Localization is essential for Underwater Sensor Networks (UWSN), as the information obtained by UWSNs must have accurate positions. Most of the existing location algorithms rely on the range, but the particularity of underwater acoustic communication seriously affects the ranging accuracy, which greatly reduces the availability of range-based localization algorithms. The malicious node interference also affects the positioning accuracy and causes harm to UWSN. To solve these problems, a Range-free and Level-based Localization scheme (RLL) is proposed in this paper, which establishes a tree structure for the nodes by moving beacons. Based on the tree structure, it divides these nodes into levels and zones. Signal strength is not used for ranging, but to calculate the relative position relationship between nodes and narrow the zones, localization is achieved by cooperation of nodes between adjacent zones. In addition, an effective protection mechanism to identify malicious nodes by level and zonal relationships is also proposed. RLL could be used in both static and dynamic situations. The simulations show that the localization effect of RLL is much better than the existing algorithms, and it can effectively detect malicious nodes to protect network security.

Information processing of guide dog robot requires expensive computing resources to meet real-time performance. We propose an edge computing framework based on Intel Up-Squared board and neural compute stick. Image processing and real-time control are performed on the framework. In addition, the voice recognition and commanding are also implemented, which are processed on Amazon cloud service. Vision, speech, and control services are integrated by ASU VIPLE (Visual IoT/Robotics Programming Language Environment). A prototype is developed, which implements the guide dog's obstacle avoidance, traffic sign recognition and following, and voice interaction with human.

This paper studies the nonlinear model predictive control problem (MPC) for networked control system (NCS) with delays in both channels. To compensate for the delays and reduce the computation, we first transform the NCS with delays into a non-delay system. Moreover, an explicit feedback control strategy is applied to provide a possible implementation for package lost and the delays. To avoid the prohibitively high online computational demand, the nonlinear MPC scheme is applied and the original problem is reduced to the nonlinear-quadratic regulator problem for an off-line system on every sampling time based on a vector iterative approximation approach. The validity of the proposed approaches is illustrated by a numerical example.

Data aggregation is a crucial technique for energy constrained sensor networks. Previous researches on data aggregation are featured as query-oriented, which only provide partial information of the event happened in the deployment area at the base station. In this paper, we propose an event-oriented data aggregation approach, called EDA. EDA presents the distributed algorithm by exploiting Cloud Membership model of fuzzy logic to aggregate the information of events in the sensor networks. It also presents a distributed algorithm to collect and aggregate the event information. The base station will restore the whole event information when receiving the aggregated packets of event features. EDA can balance the tradeoff between delay, traffic savings, and precision of the restored events. The performance has been evaluated through both theoretical analysis and simulations. We also confirm the performance with the traces of our offshore sensor network testbed (OceanSense).

Duty cycling is an important method for energy constrained sensor networks to prolong its lifetime. Current researches on duty cycling are often based on such assumption that all the sensory coverage should be maintained while some nodes are sleeping. For the applications of target monitoring, however, it is not necessary anymore to keep the whole sensory coverage of the sensor networks. It only needs to make sure that such kinds of nodes are active which can perceive the activities of the monitored targets. This observation brings new challenges to the design of duty cycling for the sensor networks. This paper proposes the novel duty cycling design for the sensor networks of target monitoring, which includes two algorithms of the sleep scheduling algorithm and the gradient query algorithm based on sleep periods. Under the proposed design, most of the sensor nodes can be in sleep, while still keep the functions of target monitoring and target query in the sensor networks. The performance of our design has been evaluated through both theoretical analysis and simulations, which prove the functionality of the proposed design on the reduction of energy consumption.

Existing localization approaches are divided into two groups: range-based and range-free. The range-free schemes often suffer from poor accuracy and low scalability, while the range-based localization approaches heavily depend on extra hardware capabilities or rely on the absolute RSSI (received signal strength indicator) values, far from practical. In this work, we propose a mobile-assisted localization scheme called perpendicular intersection (PI), setting a dedicate tradeoff between range-free and range-based approaches. Instead of directly mapping RSSI values into physical distances, by contrasting RSSI values from the mobile beacon to a sensor node, PI utilizes the geometric relationship of perpendicular intersection to compute node positions. We have implemented the prototype of PI with 100 TelosBmotes. Through comprehensive experiments, we show that PI achieves high accuracy and low overhead, significantly outperforming the existing range-based and the mobile-assisted localization schemes.