value functions are estimated iteratively through TD(0) scheme, and the policy is updated by the policy iteration. Simulation results show that the proposed algorithm con-verges within finite number of iterations and has better performance of delay and transmit power.