研究无线网络下行链路中平均功率约束下联合跨层多用户平均时延最小化问题,基站根据不同用户的信道增益和队列长度信息,动态调整发射功率和速率,以实现系统平均时延最小化,将该优化问题转化为非约束马尔可夫动态决策过程,证明了在最优策略下系统平均时延和功率具有凸关系,仿真结果表明这种关系成立,考虑到原优化问题的特性和求解的复杂度,进而将该优化问题转化为凸优化问题,并给出了实际的求解算法和仿真结果。
In order to minimize average delay of wireless network in the downlink environment under average power constraint, base station allocates transmission rate and power to individual users depending upon the buffer lengths and the channel of each user. We first formulate it as a constrained Markov decision process (MDP) and then convert it to an unconstrained one. Next the convexity property of the power-delay function of the optimal control strategy is given and verified by simulating results. Considering the huge complexity and the characteristic of the problem, we further transform it to a convex optimization problem and a practical algorithm is presented.