ACCEPTED FROM OPEN CALL

An SDN Architecture for AUV-Based

Underwater Wireless Networks to
Enable Cooperative Underwater Search
Chuan Lin, Guangjie Han, Mohsen Guizani, Yuanguo Bi, Jiaxin Du, and Lei Shu

Abstract scale realization of offshore industries and under-

water military/civil applications (e.g., resource
With the emergence of new underwater ICTs, exploration, sea condition monitoring, deep-sea
UWNs based on AUV have become the main- detection, and so on) [2]. As a result, this phe-
stream technology for underwater search tasks. nomenon also gives rise to a vision of a ‘smart
These advanced underwater searching technolo- ocean’ that integrates new types of underwater
gies are leading to smart perceptual ocean tech- searching devices and ICTs into a unified system,
nologies. However, to support precise multi-AUVs such that the ocean can be intelligently perceived,
cooperative underwater search and provide intel- investigated and surveyed. In AUV-based UWNs,
ligent data collection and data transfer among the AUVs cooperate with each other to achieve
the AUVs, one of the challenging issues is to a unifying searching target [3]. For instance, in
design a scalable network architecture capable of marine oil and gas industries, AUV-based UWNs
fine-grained control and smart underwater data are deployed on the seabed and aim to search
routing. In this article, we employ the paradigm the border of the oil and gas-producing area in a
of SDN technology and propose an SDN-based cooperative manner, such that the infrastructures
underwater cooperative searching framework for and pipelines for oil and gas can be efficiently
AUV-based UWNs. In particular, we propose a built. Consequently, AUVs require to communi-
software-defined beaconing framework integrating cate constantly with each other to share and syn-
two categories of defined beacons to synchronize chronize the searching information for performing
network information and execute network oper- cooperative control [4].
ations. Based on the software-defined beaconing Currently, underwater communications are
framework, we introduce the USBL positioning sys- predominated by acoustic-based/optical-based
tem and propose a hierarchical localization frame- methods, among which the acoustic-based com-
work to localize/track each AUV in the network. munication technologies have gained more atten-
Then, we utilize the potential field theory to model tion due to its long communication range (ranging
the multi-AUV cooperative operation, leading to a from 5 to 100 km at the bandwidth of 2-5 kHz)
cooperative control framework. Finally, to guaran- and simplicity for deployment and implementation
tee the potential data transfer among the AUVs, [5]. However, acoustic-based communication in
we also propose a software-defined hybrid data underwater suffers seriously from long propaga-
transfer scheduling framework. Simulation results tion delay, limited bandwidth, and is vulnerable
demonstrate that our proposed scheme performs to malicious attacks. By comparison, optical-based
more efficiently than some existing schemes espe- underwater communication has advantages in
cially the distributed control policy. propagating delay and bandwidth, while its com-
munication range is limited (ranging from 30 m
Introduction to 100 m based on Light Emitting Diode (LED)
The ever increasing human interest in ocean blue light) [6]. To efficiently control the AUVs and
exploration is promoting the deployment of schedule the data transfer, it requires to dynamical-
more advanced technologies [1]. Recently, with ly switch the communication mode according to
the rapid development of underwater Informa- the network status, the delay-sensitive requirement
tion and Communication Technologies (ICTs), of the cooperative control messages, and even the
and smart devices for underwater search and sur- Quality of Service (QoS) requirement of the data
veillance (e.g., underwater robots, Autonomous transfer among the AUVs [7]. By selecting a feasi-
Underwater Vehicles (AUVs), underwater sen- ble communication mode, the network states, the
sors, and so on), Underwater Wireless Networks sensed data and the location of each AUV can be
(UWNs) especially based on AUVs are increasing- timely and accurately gathered at the control unit,
ly being considered as cost-effective methods of such that the cooperative searching strategy can
ocean searching and monitoring. These emerging be determined and deployed. From the above, a
underwater searching technologies enable large scheme integrating information synchronization,
Chuan Lin is with the Key Laboratory for Ubiquitous Network and Service Software of Liaoning province, Dalian University of Technology;
Guangjie Han (corresponding author) is with the College of Engineering, Nanjing Agricultural University, and the School of Software,
Dalian University of Technology; Mohsen Guizani is with Qatar University; Yuanguo Bi is with Northeastern University, Shenyang,
Digital Object Identifier: China; Jiaxin Du is with the School of Software, Dalian University of Technology;
10.1109/MWC.001.1900387 Lei Shu is with the College of Engineering, Nanjing Agricultural University.

