Research On Nodes Scheduling For Target Tracking Based On Swarm Intelligence Optimization

Wireless sensor networks(wireless sensor networks, WSN) is the integration of distributed information processing monitoring, network control, and wireless communication systems.The characteristics of distributed information processing and rapid deployment and invulnerability have made the WSN widely used in military reconnaissance, environmental monitoring, target tracking, space exploration and disaster rescue and other special areas.Specially, the target tracking becomes the most representative one. Target tracking is a key means to realize the regional monitoring, which can provide effective real-time decision support for many civil and military command systems, sush as fire control, enemy situation analysis, threat assessment and situation assessment. Therefore, it is important to study the target tracking algorithm in WSN. Because the WSN is constrained by energy, bandwidth, and storage. WSN must dynamically manage network resources to meet the needs of the tracking performance while extending the network lifetime. This thesis takes the target tracking as the background, mainly studies the node scheduling in WSN.Node scheduling for target tracking in WSN is a combinatorial optimization problem with constraints, which select appropriate sensors dynamically in each sensing moment. The model of node scheduling has important significance for the development of WSN. It is closely associated with network overhead and tracking performance. The swarm intelligent optimization algorithm has the characteristic of fast convergence and it has become an important tool for solving this kind of complex optimization problem. There is also a local optimum problem. Therefore, it is necessary to find the reason of local optimum and make an improvement. This thesis research from the aspect of improving the swarm intelligent optimization algorithm and building node scheduling models, which is important to the development of target tracking technology in WSN.This thesis is based on the swarm intelligent optimization algorithm and aims at the needs of target tracking in WSN. The main work and contributions are outlined as follows:1. Research from the aspect of targets prioritizing, reveal the nonlinear relationships between target property and priority. A weighted tactical significance map(WTSM) based targets prioritizing algorithm is proposed.The algorithm combines the ideas of linear weighting method and tactical significance map and divides the factors affecting the targets priority into two parts, relative attributes and target’s own property. The analytical hierarchy process is used to calculate the weights of these two parts. Simulation results show that, the WTSM algorithm effectively solved the nonlinear mapping problem between the target property and priority, and avoid a single attribute controlling the overall priority. It provides a theoretical support for establishing models of node scheduling for multi-target tracking.2.Research from the aspect of improving the search algorithm in energy untrammeled WSN, This thesis aims at the problem of node scheduling for target tracking in WSN, and a binary particle swarm optimization(BPSO) based sensor scheduling algorithm is proposed to maximize the tracking accuracy. Sensor scheduling model is based on the target predicted coordinate and target priority. The determinant of fisher information matrix(FIM) is used as accuracy measurement. The node scheduling model is built to maximize the tracking accuracy. A modified form of binary particle swarm optimization(MBPSO) is proposed to solve the model, which is designed via employing binary vector coding manner, constraint satisfaction cyclic shift population initialization method, particle position updating rules with V-shaped transfer function and guidance factor. The simulation results show that the proposed sensor scheduling algorithm can efficiently applied in multi-target tracking problem. Compared typical intelligent optimization algorithms, MBPSO algorithm achieves a balance between global optimization and local exploration, and can effectively avoid the local optimum. Moreover, the proposed algorithm can effectively applied to the node scheduling problem which maximize the tracking accuracy in energy untrammeled WSN.3.In order to balance the tracking performance and the network lifetime in target tracking of WSN, two adaptive node scheduling models are proposed under the dynamic clustering structure in energy limited network,in which node scheduling model was built composing of tracking accuracy and energy consumption and energy balance.(1)In high density network,through the in-depth analysis get that the key factors of network lifetime lie in the intra cluster communication and the number of clusters and the energy distribution of nodes.Therefore, a cluster keep model based on geometry tolerance is proposed;(2)Due to one step forward based node scheduling may only get the local optimal scheduling result.Research from the aspect of multi-step prediction,analyzing the relationship between the target trajectory and the predicted position. A multi-step prediction based node scheduling model is proposed. Moreover, the binary bat algorithm(BBA) with global optimization capability is used to solve the models. The simulation results show that the cluster keep based node scheduling algorithms can content the tracking performance, reduce energy consumption and balance the node energy to prolong the network lifetime.In addition, the algorithm can effectively solve the problem of traffic congestion and information overload bring by the frequent changes of task clusters,and can be effectively applied to large-scale WSN. The multi-step prediction based node scheduling algorithm can content the tracking performance, reduce energy consumption and balance the energy distribution of nodes. Moreover, this algorithm can not only reduce the number of task cluster,but also can reduce the network overhead bringing by local optimal node scheduling result,and can be effectively applied to non maneuverable target tracking scenario.4. Research from the aspect of the node scheduling algorithms for target tracking in WSN are lack of open platform.A extensible simulation system is developed based on C# and GMAP environment, which intggrates tracking, positioning and node scheduling. The system mainly includes the map module, node layout module, observation module, state prediction and estimation module, node scheduling module and performance evaluation module.The analyses and experiments of algorithms in this thesis on the simulation platform verifies the validity of our models and algorithms. The system has strong scalability, and can provide a general simulation platform for target tracking in WSN.