《无线传感器网络节能优化策略研究》由会员分享，可在线阅读，更多相关《无线传感器网络节能优化策略研究（51页珍藏版）》请在金锄头文库上搜索。
1、合肥工业大学 硕士学位论文 无线传感器网络节能优化策略研究 姓名：毛沙 申请学位级别：硕士 专业：计算机应用技术 指导教师：唐昊 2011-04 I 无线传感器网络节能优化策略研究 无线传感器网络节能优化策略研究 摘 要 摘 要 无线传感器网络（Wireless Sensor Network, WSN）中的关键问题就是传感 器节点能量的有限性，节能高效地使用节点能量是 WSN 面临的一大挑战。 本文研究 WSN 中传感器节点能量高效控制问题，其优化目标是通过合理 选择节点的工作模式以及适应性调整节点发射功率和调制级，实现节点平均每 单位能耗的吞吐量最大。论文首先研究了衰落信道下高效的传输策略问
2、题。为 了避免节点在感知、接收以及发送数据时不必要的能耗，本文引入了动态功率 管理（Dynamic Power Management, DPM）技术。并把单个节点的跨层能量管 理问题建模成平均或折扣报酬 Markov 决策过程（Markov Decision Process, MDP） ， 运用强化学习理论中的 Q 学习算法， 结合模拟退火 （Simulated Annealing, SA）思想，求解出了节点高能效的控制策略。最后，论文通过仿真实验验证了 该算法的有效性。学习结果表明，节点能耗得到降低，节点的吞吐量也得到了 改善。 进一步，本文研究多节点同步通信问题，与点对点通信情况的不同点在
3、于 某个节点的链路质量要受其他节点决策的影响，因此，近似最优的平衡解也依 赖其他节点所采取的策略。本文采用分布式 Q 学习的方法来近似求解多节点通 信场景下的节能优化问题，实验结果表明该算法能获得较好的性能。 关键词: 关键词: 传感器节点、动态功率管理、跨层优化、Markov 决策过程、Q 学习 II Energy-Efficient Optimization Strategies for Wireless Sensor Network ABSTRACT Energy constraints in a wireless sensor network (WSN) are crucial iss
4、ues. Energy saving and energy efficiency for a sensor node are one of the key challenges in WSN. In this master dissertation the energy optimization problem for the sensor node is concerned, and the objective is to obtain the long-term average maximum throughput per energy consumption. First, we ana
5、lyze the energy conservation optimization problem in the background of energy efficient transmission strategy over fading channels. For less energy consumption on data sensing, receiving and sending we refer to a cross-layer mechanism which dynamically turns on or off different components, adjusts t
6、ransmit power and modulation level of the sensor node while maintaining required performance. Then, the problem is modified by introducing dynamic power management (DPM) technique and modeled as an average reward Markov decision process (MDP). Combined with simulated annealing (SA), Q learning algor
7、ithm is proposed to solve the energy conservation optimization problem with average performance criteria. Finally, the simulation results show that the approach in this dissertation is more efficient than adaptive transmission policy or DPM policy. As the energy consumption of the sensor node can be
8、 well balanced while the throughput doesnt decrease significantly. In contrast to the point-to-point communication, we consider transmitters simultaneously communicate to one receiver in multinode scenario. The channel link experienced by one node depends on the decision employed by other nodes in m
9、ultinode scenarios. Hence, the optimal equilibrium solution generally depends on the policy employed by the other nodes. We further extend the algorithm to solve the optimization problem in a multinode scenario by distribution Q learning. The simulation results show that the algorithm can achieve go
10、od performance. Keywords: Sensor node, Dynamic Power Management (DPM), Cross-Layer Optimization, Markov decision process (MDP), Q-learning VI 插图清单插图清单 图 1.1 无线传感器网络系统架构 .1 图 1.2 无线传感器网络协议栈层次.3 图 1.3 传感器节点构造.6 图 1.4 传感器节点的能耗情况.6 图 3.1 有限状态 Markov 信道 .21 图 3.2 不同模式间转换图 .22 图 3.3 传感器节点的结构 .23 图 3.4 =0.
11、9，初态 0 (0,1)X =下优化结果.27 图 3.5 平均准则下优化结果.27 图 3.6 =0.9，初态 0 (0,1)X =下优化结果.28 图 3.7 平均准则下下优化结果.28 图 3.8 Cross Layer 优化策略 .29 图 3.9 不同包到达率下的能量效率变化情况.29 图 3.10 另一参数下的能量效率变化情况.30 图 3.11 不同包到达率下的吞吐量变化情况.30 图 3.12 不同包到达率下的能量消耗情况.31 图 3.13 节点工作模式使用频率.31 图 4.1 多节点通信情形.33 图 4.2 网络结构图.35 图 4.3 多节点的学习优化结果.36 图 4.4 节点 2 和 3 的学习优化结果.