132 1536-1284/20/$25.00 © 2020 IEEE IEEE Wireless Communications • June 2020

Authorized licensed use limited to: Thapar Institute of Engineering & Technology. Downloaded on March 29,2024 at 06:14:38 UTC from IEEE Xplore. Restrictions apply.
HAN_LAYOUT.indd 132 6/9/20 3:54 PM
 node localization or tracking and Traffic Engineer- The remainder of this article is organized as fol- The SDN technique
ing (TE) decision functions is necessary to guaran- lows. The proposed SDN-based architecture for provides network cen-
tee the cooperative search [8]. AUV-based UWNs is presented in the following tralized management
It is a challenging issue to enable the AUV- section. Then we present the SDN-based underwa- by using a centralized
based UWNs to perform the cooperative under- ter cooperative searching scheme. Following that
water search. This is due to that current computer we present the simulation results. The final section SDN ‘controller’ and
networks are usually based on distributed network concludes the article and discusses possible future standardizing the
architecture, which does not allow comprehen- works. uniform program-
sive and centralized network management. Con- ming interface for the
sequently, each AUV in AUV-based UWNs needs
to be independently configured for cooperative
The Proposed SDN-Based Architecture for network control. This
operation. This cannot guarantee precise or time- AUV-Based UWNs enables the network
ly cooperative search. In [9] of the revised man- As already stated, in order to provide real-time operator to program
uscript, Shiliang et al. propose a multi-robot path network control and perform efficient underwater the operation of the
planning scheme based on Artificial Potential Field. search, the network architecture ought to have AUV-based network for
Their proposal can perform exact planning and the ability of centralized management or control deploying intelligent
obstacle avoidance for the multi-robot system, of the network resources, which is impossible in
which can be promoted in the area of path plan- traditional IP-based networks. The SDN technique underwater search and
ning for the AUVs. However, their proposal is avail- provides network centralized management by surveillance policies.
able under strict condition that each robot/AUV using a centralized SDN ‘controller’ and standard-
can acquire the states of their neighbors, and a izing the uniform programming interface for the
global view of the entire network can be timely network control. This enables the network oper-
maintained. Therefore, according to the aforemen- ator to program the operation of the AUV-based
tioned factors, a centralized network management network for deploying intelligent underwater
platform is indispensable to improve the scalability search and surveillance policies. With this moti-
of the network, such that exact cooperative under- vation, in this section, we propose an SDN-based
water search can be performed. architecture for AUV-based UWNs which consist
Recently, the paradigm of Software-Defined of three layers (i.e., the data layer, the local con-
Networking (SDN) has emerged and is considered trol layer, and the main control layer) as shown
to be a highly anticipated technology to address in Fig. 1 which can briefly be depicted as follows.
the network scalability issues. SDN aims at decou-
pling the network control plane from the data Data Layer
plane by using high-level abstraction interfaces, The data layer of the proposed SDN-based net-
such that centralized network control can be real- work architecture is a network of mobile nodes
ized and the flexibility or simplicity of the network based on AUVs (consisting of data Collecting
can be improved [10]. Based on SDN, flexible net- AUVs (C-AUVs) and data Storing AUVs (S-AU-
work control and complicated TE strategy can be Vs)) which are distributed in a particular area of
determined and deployed. For instance, in [11], the ocean to execute the underwater search or
Chuan et al. utilize the advantage of SDN and pro- surveillance. Note that the C-AUVs equipped with
pose a QoS-aware routing engine for performing advanced sensing devices (e.g., salinity and tem-
intelligent data routing when the spatial-temporal perature monitoring sensors, multi-beam and side-
features of SDN-based UASNs are considered. scan sonar) are dedicated to collecting the data
Motivated by the SDN paradigm, in this arti- by following a pre-defined policy. For instance,
cle, we employ SDN technology and propose an when the AUVs are scheduled to collect data
SDN-based distributed architecture for AUV-based from the underwater sensors divided by differ-
UWNs. Based on the proposed network archi- ent clusters, the collecting path for C-AUV can
tecture, we propose an SDN-based underwater be adaptively scheduled by ranking the priority
cooperative searching scheme including a soft- of each cluster, for example, ranking the priority
ware-defined beaconing framework, a hierarchi- based on the average (residual) power of each
cal localization framework, a cooperative control cluster [13]. The S-AUVs are cruising along a fixed
framework, and a software-defined hybrid data track and are only in charge of storing the data
transfer framework. The main contributions of this transferred from C-AUVs. To enhance the cruis-
article are summarized as follows: ing ability of the S-AUVs and perform efficient
• We propose a scalable SDN-based architecture data storing, the S-AUVs are usually equipped
for AUV-based UWNs, which divides the net- with large mass storage devices as well as a
work architecture into three layers: the data high-capacity battery. The S-AUVs are periodically
layer, the local control layer, and the main con- scheduled to the AUV-based Local Controllers
trol layer. In particular, the data layer of the (AUV-LCs) in the local control layer for maintain-
proposed network architecture consists of two ing (e.g., battery charge), and data unloading or
categories of AUVs. transfer scheduling. Notably, the sensed data can
• We define two categories of software-defined be routed through hop-by-hop routing policies,
beacon messages, BEA_OPE and BEA_SYN, to and the communication mode (acoustic/optical)
provide information synchronization among the can be dynamically switched based on multiple
AUVs and active request for the AUVs in the factors, for example, the QoS, data scale, loca-
data layer, leading to a proposed hierarchical tions of the AUVs, available channel, and so on. 1 For instance, the LED array
localization framework. For instance, the optical-based communication with collimating aspheric
• We propose a cooperative control framework mode based on blue light LED1 can be adopted, if lenses can be selected as a
candidate paradigm to per-
based on potential field theory and a hybrid the routing nodes (AUVs) are within a short range form LED-based data transfer
(optical/acoustic) data transfer policy based on and the scale of the data is large. Otherwise, the through high-energy conver-
Saaty’s AHP algorithm [12]. acoustic-based communication mode is more suit- gent beam.

