UDTN-RS: A New Underwater Delay Tolerant Network Routing Protocol for Facilitating Coastal Patrol and Surveillance Networks

The Coastal Patrol and Surveillance Application (CPSA) is developed and deployed to detect, track and monitor water vessel traffic using automated devices. Latest advancements of marine technologies, including the Automatic Underwater Vehicles, have allowed the development of this type of applications. To facilitate their operations, installation of a Coastal Patrol and Surveillance Network (CPSN) is mandatory. One of the primary design objectives of this network is to deliver adequate amount of data within an effective time period. This is particularly essential for reporting a detection and notifying the current status of an intruder’s vessel through the adverse underwater communication channels. Additionally, intermittent connectivity of the nodes remain another important obstacle to overcome to allow smooth functioning of CPSA. Taking these objectives and obstacles into account, this work proposes a new protocol, named UDTN-RS, which is developed by ensembling forward error correction technique (namely Reed-Solomon codes or RS) in Underwater Delay Tolerant Network (UDTN) routing protocol with probabilistic spraying technique. In addition, the existing binary packet spraying technique is enhanced for supporting encoded packet exchange between the contacting nodes. A comprehensive simulation campaign is performed in identifying the effectiveness of the proposed protocol. The obtained results suggest that the proposed UDTN-RS protocol can be considered a suitable alternative of the existing protocols for sparse networks like CPSN.


Introduction
Due to the advancement of marine technologies and the contributions by several marine surveillance projects in recent years -detecting, tracking, and monitoring of water vessel traffic come to an existence using an application, called Coastal Patrol and Surveillance Application (CPSA). It is a mission-critical application based on the cutting-edge software and equipment that accumulates, analyzes, and visualizes real-time information on the activities within a coastal zone under surveillance, including harbours against sabotage or terrorism and asymmetric threat from known and unknown enemies [1,2]. For facilitating the operations of CPSA, the CPSN is mandatory to install. Again, among various CPSN topologies (e.g., static, mobile, or hybrid), in this paper, a hybrid CPSN is considered alike in [3] where the sinks remain static and a fleet of AUVs patrol an area of interest and deliver • performing a comprehensive simulation campaign to identify the effectiveness of the proposed technique.
The rest of the paper has been organized as follows. In Section 2, we discuss the relevant DTN protocols with their error correction techniques. Section 3 elaborates the target application and its network architecture that is taken into account in this paper. Afterwards, Section 4 explains the UDTN-Prob protocol, which is the base protocol, and the proposed enhancements of UDTN-RS. Simulation scenarios and the acquired results are presented and analyzed in Section 5 and 6, respectively. This paper ends with concluding remarks in Section 7.

