Aiming at the problems of low coverage, high energy consumption and long delay of static Wireless Sensor Networks(WSNs) that are deployed randomly, a cellular perceptron coverage algorithm for WSNs based on triangulation is proposed. The algorithm applies improved Delaunay triangulation to achieve unique division of the network area. Then, it uses the cellular idea to confirm the neighborhood relationship between nodes and realizes network topology control by setting the node perception radius. Finally, the remaining energy of the node, the time delay and the historical forwarding probability are trained as the input data of the machine learning perceptron to find the optimal data forwarding communication path. The algorithm combines network coverage with data transmission. Compared with other algorithms, the coverage is increased by 13%~34%, the node energy consumption is reduced by 2.25~2.5 J, the network life cycle is prolonged by 25%, and the network delay is reduced by 0.25~1.18 s.

Under the existing rule inference mechanism, the amount of real-time feature calculation during rule matching caused by a large amount of sensor data reduces the inference real-time performance. At the same time, the limited memory resources of edge devices cannot cope with such a huge amount of data. For this reason, this thesis designs the Data Pre-Deployment Scheme(DPDS). By utilizing the rule network and Light-weight Characteristic Table(LCT) obtained by the rule analysis and preprocessing module, this scheme enables the rule network to directly reference the characteristic values in the LCT during inference without real-time feature calculations, which significantly improves efficiency and real-time and greatly reduces the memory usage of the inference process. The experimental results show that even in the case of a large amount of data and rules, DPDS still has high time efficiency and space efficiency.

Aiming at the problem that the large error of underwater localization is caused by variable sound velocity and obstacles, a parabolic model ranging localization was proposed in this paper. Firstly, based on AoA(Angle of Arrival), a ranging method with parabolic models that is suitable for the curved sound ray was proposed. Then, based on ToA(Time of Arrival), the ranging recognition method of the non-ideal path and the ranging correction method of the ideal path were proposed respectively. Finally, the projection method and the least square method were used to complete the localization. The simulation results show that the algorithm can effectively reduce the localization error in three aspects, curved sound ray, path recognition, and ranging correction.

To improve the locating accuracy for wireless sensor networks (WSN) in the mixed propagation environment of non-line-of-sight (NLOS) and line-of-sight (LOS) during passive localization,an algorithm named residual test based on arc-edged convex hull (RTAC) is proposed. RTAC algorithm uses the distribution characteristics of ranging residuals of sensors to construct an offset circle model reflecting the distribution of residual points in the polar coordinate system, and uses the minimal arc-edged convex hull to group and identify the sensors in order to realize the LOS sensor identification in the network. The simulation results demonstrate that RTAC algorithm can realize the correct identification of LOS sensors under a lower computational complexity, and can obtain better location performance. RTAC algorithm is an efficient algorithm for LOS sensor identification in mixed propagation environment.

RF energy harvesting wireless sensor network (RFEH-WSN) consists of dedicated energy transmitter (ET) and sensor nodes with RF energy harvesting technology. The RFEH-WSN solves the problems of the battery replacement and node energy depletion, which makes it has more advantages in the future application. How to place ET effectively with minimum energy consumption and maximize overall charging utility is one fundamental issue in RFEH-WSN. In this paper, a new multiple object model is proposed, and the optimization aims of the model are to minimize the charging time and to maximize the coverage. An approximate with low complex algorithm is proposed to solve this multi-object function by PSO optimizer, and from it an optimum pareto solution set is obtained. The simulation results show that the new methods can improve the charging efficiency obviously and satisfy the different demands for lots of application environments.

Data storage is a basic operation of data management in wireless sensor networks. In mobile low-duty-cycle sensor networks, due to the mobility of the nodes, each node needs to frequently update the set of its neighbor nodes, which making energy consumption of the node too large. At the same time, each node is sleeping in most of its time, and wakes up to work in only a small portion of time. This sleeping/working mode results in excessive communication delay for data backup. A fast data storage mechanism with low energy consumption is proposed. First, each source node performs piecewise linear fitting compression on its sensing data based on continuous time series. Then, the node calculates a reasonable number of compressed data backups based on an estimated failure probability and the size of its storage space. On this basis, a dynamic adaptive transmission protocol is designed. Experimental simulations show that this mechanism has lower energy consumption of transmission and lower communication delay compared with existing storage algorithms.

Designing accurate localization algorithms is a research focus in WSNs (Wireless Sensor Networks). Aiming at the problem that the large estimated distance error between nodes could result in inaccurate localization in Distance Vector-Hop (DV-Hop), we proposed an improved three-dimensional DV-Hop algorithm with continuous hop value in this paper. Firstly, the algorithm explored the relationship between the distance of neighbor nodes and the volume of intersecting spheres which is composed of the corresponding node locations and communication radius. In addition, the definition and parameter modification calculation formula of continuous hop value were put forward. Then we explored the influence of network environment on parameters and determined parameters value through simulation. The accurate continuous hop value was used instead of the hop in DV-Hop to reduce the estimated distance error between nodes. The simulation results show that the algorithm effectively reduces the localization error without increasing the complexity of the algorithm and additional hardware.