Research On Routing Decision And Data Transmission Mechanism Of Underwater Acoustic DTN Sensor Networks

Underwater acoustic sensor networks(UASNs)is characterized by long propagation delay,narrow bandwidth,limited node resources and frequently disconnected interruption,which can be viewed as delay/disruption tolerant networks(DTNs).Therefore,DTN protocol can be introduced into UASNs to solve the above problems.However,the application of traditional DTN protocol for terrestrial communication in UASNs can increase network delay and fail to solve the problem of limited node resources,which leads to low transmission efficiency,high energy consumption and short network lifetime of underwater acoustic communication.In this context,this paper proposes a new cross-layer protocol mechanism of UASNs based on the protocol architecture of DTN,which combines the network layer and transport layer protocol for comprehensive optimization design,so that the protocol mechanism of DTN can be applied to the UASNs to ensure the effectiveness and reliability of underwater communication.In this paper,we propose an adaptive Deep Q-Network-based energy-and latency-aware routing decision algorithm(DQELR)to prolong network lifetimes in UASNs.In this algorithm,a Deep Q-Network algorithm with both off-policy and on-policy methods is adopted to make globally optimal routing decisions.Based on both the energy and depth states of nodes at different communication stages,nodes with the maximum Q-value can be selected as forwarders adaptively considering both energy and latency.A hybrid of the broadcast and unicast communication mechanisms is also designed to reduce network overhead.In addition,network topology changes can be addressed through an on-policy method that makes a new routing decision when the current route becomes corrupted.With less energy consumption and strict latency limitations,DQELR can prolong network lifetimes in UASNs.Simulation results show that DQELR routing algorithm can achieve a superior network lifetime with better latency and energy efficiency performances relative to other general schemes applied in UASNs.And then,we propose a transport mechanism,called optimal retransmission timeout(RTO)interval stop and-wait transmission(ORIT).In most general transport layer protocols,the RTO timer is set to longer than the RTT to avoid pseudo-retransmissions.However,great network latency could be introduced by applying such a mechanism in underwater communication networks.In ORIT mechanism,we propose an RTO optimization algorithm to maximize the remove redundant goodput(RRG)performance of the network by reducing the RTO timer.Meanwhile,we adopt the interval stop-and-wait transmission mechanism to avoid data pseudo-retransmissions caused by setting the RTO timer shorter than the RTT in narrow acoustic channels.Through adjusting the RTO timer,this mechanism can achieve the best RRG performance under an RTO timer,which we call the optimal RTO timer.We compare the ORIT with other transport layer protocols in the DTN network.The results show that this transmission mechanism with the optimal RTO timer can shorten the long propagation delay and effectively increase the data delivery rate in the underwater DTN sensor networks.