IEEE Wireless Communications • June 2020 133

Authorized licensed use limited to: Thapar Institute of Engineering & Technology. Downloaded on March 29,2024 at 06:14:38 UTC from IEEE Xplore. Restrictions apply.
HAN_LAYOUT.indd 133 6/9/20 3:54 PM
 The proposed coop-
erative control model Data relay
is composed of three Satellite
Control synchronization
parts: the task model, Control decision
USV-MC
Satellite base Data analysis
and control models Station
for S-AUVs, C-AUVs, East/West-bound API
AUV management
respectively. Notably,
Traffic engineering
all the operations or
Topology awareness
the searching tasks are Internet
ernett Node localization
South-bound API
under the direction
of the SDN control- Data collection
lers, based on which Data routing
the C-AUVs perform Data storage
Cloud computing Data fusion
underwater search
according to the task
model and transfer S-AUV C-AUV AUV-LC
Acoustic-based
i b Optical-based Acoustic-based
control message data transfer data transfer
the sensed data to the
S-AUVs.
FIGURE 1. Diagram displaying the proposed SDN-based architecture for AUV-based UWNs.

able. All the actions including routing decision, both power charging and data high-quality trans-
data fusion, and so on, are directed by the AUV- fer, otherwise, acoustic-based communication
LCs in the local control layer through dedicated can provide data transfer of large range.2 Simul-
acoustic tunnels/channels. taneously, the USV-MC also maintains a global
view of the entire AUV-based UWNs (e.g., the
Local Control Layer entire network topology) and synchronizes the
The proposed SDN-based architecture for UWNs control messages for multi-AUVs cooperative
is with the distributed control layer, that is, the search. Consequently, the USV-MC is treated
control layer is divided into local and main con- as the control decision center when the control
trol layers. The local control layer aims to manage strategy across multi-domains is to be deployed.
the operations and missions in the data layer by Furthermore, the USV-MC also serves as the gate-
actively sending or passively receiving the inquir- way between AUVs and the data center on the
ing messages to/from the C-AUVs and S-AU- ground. Namely, the USV-MC gathers the sensed
Vs, through the dedicated acoustic channel for data from the local control layer and transfers the
control messages. All the operations in the data data to the computing center (e.g., cloud com-
layer are directed by the AUV-LCs which are with puting unit) for further industrial analysis through
particular control units (e.g., high-performance the satellite networks, the Internet, and so on. It
CPU, high-energy-density battery, and so on). The should be clarified that the data layer and control
AUV-LCs are responsible for managing the S-AU- layer are allocated with different ranges of acous-
Vs/C-AUVs (e.g., cruising direction, speed, coop- tic frequency when the acoustic communication
erative search, data relay for the S-AUVs, charging model is utilized to perform underwater data
for the AUVs, and so on) within a specified transfer or control message broadcasting.
domain. The number of C-AUV/S-AUV controlled
by one AUV-LC depends on the control efficiency
requirement of the control plane, which is speci-
The Proposed SDN-Based Underwater
fied according to the maximum load of the con- Cooperative Searching Scheme
trol channel, the distributed range of the AUVs, To achieve intelligent and cooperative underwa-
and the features of the computing/communica- ter search, we propose an SDN-based underwa-
tion units. Meanwhile, the AUV-LCs also provide ter cooperative searching scheme for AUV-based
the capability of traffic engineering for data trans- UWNs, as the cooperative control model present-
fer scheduling in the data layer (e.g., determining ed in Fig. 2a. The proposed cooperative control
the communication mode of the AUVs), topology model is composed of three parts: the task model,
awareness for the AUV-based network, and local- and control models for S-AUVs, C-AUVs, respec-
izing or tracking the trajectory of AUVs. Further- tively. Notably, all the operations or the searching
more, the AUV-LCs will ask the USV-based Main tasks are under the direction of the SDN control-
Controller (USV-MC) in the main control layer for lers, based on which the C-AUVs perform under-
further analysis or cooperative search performed water search according to the task model and
by the AUVs across multi-domains. transfer the sensed data to the S-AUVs.

Main Control Layer Task Model

In the main control layer, a USV-MC is deployed The task model is responsible for defining the task
on the water surface, and the AUV-LCs can work format including the timer selector leading to a
2 Here, we assume the USV- in two manners by tethering it or not to the USV- uniform synchronization mechanism with efficient
MC is powered by solar MC, according to the category of real applica- task scheduling strategy. The task model provides
panels and is available to
supply continuous power for tions. If tethered, the cables integrating optical the control model for C-AUVs with standards to
the AUVs. fiber/power cable [14] can be selected to provide perform cooperative underwater search.