Related Works
Till date, many routing protocols for underwater DTNs are proposed in the literature. One of the typical ways of classifying these protocols is whether or not packets are replicated. Among them, routing protocols that replicate packets are called replication-based routing protocols [15], as opposed to forwarding-based protocols, which never replicate packets [16]. As a general observation, replication-based protocols achieve better packet delivery ratios over forwarding-based protocols, at the price of the greater overhead by injecting multiple replicas into the network. Again, the former also ensures the minimum end-to-end delay by increasing the chances of transmitting at least one packet among many replicas through the shortest path. Consequently, between both these classes, replication-based protocols are more suitable for CPSN due to the constraint of receiving adequate number of packets within the effective time period. Hence, it is taken into account in this paper. Most prominent protocols in this class are investigated below.
Epidemic routing protocol [17] is one such replication-based protocol, which performs massive replication by replicating a packet to each newly discovered contact that does not already own a copy of that packet. Therefore, similar to the flooding technique, epidemic routing is likely to achieve the best packet delivery ratio among all the replication-based approaches, albeit at the price of a very large replication overhead. Conversely, other routing protocols of this class endeavor to limit the replication overhead by generating and delivering replicas to only a handful of contacting nodes. For instance, Max Prop [18] prioritizes the packets to be transmitted or dropped upon contact with a peer based on a number of parameters, including packet generation time and lists of previous encounters. Even though, it incorporates a controlled strategy; however, it still experiences a considerable amount of overhead due to its flooding-based nature.
On the other hand, in Spray-And-Wait (SAW) protocol [19], the replication of each packet is restricted to a fixed number of copies. In the vanilla version of SAW, only the source is allowed to replicate, whereas the binary version allows intermediate relays also to replicate packets. However, in the latter version, the maximum number of replicas allowed is evenly split between the current and the next relay except the last replica, which it tries to deliver to the destination by itself. Another variant of it [20] considers overdue contacts as one of the criteria of selecting relays. The assumption that is considered here is that the nodes that did not encounter the destination for a long time are more likely to encounter it soon and hence, become the preferred relays.
PROPHET [21] is another replication-based routing protocol which limits the packet replication by forwarding a packet only to those neighbours that exhibit higher probabilities of delivering the packet to the destination within a short period of time. Another analogous protocol, named Resource Allocation Protocol for Intentional DTN (RAPID) [15] also limits the replication by computing the utility of the packets employing a list of global routing metrics, including the average delay or the number of missed deadlines and then, replicating packets with the highest utility only. Alternatively, single replication is advocated in the Prediction Assisted Single-copy Routing (PASR) protocol [23], which outperforms multi-copy routing in some resource-constrained underwater network scenarios.
In [22], a new DTN protocol is proposed, named UDTN-Prob protocol, which exploits the contact duration between the nodes and their contact knowledge to increase the packet delivery ratio and decrease the end-to-end delay. In the UDTN-Prob protocol, the probabilities of the contacts between intermediate nodes and sinks are estimated based on the relevant information exchanged. Afterwards, this knowledge is utilized in selecting the list of the packets to be exchanged with a prescribed probability. The statistics of future meeting times are inferred initially from synthetic mobility models, which well approximate the behavior of actual nodes and are updated as the network runs. This allows making the best use of the infrequent contacts among the nodes. Through the knowledge of the statistics of inter-contact intervals with the sink, the nodes can transmit only those packets that have a sufficiently high chance of being delivered to the sink before their lifetime expires.
In case of error correction, most of the aforementioned protocols employ a retransmission-based (aka ACK-based) error correction technique. However, due to the long inter-packet transmission delay for the long round trip time (RTT), this approach offers limited performance. For overcoming this issue, UDTN-Prob and a few other protocols employ a juggling-based packet retransmission approach where multiple packets are transmitted within a single RTT [11]. However, the juggling-based approaches impose higher implementation complexity, time synchronization problem, overhead for injecting ACK packets, and wasting time in the form of guard time to accommodate ACK packets. Therefore, this paper proposed a new protocol by ensembling RS [24] in UDTN-Prob protocol and incorporating a new packet spraying technique. This protocol is simple in terms of implementation, does not require time synchronization, no ACK required, no guard time required, and can overcome most of the aforementioned limitations.

Network Scenario
In recent times, coastal authorities are encouraged in employing automatic devices in marine activities, including CPSA (which is the focus of this paper) due to the advancements in the design of AUVs and other relevant devices over the last decade. In this paper, for imitating a realistic CPSN, a fleet of AUVs are considered patrolling autonomously an area of interest, inspecting surface ships or underwater assets, and delivering collected data to sinks, which are further connected with a control center as demonstrated in Fig. 1. Since the area of CPSNs are considerably large and generally, only a few AUVs are deployed to cover the area, they remain out of the contact of the control center and their fellow AUVs most of the time. Therefore, the inter-contact interval between the AUVs are generally high, and hence, discovering a fixed route from a source to a destination is impractical. In other words, AUVs experience intermittent connectivity; and therefore, they need cooperation from the other fellow nodes to deliver their packets to the destination. Again, since the connection between the nodes are Figure 1. A hybrid coastal patrol and surveillance network where AUVs are patrolling an area of interest and collecting respective data. Since nodes are not within the communication range of each other and experience intermittent connectivity, and thus, form a DTN network. Hence, for data exchange, they utilize store-and-forward based routing protocols where the data are exchanged opportunistically with an objective of delivering them to the sink, which is further connected with the control center. intermittent, they have to perform the task in a SAF-based manner by employing a DTN routing protocol.
In the CPSA, when a vessel enters into a surveillance area, one or more AUVs start following it [14]. Here, the responsibility of the follower(s) is to acquire desired data of the target (e.g., timestamped trajectory data) and report to the shore-based control center via sinks. For this, whenever an AUV detects a contact with a fellow AUV, they opportunistically exchange data with each other with a hope that the other node will deliver the data to the destination. Since this kind of application is mission-critical, timely delivery of adequate amount of data is immensely important for realizing the motive of the target.

Proposed Protocol: UDTN-RS
This section includes the details of the existing UDTN-Prob protocol [22], the technique of ensembling RS in the UDTN-Prob, and the enhanced packet spraying technique.

