Two kinds of sensors will be discussed in some sort of wireless sensor networks (WSNs). One is named a normal sensor (called A_nodes) with fixed initial energy, which can get perception data from the surrounding environment and have functions of storage and forwarding. The other is named relay sensor (called B_node) with sufficient energy, which only can store data and forward data. Cluster heads (called A_heads) are chosen among A_nodes by probability mode to create clusters first. Local adjustments would be done inter-cluster. Then the sink is selected as the root and a backbone shortest path tree with hops limited is built dynamically from A_heads, B_Nodes and sink. Inside the backbone shortest path aggregation tree, two steps are done. The first step, select some B_node as cluster head (denoted as B_node) and make local adjustment intra-clusters if it is possible. The second step, adjust the shortest path aggregation tree. By using both particle swarm optimization (PSO) and ant colony optimization (ACO), a traveling salesman problem (TSP) cycle with shorter path length is obtained (this method is noted as TSP_PSO_ACO). This TSP cycle consists of all nodes in the aggregation backbone tree. Along this cycle, energy is offered to nodes except sink. The above strategy would be discussed for its feasibility and time complexity. The experimental results imply that the energy consumption is reduced by using local adjustment. The relative failure rate of nodes can be reduced in the later stage of WSN survival. Offering energy to a small number of nodes several times has no obvious impact on the overall energy consumption and survival time of WSN. The case with some small-sized obstacles in WSN area is also discussed. There are little affections on TSP cycles both from results on theoretical analysis and simulation if there are some obstacles with small size in WSN area.

我们将讨论某种无线传感器网络 (WSN) 中的两种传感器。一种称为普通传感器（称为A_nodes），具有固定的初始能量，可以从周围环境获取感知数据，并具有存储和转发功能。另一种称为中继传感器（称为B_node），具有足够的能量，只能存储数据和转发数据。通过概率模式在 A_node 中选择簇头（称为 A_heads）来首先创建簇。本地调整将在集群间进行。然后选择接收器作为根，并从 A_heads、B_Nodes 和接收器动态构建跳数有限的骨干最短路径树。在主干最短路径聚合树内部，完成了两个步骤。第一步，选择某个B_node作为簇头（记为B_node），如果可能的话，在簇内进行局部调整。第二步，调整最短路径聚合树。通过同时使用粒子群优化（PSO）和蚁群优化（ACO），获得了路径长度较短的旅行商问题（TSP）循环（该方法记为TSP_PSO_ACO）。该TSP循环由聚合骨干树中的所有节点组成。沿着这个循环，能量被提供给除汇点之外的节点。将讨论上述策略的可行性和时间复杂度。实验结果表明，通过局部调节可以降低能耗。在WSN生存的后期可以降低节点的相对故障率。多次向少数节点提供能量对WSN整体能耗和生存时间没有明显影响。还讨论了无线传感器网络区域中存在一些小尺寸障碍物的情况。从理论分析和仿真结果来看，如果WSN区域内存在一些小尺寸障碍物，对TSP周期的影响很小。