134 IEEE Wireless Communications • June 2020

Authorized licensed use limited to: Thapar Institute of Engineering & Technology. Downloaded on March 29,2024 at 06:14:38 UTC from IEEE Xplore. Restrictions apply.
HAN_LAYOUT.indd 134 6/9/20 3:54 PM
 FIGURE 2. Proposed SDN-based underwater cooperative searching scheme: a) Cooperative control model; b) Partitioned searching
policy; c) Spiral elliptic trajectory.

controL ModeL for s-AuV zation policy as they did in [13]. By comparison,
The control model for S-AUV aims to define the S-AUV is pre-defined to be cruising along the fixed
format for storing the data which is gathered from spiral elliptic trajectory from top to down, which
the C-AUVs. The S-AUVs will adaptively adjust traverses each cuboid (Fig. 2c). Note that the tra-
the data storing mode based on the type and size jectory of each AUV can be dynamically revised by
of the data. The scheduling mode in the control an inertial/acoustic-based navigation system.
model for S-AUV defines the data transfer (to the
AUV-LC) period, path, strategy, and so on. the softWAre-defIned beAconIng frAMeWork
To provide information synchronization and shar-
controL ModeL for c-AuV ing, we introduce a software-defined beacon-
In the control model for C-AUV, all the opera- ing framework led by the AUV-LCs or USV-MC
tions for C-AUVs are coordinated by a centralized together, which can be promoted to localize/
task controller which is designed to be deployed track the AUVs and synchronize the network
on the AUV-LCs. The control model for C-AUV information among the AUVs (including the AUV-
realizes the cooperative underwater search in the LCs). We define two categories of beaconing
following steps: messages which are directed by the AUV-LCs or
• Broadcasting the beacon to synchronize the USV-MC and performed in a request/reply man-
information (e.g., the depth, speed, left power, ner:
and so on) among the C-AUVs or S-AUVs, such BEA_OPE: Is defined to perform a series of
that a global network view can be acquired. operations triggered by the AUVs in the data
• Based on the developed underwater localiza- layer/AUV-LCs, for example, network topology
tion system, the localization of each AUV in the awareness, AUV localization/tracking, exception
AUV-based UWNs can be tracked, leading the handling, and so on. For instance, in Fig. 3a, the
network topology to be predictable (e.g., the AUVs in the data layer/AUV-LCs request the AUV-
spatiotemporal features of the network topol- LCs/USV-MC for performing localization for them-
ogy) and cooperatively controllable for under- selves by broadcasting Operation_request. After
water search (e.g., the synchronization strategy the requested AUV-LCs/USV-MC receive Opera-
based on the relative position of the AUVs). tion_request, AUV-LCs/USV-MC execute the local-
• After executing the above steps, the coopera- ization phase for the requested AUVs and reply
tive control strategy can be determined based the localization or trajectory estimation results to
on the pre-defined cooperative models, and the requested AUVs by BEA_OPE. More details, we
the searching trajectory for each AUV can be present a hierarchical localization framework based
generated. on BEA_OPE .
• Meanwhile, the global optimal traffic engineering BEA_SYN: Is defined to synchronize the infor-
strategy can be determined to schedule the traf- mation among all the components. Meanwhile,
fic between C-AUVs and S-AUVs dynamically. BEA_SYN can also be utilized to synchronize/
In the following, we will detail each step and deploy the control message from AUV-LCs/USV-
and present the new approaches proposed in this MC to the AUVs in the data layer/AUV-LCs. As
study. We refer to our previous work in [2] and presented in Fig. 3b, the BEA_SYN (with request_
propose a partitioned searching policy to sched- type=synchronization) is utilized to synchronize
ule the searching trajectories for both C-AUVs and the information from AUV-LCs to USV-MC, to
S-AUVs. As shown in Fig. 2b, given a fixed search- maintain a global view of the entire network and
ing area that can be divided into a set of cuboids provide the optimal cooperative search with com-
of equal size, each of which is with a group of prehensive information. While request_type=ex-
C-AUVs for performing data collection from the ception checking as shown in Fig. 3c, the C-AUV is
sensors or target searching. In each cuboid, the required to report its existing exception to AUV-LC,
C-AUV’s trajectory can follow an adaptive optimi- based on BEA_SYN. Note that the AUV is judged

IEEE Wireless Communications • June 2020 135

Authorized licensed use limited to: Thapar Institute of Engineering & Technology. Downloaded on March 29,2024 at 06:14:38 UTC from IEEE Xplore. Restrictions apply.
HAN_LAYOUT.indd 135 6/9/20 3:54 PM
 FIGURE 3. Proposed framework for software-defined beaconing: a) BEA_OPE; b) BEA_SYN (synchronization); c) BEA_SYN (exception
checking).