UDTN-Prob protocol
As mentioned earlier, the UDTN-Prob is a replication-based routing protocol with a list of distinguishing features that are highlighted previously and will be explained briefly in this section. Unlike other plain replication-based protocols, it employs statistical knowledge for restricting replication. In details, it leverages the knowledge of inter-contact intervals of the nodes to calculate their chances of meeting the destination in the future. Once the estimation of probable sink meeting time is performed, it is exploited in identifying the packets that have the greatest chance of being delivered before the expiry. Accordingly, those packets or a subset of those packets are exchanged between the contacting nodes. In addition, UDTN-Prob also calculates the probable contact duration between the nodes and fairly divides this duration between the contacting nodes.
For enabling these strategies, the UDTN-Prob introduces a new messaging scheme, which is comprised of three phases: 1) contact discovery via BEACON packets, ii) analysis of contacts via INFO packets, and iii) contact establishment via RESPONSE packets. In the first phase, every node periodically broadcasts BEACON packets for discovering other contacting nodes. Generally, in DTNs, deployed nodes generally experience intermittent connectivity due to infrequent contacts, and may also experience prolonged periods of isolation and hence, this phase is important. After transmitting a BEACON packet, a node, denoted as A, starts a timer and keeps waiting for receiving corresponding INFO packets from its neighbors, if any.
On the other hand, if another node, denoted as B, receives a BEACON packet, it replies with a corresponding INFO packet, as demonstrated in Fig. 2. This packet contains a number of information, including its current position and velocity (necessary for estimating the contact duration), a subsampled version of the distribution function of the inter-contact time between itself and the destination and other relevant information. After transmitting the INFO packet, it starts waiting for the corresponding RESPONSE packet for a fixed waiting time. If no RESPONSE packet is received before expiring the waiting time, B moves to the idle state. In case of A, it keeps collecting all the INFO packets from the neighbors and stores them in a buffer until the relevant waiting time is over. This way, it gets an opportunity to select the best from the available options. Once the timer expired, it fetches all the INFO packets from the buffer and calculates the approximate inter-contact durations employing the position and velocity information that are shared in the INFO packet using the following equation [22]: AB is the relative position and could be found as: AB is the relative velocity and could be found as: δ TX is the transmission range of a transducer and could be calculated as [22]: where, α A =⇒ position of node A α B =⇒ position of node B ζ α =⇒ estimations errors for the relative position β A =⇒ velocity of node A β B =⇒ velocity of node B ζ β =⇒ estimations errors for the relative velocity b −→ spreading factor -geometry of the propagation [26] a( f ) =⇒ thorp absorption coefficient for f in kHz [25] W(x) =⇒ principal branch of the Lambert function [22] γ tgt =⇒ target Signal-to-Noise Ratio (SNR) [25] L =⇒ data packet length ρ TX =⇒ transmit source level [22].
Once the inter-contact durations for all the INFO transmitted nodes are calculated, A selects the node that exhibits the longest probable contact duration, which must also satisfy the minimum threshold condition, i.e., τ c AB > τ c min . Here, τ c min = 2T D + 2T A + 2∆, where T D , T A , and ∆ are the transmission times of a DATA packet, an ACK packet, a fixed guard time (equivalent to a short propagation delay among the nodes), respectively. Let us assume that, B satisfies all the above conditions and gets selected. Consequently, A transmits a RESPONSE packet to B, which includes the share of its own contact duration, simply computed as τ A = ητ c AB . Here, the factor η could be employed for implementing priority policies. If there is no accountable priority policy, the estimate contact duration can be equally divided among the contacting nodes by setting, η = 0.5 (for AUV) and η = 1 (for SINK, since it has no packet to transmit for the application considered in this work), and When B receives the corresponding RESPONSE packet, it calculates its portion of the data transmission duration employing a simple equation, τ B = τ c AB (1 − η). Here, η = 1 for a SINK, since it has no packet to transmit as mentioned earlier; and hence, moves to WaitDataRx state. Otherwise, it moves to DataTx state and starts transmitting packet for its own assigned portion of the duration. In DataTx state, the sender first employs a packet selection technique that commences with the estimation of number of packets, ν that can be delivered within τ B epoch. Afterwards, from the buffer, only those packets are selected whose lifetimes are more than one-hop sink θ-meeting time prediction (see [22]). Note it imposes a large computational complexity with a very limited performance gain and hence, not selected in this paper. Again, if this condition selects more than ν packets, only top-ν packets are selected based on the packet lifetime.
After finishing the packet transmission of its own stake, the sender switches its role and move to WaitDataRx state or Idle state based on the mode of the INFO transmitting node. Unlike UDTN-Prob protocol, thanks to the FEC technique, a node does not have to wait for the ACK packets as demonstrated in Fig. 3. Again, when A is in WaitDataRx state, it keeps receiving packet until its time for data transmission starts. This way, estimated contact duration is fairly utilized by both parties. Fig. 4 shows the state transition diagram of UDTN-RS protocol where a complete list of states and the events that triggers state changes (i.e., transitions) are provided.

