Node Depth Adjustment Based Target Tracking in UWSNs Using Improved Harmony Search

Meiqin Liu, Duo Zhang, Senlin Zhang, Qunfei Zhang
2017 Sensors  
Underwater wireless sensor networks (UWSNs) can provide a promising solution to underwater target tracking. Due to the limited computation and bandwidth resources, only a small part of nodes are selected to track the target at each interval. How to improve tracking accuracy with a small number of nodes is a key problem. In recent years, a node depth adjustment system has been developed and applied to issues of network deployment and routing protocol. As far as we know, all existing tracking
more » ... mes keep underwater nodes static or moving with water flow, and node depth adjustment has not been utilized for underwater target tracking yet. This paper studies node depth adjustment method for target tracking in UWSNs. Firstly, since a Fisher Information Matrix (FIM) can quantify the estimation accuracy, its relation to node depth is derived as a metric. Secondly, we formulate the node depth adjustment as an optimization problem to determine moving depth of activated node, under the constraint of moving range, the value of FIM is used as objective function, which is aimed to be minimized over moving distance of nodes. Thirdly, to efficiently solve the optimization problem, an improved Harmony Search (HS) algorithm is proposed, in which the generating probability is modified to improve searching speed and accuracy. Finally, simulation results are presented to verify performance of our scheme. of 16 in an area with a low coverage rate, the network will fail to track it accurately, and it lacks the adaptive adjustment ability to dynamic events. Therefore, we need to improve the flexible sensing and communication ability of UWSNs. As we know, underwater nodes are floating underwater with the help of buoys and mooring lines. In this paper, we consider that the nodes are equipped with a node depth adjustment system. Detweiler et al. [8] presented a depth adjustment system that connects to underwater nodes. Wu et al. proposed a depth adjustment scheme to maximize the coverage in 3D space [9] . The depth adjustment for underwater nodes is always applied to issues of network deployment and routing protocol. It can improve the network coverage rate and data packet delivery ratio, thus increasing network reliability [10, 11] . Inspired by these research results, this paper proposes a novel scheme to improve target tracking accuracy. It is assumed that nodes are located at their original position when they are not working. Once some nodes are woken up by fusion center, they adjust their depth according to commands from the fusion center and sense the target at the optimal depth, in order to improve tracking performance. Thus, the key problem is how to determine optimal depth of nodes during the tracking task, which can be converted to a dynamic optimization problem under the constraint of moving range. In our previous work, we provided some solutions to select the optimal node cluster for target tracking [12, 13] . Fisher Information Matrix (FIM) and its inverse matrix posterior Cramer Rao Low Bound (PCRLB) can reflect estimation accuracy, and we employed them as criteria to select the optimal node cluster. Hence, we will take FIM as an objective function in this paper and compute the optimal nodes' depth in the framework of optimization problem. As an optimization problem, finding the global optimal analytical solution becomes extremely difficult. In theory, the optimal depth can be determined if we discretize the constraint moving range and perform an exhaustive search. However, this method is not practical because of the heavy computational burden. For this reason, an improved Harmony Search algorithm (HS) is proposed to solve this optimization problem. HS is a meta-heuristic optimization algorithm that mimics the improvisation process of music players. It is a population-based search algorithm that can successfully solve optimization problems [14] [15] [16] . In this paper, each harmony represents nodes' moving distance, which is related to final position of nodes. For traditional HS, a harmony memory matrix is randomly initialized within moving range. Some harmonies may lead a node's final position to be far from the target; however, such harmonies are not that useful. The closer the distance between node and target, the better the performance may be. Therefore, we adjust the probability of a new harmony being generated as increasing as the distance to the target decreases, in order to improve tracking accuracy and searching speed. The main contributions of this paper are threefold. Firstly, we propose a node depth adjustment scheme to dynamically improve tracking accuracy in UWSNs. This is a novel idea that has not been done before. Secondly, we employ the relationship between FIM and nodes' depth as an objective function, and determine moving distance in the framework of optimization problems. Thirdly, an improved HS is proposed to solve this optimization problem, improve tracking accuracy and searching speed. The rest of the paper is organized as follows. In Section 2, an overview of the related work about this paper is provided for readers. In Section 3, the target tracking problem in UWSNs is formulated based on node depth adjustment. In Section 4, the relationship between FIM and nodes' depth is employed to determine nodes' depth, and an optimization optimization problem is formulated. An improved Harmony Search algorithm is proposed to solve the optimization problem in Section 5. Simulation results are presented in Section 6 and conclusions of this paper are drawn in Section 7. Related Work As for the development of UWSNs, the research of target tracking in UWSNs has gradually become an important issue. Huang et al. [17] presented a two cluster-based distributed particle filter tracking algorithms. The first algorithm focused on tracking accuracy, while the second considered the
doi:10.3390/s17122807 pmid:29207541 pmcid:PMC5751611 fatcat:ubnj4ff7kfdonew7s5h6cl3lhi