3. Assuming the AUV-LCs and USV-MC are

equipped with Ultra Short Base Line (USBL)
Satellite
positioning system, the USV-MC utilizes the
acoustic transducer array with a set of hydro-
phones to receive Operation_request and
1. Localization based on GPS computes the difference of distance from the
sender to different hydrophones. Then, USV-
MC can utilize USBL to localize AUV-LCs.
3. Localization based on USBL 4. The USV-MC sends the computed position
USV-MC
to AUV-LCs by BEA_OPE
4. BEA_OPE 5–7. Similarly, as the benchmarks, the AUV-LCs can
assist the AUVs in the data layer with localiza-
tion, by utilizing USBL positioning system.
2.Operation_request 6. Localization based on USBL
(localization) the cooPerAtIVe controL frAMeWork
AUV-LC To cooperatively control the AUVs or AUV-based
7. BEA_OPE UWNs for searching or surveilling a common
target, we employ the potential field theory to
model the multi-AUV cooperative operation.
5.Operation_request Here, we take the target hunting as an example.
(localization) C-AUV Based on the SDN technology, the cooperative
control framework for underwater search can be
carried out as follows:
1. C-AUV u detects the target by utilizing the
FIGURE 4. Hierarchical localization framework. multi-beam and side-scan sonar and reports it
to the AUV-LC.3
2. AUV-LC localizes u, and shares the network
to be going through a downtime, when the AUV- information together with the searching target
LC continuously sends n synchronization_request with the other C-AUVs or S-AUVs
beacons to the AUV and receives no response. 3. Based on the information for the entire net-
This will result in a serious exception handle pro- work as depicted in Step 2, constructing the
cedure. grid-based active model [15] for the target and
building the attraction-repulsion model for each
the hIerArchIcAL LocALIZAtIon frAMeWork searching AUV. The policy is heuristically exe-
To accurately control each AUV (in the data cuted until each searching AUV arrives in a grid
layer) leading to a cooperative searching plane, around the target.
the position of each AUV ought to be periodical- 4. AUV-LC deploys the cooperative searching
ly acquired or predicted, such that the trajecto- strategy (e.g., the direction, speed, depth, and
ries of AUVs can be tracked and the cooperative so on) by sending the acoustic-based control
searching strategy can be efficiently deployed. In messages to each corresponding C-AUV.
this article, we propose a hierarchical localization 5 During the hunting, each AUV utilizes BEA_OPE
framework based on BEA_OPE of the proposed beacon to report to AUV-LC for some partic-
software-defined beaconing framework. Refer- ular issues, for example, obstacle avoidance,
ring to Fig. 4, the localization for AUVs (including additional exception, and so on, and AUV-LC
AUV-LCs and the AUV in the data layer) can be dynamically adjusts the hunting model.
carried out in the following seven steps: 6 Periodically, Steps 1–5 are re-conducted until
1. The USV-MC is localized with the assistance the target is hunted.
3 Here, we only consider the of GPS and is regarded as the benchmark
scenario where only a local
area AUV-based network
for localizing the AUV-LCs. the softWAre-defIned hybrId dAtA trAnsfer frAMeWork
2. With Operation_request (Request_type=lo- As mentioned above, we propose to utilize a
directed by an AUV-LC is
adopted for scheduling the calization), AUV-LCs periodically request hybrid (acoustic/optical) communication mode
search. the USV-MC for localization assistance. in the data layer of the proposed SDN-based

136 IEEE Wireless Communications • June 2020

Authorized licensed use limited to: Thapar Institute of Engineering & Technology. Downloaded on March 29,2024 at 06:14:38 UTC from IEEE Xplore. Restrictions apply.
HAN_LAYOUT.indd 136 6/9/20 3:54 PM
 UWNs. To achieve this purpose, we propose a
software-defined hybrid data transfer framework,
as presented in Fig. 5. As shown on top of Fig. 5, the
TE scheduling module includes network topolo-
gy awareness, monitoring, and dynamic schedul-
ing modules. Based on the abstract global view
acquired by the topology awareness module, Dynamic scheduling
Network topology
the network monitoring module synchronizes awareness
Network monitoring based on SaatyĆs
the environmental sensing information (optical/ AHP algorithm
acoustic communication condition, currents, and
so on), self-health status, and so on, from the
C-AUVs/S-AUVs to AUV-LC. Thus, the dynamic
scheduling module can dynamically determine
the multi-parameters optimization policy (e.g., uti- src_addr dst_addr QH[WBDGGU action src_addr dst_addr QH[WBDGGU action

lizing Saaty’s AHP algorithm [12]) for scheduling 

ĂĂ

ĂĂ

ĂĂ
xxx
ĂĂ

ĂĂ

ĂĂ

ĂĂ
xxx
ĂĂ
the traffic, and deploy intelligent flow tables for ĂĂ ĂĂ ĂĂ ĂĂ ĂĂ ĂĂ ĂĂ ĂĂ
specifying the data transfer manner (acoustic/
optical) to unload the data. Meanwhile, all the
analytic hierarchy process is performed by AUV- b