Ensembling RS in UDTN-Prob
This section discusses the ensembling technique of RS in UDTN-Prob and the mechanism of recovering erroneous packets using the RS(n, k, t) code where k is the unencoded packets, n is the encoded packets and n > k, and t is the number of packets that can be recovered if contain errors. Now, when a UDTN-RS enable node receives a packet from the upper layer, it stores the packet in a buffer until there are k unencoded packets for the identical destination and application pair. Again, this approach may lead to the starvation problem that may occur due to the lack of adequate number of packets for indefinite time. For resolving this problem, a timer is introduced that forces a sender to transmit packets without any encoding after expiration. Conversely, once the count of unencoded packets for an identical destination and application pair reaches to k, they are fetched from the buffer and encoded to n packets. To incorporate these changes, two new fields are introduced in the routing header, namely ground_id (GID) and packet_id (PID), where GID ∈ Z+ and any negative GID value indicates unencoded packet, and 0 ≤ PID < n.
Note that during the transmission, encoded and unencoded packets receive an identical attention. Thereby, this approach incorporates the RS FEC technique at a cost of a low buffering delay at the transmitting side. In addition, it is noteworthy to mention that encoding only occurs at the source node  Conversely, the destination waits to receive at least k correct packets from n encoded packets of a group before decoding and sending to the upper layer.

Enhanced Packet Spraying Technique
For UDTN-RS, an enhanced packet spraying technique is designed to incorporate encoded packets; whereas unencoded packets are already taken care of in UDTN-Prob protocol, which is also adopted in this new protocol. When two nodes come to an agreement of packet transmission after exchanging control packets, this enhanced packet spraying technique decides: how many encoded packets of a group to transmit and which members to select in case of this fragmentary transmission.
It is noteworthy to mention that UDTN-Prob employs a packet spraying technique similar to the one in [26]. More specifically, the binary technique of SAW is employed where a node transmits half of the copies of a packet and keeps half; except the last packet, which it tries to deliver by itself. For instance, if a node carries L copies of a packet, in each encounter, only L/2 copies will be delivered to the contacting node until the last copy.
When this spraying technique is enhanced for the proposed protocol, the demand of the application (i.e., CPSA) is taken into account, i.e., delivering adequate number of packets within the effective time period. For satisfying these constraints, it is necessary to maintain the tradeoff between packet delivery ratio and end-to-end delay. For that, instead of transmitting all n encoded packets of a group, UDTN-RS sprays only k of them. The rationales of such selection are as follows. Firstly, the contact durations between the nodes are considerably short with respect to inter-contact intervals, and hence, when a node comes to a contact with another node, generally it carries many packets to transmit. Consequently, transmitting k encoded packets allows a node to spray considerably more groups of packets to the other contacting nodes and thus, it makes an effort in reducing end-to-end delay. In addition, limiting spraying to k packets also increases chances of spraying a single group of packets to more number of intermediate nodes; and thus, increases the chances of delivering adequate number of packets to the destination. Again, to choose k encoded packets from n encoded packets, a round robin technique [27] is employed in UDTN-RS since all packets have equal priority.

