Methodologies and algorithms are presented for the secure cooperation of a team of autonomous mobile underwater sensors, connected through an acoustic communication network, within surveillance and patrolling applications. In particular, the work proposes a cooperative algorithm in which the mobile underwater sensors (installed on Autonomous Underwater Vehicles—AUVs) respond to simple local rules based on the available information to perform the mission and maintain the communication link with the network (behavioral approach). The algorithm is intrinsically robust: with loss of communication among the vehicles the coverage performance (i.e., the mission goal) is degraded but not lost. The ensuing form of graceful degradation provides also a reactive measure against Denial of Service. The cooperative algorithm relies on the fact that the available information from the other sensors, though not necessarily complete, is trustworthy. To ensure trustworthiness, a security suite has been designed, specifically oriented to the underwater scenario, and in particular with the goal of reducing the communication overhead introduced by security in terms of number and size of messages. The paper gives implementation details on the integration between the security suite and the cooperative algorithm and provides statistics on the performance of the system as collected during the UAN project sea trial held in Trondheim, Norway, in May 2011.

This paper examines a cooperative localization algorithm for teams of Autonomous Underwater Vehicles, as developed within the project THESAURUS. The localization algorithm is strongly dependent on the underlying acoustic communication network, as it only uses acoustic data (range, USBL measurements and transmission of navigation information) to reduce the error and the uncertainty. The localization algorithm is based on the use of an Extended Kalman filter to maintain a model of the entire swarm, while adapting the measurement equation on the basis of the specific measurement available. Simulative results show how the algorithm is effective in fusing the information. The underwater communication networks is based on a time-division multiplexing scheme and on the use of the MOOS pub/sub for application integration. Implementation details of the network are provided, together with a description of how the positioning systems (e.g. USBL) have been integrated within the constraints of the network structure. Preliminary experimental results on localization and communication are provided as obtained from a recent engineering sea trial.

This paper proposes two novel strategies to estimate the complete attitude of a stationary body, exploiting the measurements of a commercial Micro Electro-Mechanical System (MEMS) composed of a triaxial gyroscope and a triaxial accelerometer. These methods, both based on the Kalman filter framework, can solve the problem of the North seeking with high-accuracy, without the use of digital magnetic compasses (DMCs) or fiber optic gyrosopes (FOGs). The first method, called one axis carouseling, is an extension of the classic carouseling approach, and makes use of a non-continuous rotation about one axis and of an active roll and pitch compensation system. The two axes carouseling is based instead on two non-continuous rotations about different axes. Both methods have been validated with simulation tests, based on real sensor and rotary stages characteristics, where the body heading was estimated with high accuracy after convergence. Preliminary experimental tests with the novel one axis strategy have been performed; the results are very promising in light of future developments.

This paper deals with the problem of transient events in model-based observers for dynamic positioning of marine surface vessels. Traditionally, model-based observers experience a deterioration of performance during transients, and there is a give or take relationship between transient and steady state performance. To remedy this problem, we propose to use time-varying gains for a model-based observer. The gains are aggressive during transients to improve transient performance, and relaxed in steady state to lower the oscillations of the estimates. The proposed observer is analyzed with regard to stability. Its performance is verified in both a high-fidelity simulation model, and on experimental data with the research vessel (R/V) Gunnerus. In addition, a partial closed-loop validation with R/V Gunnerus has been performed.