LCs or USV-MC. For instance, an S-AUV is per- 1.Light brightnessχ
forming the data storing task among the C-AUVs. src_addr dst_addr QH[WBDGGU action
   [[[ 2.Light angleχ
While the distance from the C-AUV to S-AUV or ĂĂ ĂĂ ĂĂ ĂĂ
3.Light duration
between the C-AUVs is less than the given com- ĂĂ ĂĂ ĂĂ ĂĂ

munication range, as shown on the bottom of

Fig. 5, the LED blue light is adopted for transfer- FIGURE 5. Proposed software-defined hybrid data transfer framework.
ring the data from node/C-AUV u (IP: 10.0.0.1)
to node/S-AUV b (IP: 10.0.0.4) along the path
<u, v, a, b>. The flow entries for scheduling the 6a–6j, we can see that each AUV’s cruising path
traffic are shown in the red squares on the bottom can be exactly scheduled and closely follows the
of Fig. 5. trajectory of AUV-LC, when the proposed scheme
is adopted. In particular, with the ability of central-
Simulation Results ized management, the cooperative control frame-
In this section, we present some simulation results work can be efficiently deployed to perform exact
to demonstrate the availability of the proposed obstacle avoidance. As a comparison, our previous
scheme in this article, especially in performing proposal requires strict conditions (e.g., each AUV
the cooperative underwater cruising (in order to acquires a global view of the entire network), but
perform ocean exploration). All the simulations performs worse, especially in the aspects of exe-
were conducted on an Intel® Core i7-8565U 1.8 cution efficiency and system stability. This is due to
GHz machine of 8 GM memory. We use Matlab that the work in [9] is based on a distributed con-
R2014b to simulate the underwater cruising in trol mechanism, and each AUV plans each own
an area of 300 m x 300 m, especially when the path ignoring the others. The results in Figs. 6a–6j
potential underwater obstacle avoidance is taken demonstrate that our proposal is easy to deploy
into account. In particular, the cruising or path and more efficient in planning the cruising trajecto-
planning policy is derived from the original work ry for the AUVs.
in [9] 4 which is also selected as the compared According to the results from Figs. 6a–6j, we
object. Instead of the distributed network control also test the distribution area of the AUVs, and the
mechanism as presented in [9], we utilize our pro- control delay 5 it takes the AUVs to achieve the
posal to improve the network scalability, leading path planning policies. Here, the acoustic propaga-
to a centralized network management plane. Due tion speed in underwater is fixed specified to 1450
to the limited space, all the simulation results are m/s, and the policy computing speed of the AUV
presented in the same figure. In Figs. 6a–6j, 20 in the distributed path planning scheme is assumed
C-AUVs/S-AUVs denoted by red circles are united to be 10 times smaller than the AUV-LC’s. In Fig.
managed by one AUV-LC (denoted by the red 6k, we can see that the distribution area acquired
star). As a comparison, another 20 blue circles by both of the policies are almost the same before
denote the AUVs following the distributed control 500 time slots. Then, the AUVs controlled by our
proposed in [9]. The solid green circles in Figs. proposal in this article are cruising closer than the
6a–6j represent the underwater obstacles, and compared objects. This demonstrates as well that
the arrows represent the speed vector. Note that our proposal in this article performs more efficient-
we utilize the proposed control framework from ly in path planning, that is, by our proposal, the
above by performing the potential underwater AUVs do not have to consume valuable energy to 4For more details about [9],
obstacle avoidance. We refer to the parameter revise the cruising path frequently. Furthermore, please refer to our statement
above.
settings in the work in [9], especially for the arti- from Fig. 6l, we can see that the control delay of
ficial potential field-based control policy. We take the distributed path planning is little smaller than 5 The control delay is defined
0.1 seconds as the time slot and refresh the result our proposal in this article, at a serious cost of bea- as the entire time from the
every 50 time slots while the beacon (BEA_SYN) con number.6 This is because the distance between information collection (from
period is 10 time slots. The initial speed of all the AUV-LC and the controlled AUVs is farther than the the AUVs) to path planning
policy deployment.
AUVs is set to 1.4 m/s. distance between the controlled AUVs. Although
We first show the path planning procedure, our proposal will bring additional communication 6 We compute the number
when the proposed scheme is deployed and com- cost between AUV-LC and the controlled AUVs, of all the beacons during the
pared with the policy in [9]. As shown in Figs. our proposal is more suitable to the case where cruising.

