Abstract: A cross-layer transmission control scheme based on Cognitive Radio（CR） for average power minimization under the constraints of collision with licensed user and buffer state is addressed.The scheme is formulated by Constrained Markov Decision Process（CMDP），which is solved by Linearly Programming（LP）.State aggregation is employed to solve the problem in the case that channel state information cannot be achieved in CR networks.To reduce the complexity of LP，state aggregation and action set reduction are utilized to cope with the curse of dimensionality.According to the simulation results，the scheme can afford average power minimization under constraints，and complexity reduced solution almost does not reduce performance of the scheme.

Key words: Cognitive Radio（CR）, energy-efficient, cross-layer design, transmission control, Constrained Markov Decision Process（CMDP）

摘要： 提出了一种基于认知无线电（CR）技术的跨层传输控制方案，该方案在满足与主用户冲突率约束及缓存器状态约束的前提下，最小化系统平均功率消耗。该方案被建模为约束马尔可夫决策过程（CMDP），并采用线性规划（LP）求解此CMDP。引入状态聚合来解决CR网络中信道状态信息在某些情况下不可获取的问题。为降低LP的求解复杂度，通过状态聚合和行动集缩减解决CMDP的维灾问题。仿真结果表明，方案能够在满足约束的条件下最小化功率消耗，而且低复杂度的求解方法对该方案性能的影响很小。

关键词: 认知无线电, 能效, 跨层设计, 传输控制, 约束马尔可夫决策过程