Research On Underwater Data Collection Algorithms In Uasns Based On Rbp System Model

In the 21 st century,countries have increasingly attached importance to the development and utilization of the ocean and vigorously developed Marine science and technology.Underwater Acoustic Sensor Networks(UASNs)deploy a large number of sensor nodes in specific areas to collect data via underwater acoustic channel,which has become a key technology for observing the ocean and obtaining underwater information.Hence,how to effectively collect data from UASNs has also become a key issue in marine research.As a new AUV,Glider,with its characteristics of low noise,low energy consumption,has gradually become an important platform for underwater data collection and research.In this paper,we focus on the application of gliders gathering data from underwater sensor networks.And the main research in this paper is how to schedule the sensor nodes in UASNs,effectively collect datafrom sensor nodes,and overcome the time-varying underwater environment and the delayed feedback.In this paper,we first analyze the time-varying characteristics of underwater acoustic channel and the energy change of underwater equipment.And then we formulate the dynamic evolution of system states as the multi-armed bandit(MAB)question with Markov chain and propose a distributed data collection strategy which schedules the sensor nodes.Besides,to tradeoff between the data quality(such as bit error rate(BER),etc.)and the collection delay,we design an indexable threshold policy and analyse the energy efficiency.Simulations prove that the proposed policy can effectively decreace the switching frequency between sensor nodes,reduce energy consumption,prolong the network lifetime of UASNs,and reduce the BER performance.Secondly,aiming at the problem of inaccuracy of system parameters caused by underwater noise and interference,we modified the system model as a more practical Restless Bandit Problem(RBP),and then establish the robust system model for the parameter uncertainty of system.We propose a robust index policy to solve the fuzzy of RBP system model for better robustness and the optimization of the scheduling policy and fewer system energy consumption.Besides,the Whittle index policy,which is calculated off-line and scheduling the bandit processes on-line,is used to reduce computation overhead on underwater equipment.Finally,since the low speed and multipath of underwater acoustic signals cause the delayed feedbacks,we formulate the system as a delayed feedback model based on RBP.Then we pursue an algorithmic solusion for the delayed feedback problem in RBP and propose a queue-based index policy and use the black box algorithm to apply the strategy in the environment without delay to the environment with delay,so as to effectively solve the absence of effective information caused by delayed feedbacks and improve the efficiency of decision making.In addition,the feasibility of this policy is proved.