IEEE Wireless Communications • June 2020 137

Authorized licensed use limited to: Thapar Institute of Engineering & Technology. Downloaded on March 29,2024 at 06:14:38 UTC from IEEE Xplore. Restrictions apply.
HAN_LAYOUT.indd 137 6/9/20 3:54 PM
 FIGURE 6. Test for the proposed framework for exact path planning.

large-scale AUVs are deployed. This is because the cooperative control for multi-AUVs, and intelli-
communication cost between the controlled AUVs gent data transfer scheduling, respectively. Simu-
will keep an exponential growth, when the scale lation results show that the proposed SDN-based
of the controlled AUVs is large. This will dramati- underwater searching scheme performs more
cally decrease the performance of the distributed efficiently than the normal distributed approach,
path planning scheme, while our proposal will not especially in the aspect of cooperative underwa-
do so. Furthermore, we can infer that the control ter cruising. Possible future open issues derived
delay of our proposal in this article will decrease from this article are as follows.
when the AUV-LC is keeping the right range from
the AUVs. sMArt dAtA LAyer
Unlike the data layer in a traditional network as
concLusIons presented in [11], the proposed data layer in this
In this article, we have investigated a number article is not only in charge of the data transfer
of cooperative underwater search (or surveil- scheduling, but also is responsible for adjusting
lance) issues in underwater services. We first the attitude of AUVs by controlling thrusters,
proposed an SDN-based distributed architecture actuators, and ballast for resurfacing. Therefore,
for AUV-based UWNs, which partitions the net- designing open APIs to support smart network
work architecture into three layers and provides control/intelligent strategy deployment for the
fine-grained network control. Based on the pro- local control layer is a crucial issue to solve;
posed network architecture, we proposed an
SDN-based underwater cooperative searching securIty MechAnIsM
scheme. The proposed framework adopts the Due to the high bit error rate and unreliable opti-
SDN features and supports the ability of network cal/acoustic channel in the underwater environ-
synchronization, node localization estimation, ment, UWNs may face several types of malicious

138 IEEE Wireless Communications • June 2020

Authorized licensed use limited to: Thapar Institute of Engineering & Technology. Downloaded on March 29,2024 at 06:14:38 UTC from IEEE Xplore. Restrictions apply.
HAN_LAYOUT.indd 138 6/9/20 3:54 PM
 attacks (e.g., bad mouthing attack, on-off attack, Biographies
blackhole attack and wormhole attack), which Chuan Lin [S’17] (chuanlin1988@gmail.com) is currently a post-
have numerous deleterious effects on network doctoral researcher with the Key Laboratory for Ubiquitous
Network and Service Software of Liaoning Province, Dalian
control. Therefore, exploring the security mecha- University of Technology, Dalian, China. He received the B.S.
nism by utilizing SDN features to guarantee the degree in computer science and technology from Liaoning
precise network control is also a possible future University, Shenyang, China in 2011, the M.S. degree in com-
issue that needs to be addressed. puter science and technology from Northeastern University,
Shenyang, China in 2013, and the Ph.D. degree in computer