Simulation Scenarios
The effectiveness of the proposed protocol is evaluated using a comprehensive simulation campaign. For that, DESERT underwater simulator [28] has been utilized, which is a package of ns-miracle [29] and ns2 [30]. This open-source simulator is available online, and offers a complete set of underwater networking features, which are necessary for imitating real underwater environment. It also includes the World Ocean Simulation System (WOSS) package [31], which is a framework aimed at improving underwater network simulations through a more realistic account of acoustic propagation.
For our simulation campaign, our CPSN is installed at the coordinates 39.97 • N and 11.82 • E, and spread thereafter. This network is comprised of one fixed sink, which is placed at one side of the network, and either 5 or 10 mobile AUVs are deployed. At the beginning of every simulation, all AUVs are deployed randomly over a selected area. Once the simulation begins, they start moving freely within the area. Their trajectories are simulated as random realizations of a Gauss-Markov process with fixed self-correlation parameter α = 0.8 [32]. This leads to random yet smooth trajectories that reproduces the actual trajectories sufficiently well that autonomous underwater vehicles may follow during patrol, reconnaissance or survey missions (see Fig. 5). Again, an analogous model is also assumed for the intruders. Upon detection of a intruder, a follower (i.e., an AUV) will start following it maintaining an offset distance for being stealth and avoid any unwanted collision.
All the nodes communicate via a Binary Phase Shift Keying (BPSK) modulation technique at a bit rate of 4800 bps using a central frequency of 25 kHz and a bandwidth of 9 kHz. By setting a source level of PTX = 150 dB re µ Pa relative to a distance of 1m from the source, this leads to an estimated nominal transmission range dTX = 2000m. The size of the BEACON packet is 10 Bytes, the INFO, RESPONSE, and DATA packets are 125 Bytes. All the results are presented in this paper after averaging over 25 runs.
In case of UDTN-Prob, packets are sprayed using the binary spraying technique and the replication frequency of a packet is restricted to 5. On the other hand, alike [24], n and k are selected as 3 and 2, respectively for UDTN-RS and the replication frequency of a group is restricted to 3. Three metrics are considered in evaluating the performance of the compared protocols, namely Normalized Packet Delivery Ratio (PDR), End-to-End Delay, and Normalized Throughput. Here, PDR is the ratio between the number of packets received and the number of packets transmitted; whereas, End-to-end delay is calculated based on the difference between the packet reception time and the packet generation time. And, normalized throughput of a node is the ratio of payload reception time against the simulation time.

Results Discussions
This section presents and discusses the results that are acquired from our comprehensive simulation campaign. The results are grouped together based on their scenarios, and discussed accordingly.

Scenario 1: varying data transmission rate
In this scenario, data transmission rate, λ was varied from 1 bps to 50 bps to observe the performance of the proposed protocol for various packet loads in the network. The acquired results are plotted in Figs. 6, 7 and 8 for node 5 and node 10 and compared with that of its ancestor, UDTN-Prob protocol. One of the core reasons of selecting 5 and 10 nodes is that when the area is fixed, increased number of nodes increases inter-contact frequencies in the network. Thereby, the simulation with 5 nodes exhibits sparse scenario and 10 nodes exhibits dense scenario with the likelihood of twice as many contacts as the former.
The acquired results of this scenario explores several preeminent facts. As could be observed from Fig. 6 is that the PDR declines with increasing λ. This observation is true for any protocols and any number of nodes. Again, in case of sparse network where there are a small number nodes, our proposed technique performs comparatively better than the UDTN-Prob in terms of the PDR (see Fig. 6). Thanks to the ensembling RS in UDTN-Prob and the enhanced packet spraying technique that replicated and forwards packets conscientiously. Conversely, the retransmission-based error correction technique of UDTN-Prob imposes delays in packet transmission for accommodating ACK packets, and hence, the contacting nodes can exchange relatively lower number of packets from a pile of packets. The highest normalized PDR, 0.38 is received by the proposed protocols for 5 nodes. However, when the inter-contact intervals are relatively low (in dense network), it can be observed that UDTN-Prob exhibits better performance at the expenses of higher overhead (see Section 6.3). In other words, albeit, UDTN-RS injects relatively lower overhead in the network than UDTN-Prob, it shows comparable performance in many cases. For instance, when a UDTN-Prob node generates 10 packets and the highest forwarding rate is restricted at 5, it injects 10 × 5 = 50 packets in the network. Conversely, UDTN-RS transmits 10 2 × 3 × 3 = 45 packets in the network. In case of end-to-end delay (see Fig. 7), for any λ values, UDTN-RS demonstrates the lowest delay for the sparse network. Thanks to the enhanced packet spraying technique, which ensures packet delivery within the relatively shortest period of time even when there are minimum number of nodes in the network. It even beats the end-to-end delay of UDTN-Prob with 10 nodes whose expected inter-contact frequencies are twice higher, let alone 5 nodes. Hence, UDTN-RS can be considered as a suitable alternative for CPSA in terms of delay.
As oppose to the PDR, normalized throughput (see Fig. 8) increases rapidly for any compared protocols until moderate loads or mid λ in the network and stagnate at the higher λ values. All the compared protocols reach to the threshold at around λ = 40. Again, for any network topology, UDTN-RS with 5 nodes outperforms the rest. Even it is higher than that of 10 nodes of UDTN-Prob. The highest normalized throughput received in this network is 0.0021 by the proposed protocol for sparse network scenario.

