A Novel Network Scheduling Approach Based on Genetic Algorithm for Autonomous Underwater Vehicle Control

Autonomous Underwater Vehicle (AUV) is one type of important equipments for conducting ocean exploration missions. The AUV control system typically employs a common communication network to connect various function nodes such as Doppler meter, compass/inertial navigation unit, sonar, water depth sensor, main propulsion, and rudder motor. Each of these function nodes has its own information transmission requirements that need to be served by the limited bandwidth of the common communication network. In order to obtain a good control performance, an elaborate information transmission schedule in the communication network should be planned in such a way that all information transmission requirements in the AUV control system can be satisfied with as less network-induced delay and data packet dropout as possible. This paper proposes a novel network scheduling approach based on the Genetic Algorithm (GA) and the time-triggered architecture for meeting information transmission requirements in the Controller Area Network (CAN) of the AUV control system. The proposed network scheduling approach divides the whole transmission period into a serial of time slices with equal lengths, and defines a large cycle consisting of multiple time slices. For the periodical information transmission, the transmission schedule in one large cycle is determined by solving one nonlinear integer optimization problem with the Genetic Algorithm. This GA-based network scheduling approach maximizes the available bandwidth of each single time slice in all the periodic time slices of the whole schedule period, and aims to achieve a balanced information transmission in the CAN network. Meanwhile, the available bandwidth in each time slice of the whole schedule period is then allocated for the event-triggered information transmission. In this way, the proposed network scheduling approach satisfies the transmission requirements of both periodical and event-triggered information in the AUV. Finally, simulation experiments demonstrate the performance of the proposed GA-based network scheduling approach, and verify its performance improvement on an AUV heading motion control using TrueTime toolbox.