Acknowledgments architecture in 2018. His research interests include UWSNs,

industrial IoT, and software defined networking.
This work is supported by the National Natural
Science Foundation of China under Grant No. G uangjie H an [S’03, M’05, SM’18] (hanguangjie@gmail.
com) is currently a Distinguished Professor with the Col-
61971206; the China Postdoctoral Science Foun- lege of Engineering, Nanjing Agricultural University, Nan-
dation, No. 2019M661096; and the Fundamental jing, China, and a Distinguished Professor in the School of
Research Funds for the Central Universities, No. Software at Dalian University of Technology, Dalian, China.
DUT19ZD206, and No. DUT20RC(5)016. He received the Ph.D. degree from Northeastern Universi-
ty, Shenyang, China, in 2004. He is the author of over 330
References papers published in related international conference pro-
ceedings and journals, including the IEEE COMST, IEEE TII,
[1] M. Martínez-García, T. Gordon, and L. Shu, “Extended Cross-
IEEE TMC, IEEE TVT, IEEE TIE, and so on, and is the holder of
over Model for Human-Control of Fractional Order Plants,”
125 patents. He is a Senior Member of IEEE.
IEEE Access, vol. 5, Nov. 2017, pp. 27 622–35.
[2] G. Han et al., “An AUV Location Prediction-based Data Col- Mohsen Guizani [S’85, M’89, SM’99, F’09] (mguizani@ieee.org)
lection Scheme for Underwater Wireless Sensor Networks,” is currently a professor in the Department of Computer Science
IEEE Trans. Vehicular Technology, vol. 68, no. 6, June 2019, and Engineering, Qatar University, Qatar. Previously, he served in
pp. 6037–49. different academic and administrative positions at the University
[3] R. Diamant, L. Lampe, and E. Gamroth, “Bounds for Low of Idaho, Western Michigan University, the University of West
Probability of Detection for Underwater Acoustic Commu- Florida, the University of Missouri-Kansas City, the University of
nication,” IEEE J. Oceanic Engineering, vol. 42, no. 1, Jan. Colorado-Boulder, and Syracuse University. His research inter-
2017, pp. 143–55. ests include wireless communications and mobile computing,
[4] M. V. Jamali, J. A. Salehi, and F. Akhoundi, “Performance computer networks, mobile cloud computing, security, and smart
Studies of Underwater Wireless Optical Communication grid. He is currently the Editor-in-Chief of the IEEE Network Maga-
Systems with Spatial Diversity: MIMO Scheme,” IEEE Trans. zine. He is a Fellow of IEEE and a Senior Member of ACM.
Commun., vol. 65, no. 3, Mar. 2017, pp. 1176–92.
[5] S. Rani et al., “Energy Efficient Chain based Routing Proto- Yuanguo Bi [M’11] (biyuanguo@mail.neu.edu.cn) is currently a
col for Underwater Wireless Sensor Networks,” J. Network professor with the School of Computer Science and Engineer-
and Computer Applications, vol. 92, no. 8, Aug. 2017, pp. ing, Northeastern University, Shenyang, China. He received the
42–50. B.Sc. degree from Liaoning University, Shenyang, China in 2003,
[6] C. Wang, H.-Y. Yu, and Y.-J. Zhu, “A Long Distance Under- and M.Sc. (2006) and Ph.D. (2010) degrees from Northeastern
water Visible Light Communication System with Single Pho- University, Shenyang, China, all in computer science. He was a
ton Avalanche Diode,” IEEE Photonics J., vol. 8, no. 5, Oct. visiting Ph.d. student with the Department of Electrical and Com-
2016, pp. 1–11. puter Engineering, University of Waterloo, ON, Canada during
[7] N. Xia et al., “Localizability Judgment in UWSNs based on 2007-2009. He joined the School of Computer Science and Engi-
Skeleton and Rigidity Theory,” IEEE Trans. Mobile Comput- neering, Northeastern University, China, as an associate professor
ing, vol. 16, no. 4, April 2017, pp. 980–89. in 2010. In 2019, he was promoted to be a full professor. He has
[8] M. Martínez-García, Y. Zhang, and T. Gordon, “Pattern Iden- published 30 journal/conference papers and is a member of IEEE.
tification for Feedback Tracking Control in Human–Machine
Systems,” Human Factors, Oct. 2019, pp. 1–17. Jiaxin Du (dujiaxin@mail.dlut.edu.cn) is currently persuing the
[9] S. Shiliang et al., “Distributed Fault Detection and Isolation Ph.D. degree at the School of Software at Dalian University of
for Flocking in a Multirobot System with Imperfect Commu- Technology, China. She received the B.S. degree in software engi-
nication,” Int’l. J. Advanced Robotic Systems, vol. 11, no. 6, neering from Taiyuan University of Science and Technology, Taiyu-
June 2014, pp. 86–94. an, China, in 2016. Her current research interests include privacy
[10] S. Bera, S. Misra, and A. V. Vasilakos, “Software-Defined protection and security for underwater acoustic sensor networks.
Networking for Internet of Things: A Survey,” IEEE Internet
of Things J., vol. 4, no. 6, Dec. 2017, pp. 1994–2008. Lei Shu [M’07, SM’15] (lei.shu@ieee.org) is currently with the
[11] C. Lin et al., “A Scheme for Delay-Sensitive Spatiotempo- College of Engineering, Nanjing Agricultural University, Nanjing
ral Routing in SDN-Enabled Underwater Acoustic Sensor 210095, China, and a Lincoln Professor with the University of
Networks,” IEEE Trans. Vehicular Technology, vol. 68, no. 9, Lincoln, U.K. He received the B.Sc. degree in computer science
Sept. 2019, pp. 9280–92. from South Central University for Nationalities, China, in 2002,
[12] T. L. Saaty, “Decision Making with the Analytic Hierarchy the M.Sc. degree in computer engineering from Kyung Hee
Process,” Int’l. J. Services Sciences, vol. 1, no. 1, Jan. 2008, University, South Korea, in 2005, and the Ph.D. degree from
pp. 83–98. the Digital Enterprise Research Institute, National University of
[13] M. T. R. Khan et al., “An Energy-Efficient Data Collection Ireland, Galway, Ireland, in 2010. Until 2012, he was a specially
Protocol with AUV Path Planning in the Internet of Under- assigned researcher with the Department of Multimedia Engi-
water Things,” J. Network and Computer Applications, vol. neering, Graduate School of Information Science and Technol-
135, no. 6, June 2019, pp. 20–31. ogy, Osaka University, Japan. He is currently a Distinguished
[14] R. Fan et al., “A SDN-Controlled Underwater MAC and Professor with Nanjing Agricultural University, China, and a
Routing Testbed,” Proc. IEEE Military Commun. Conf., Balti- Lincoln Professor with the University of Lincoln, U.K. He is also
more, MD, USA, 2016, pp. 1071–76. the Director of the NAU-Lincoln Joint Research Center of Intel-
[15] Z. Huang , D. Zhu, and B. Sun, “A Multi-AUV Coopera- ligent Engineering. His main research fields are wireless sensor
tive Hunting Method in 3-D Underwater Environment with networks and Internet of Things. He has published over 400
Obstacle,” Engineering Applications of Artificial Intelligence, papers in related conferences, journals, and books in the areas
vol. 50, no. 4, Apr. 2016, pp. 192–200. of sensor networks and Internet of Things.

IEEE Wireless Communications • June 2020 139

Authorized licensed use limited to: Thapar Institute of Engineering & Technology. Downloaded on March 29,2024 at 06:14:38 UTC from IEEE Xplore. Restrictions apply.
HAN_LAYOUT.indd 139 6/9/20 3:54 PM