Scenario 2: varying area size
As could be observed from the previous scenario is that the performance of the compared protocols varies for sparse and dense network topologies. To explore more about this, another set of simulation campaign was conducted by varying the network area from 4 × 4 to 10 × 10 km 2 . The acquired results for normalized PDR, End-to-End delay, and normalized throughput are presented on Figs. 9, 10 and 11, respectively.
Analogous to the observation of scenario 1 for the normalized PDR (see Fig. 9), it declines with increasing λ for any compared protocols. Likewise, the normalized throughput (see Fig. 10) also declines with increasing λ value, as oppose to scenario 1. It is because, with increasing network area, Preprints (www.preprints.org) | NOT PEER-REVIEWED | Posted: 17 November 2020 doi:10.20944/preprints202011.0453.v1 inter-contact interval also increases, and hence, considerably lower number of packets are delivered to the destination. The results of normalized PDR and normalized throughput explore the fact that when the area is short, UDTN-Prob performs considerably better than the proposed technique at the price of higher overhead (see Section 6.3). However, with the increasing area, the performance of UDTN-Prob declines. In continuation of that, after a certain area size, the performance falls below the proposed technique; and the latter continues to dominate afterwards. Thanks to the ensembling of RS in UDTN-Prob and to the enhanced packet spraying technique that assist the proposed technique in dealing with increasing area size. The area after which UDTN-RS overpowers UDTN-Prob is 7 × 7 km 2 .
Again, when the area size is comparatively small, all the compared protocols exhibit lower end-to-end delay. However, it increases sharply upto a mid area size and afterwards, it observes  slow growth for the large area size. Among all the compared protocols and two λ values, UDTN-RS outperforms the rest in higher area size for λ = 16.67. However, when λ = 8.33, the proposed protocol suffers due to not having adequate number of packets for encoding. By calibrating the timer for independent packet transmission, this problem can be resolved. On the other hand, UDTN-Prob achieves the lowest end-to-end delay for the compact scenarios.

Scenario 3: overhead ratio
One of the design goals of UDTN-RS is to reduce overhead and increase packet transmission rate per contact. For that, the RS FEC technique is chosen, which reduces overhead in the following manner: i) in this proposed technique, no ACK packet is injected in the network, and thus, reduce overhead in the network; and ii) by injecting a lower number of copies of a single packet in the network as explained in Section 5. In addition, the proposed technique also takes advantage of replication to increase PDR. However, from our experiment, it has been discovered that even with lower replication copies, the Preprints (www.preprints.org) | NOT PEER-REVIEWED | Posted: 17 November 2020 doi:10.20944/preprints202011.0453.v1  proposed technique receives comparable or higher performance as demonstrated in previously. The relationship between the number of copies with respect to the number of nodes for UDTN-RS is depicted in Figure 12 using a contour graph. Again, in Fig 13, the comparison of overhead ratios is shown for the compared protocols for sparse and dense networks. As could be observed from the figures is that UDTN-RS introduces lower overhead in the network with respect to its counterpart. A tradeoff between the encoding and replication must be maintained to achieve performance goals, which can be performed by calibration these parameters.

Conclusion
This paper proposed the UDTN-RS protocol for facilitating coastal patrol and surveillance application -a time critical application that demands delivery of adequate number of packets for realizing the activities of the coastal area. In the proposed protocol, the RS is ensembled in UDTN-Prob for compensating the inherent erroneous nature of the underwater channels. In addition, an enhanced packet spraying technique is incorporated in the protocol for facilitating delivering adequate number Preprints (www.preprints.org) | NOT PEER-REVIEWED | Posted: 17 November 2020 doi:10.20944/preprints202011.0453.v1 of packets within the effective time period. The effectiveness of the proposed protocol is evaluated by performing a comprehensive simulation campaign. Results of the simulations are acquired and compared with that of its ancestor, UDTN-Prob protocol, where the proposed protocol outperforms its ancestor in sparse networks. Hence, it can be concluded that the UDTN-RS is a better alternative to UDTN-Prob protocol for coastal patrol and surveillance networks.