Available via license: CC BY 4.0
Content may be subject to copyright.
Academic Editor: Graciliano Nicolás
Marichal
Received: 15 December 2024
Revised: 30 December 2024
Accepted: 2 January 2025
Published: 4 January 2025
Citation: Xue, L.; Lei, H.; Zhu, R. A
Collision Avoidance MAC Protocol
with Power Control for Adaptive
Clustering Underwater Sensor
Networks. J. Mar. Sci. Eng. 2025,13, 76.
https://doi.org/10.3390/
jmse13010076
Copyright: © 2025 by the authors.
Licensee MDPI, Basel, Switzerland.
This article is an open access article
distributed under the terms and
conditions of the Creative Commons
Attribution (CC BY) license
(https://creativecommons.org/
licenses/by/4.0/).
Journal of
Marine Science
and Engineering
Article
A Collision Avoidance MAC Protocol with Power Control for
Adaptive Clustering Underwater Sensor Networks
Libin Xue 1, Hong Lei 1,∗and Rongxin Zhu 2,∗
1School of Cyberspace Security, Hainan University, Haikou 570228, China; xuelibin@hainanu.edu.cn
2School of Computer Science and Technology, Hainan University, Haikou 570228, China
*Correspondence: leihong@hainanu.edu.cn (H.L.); rongxin914@163.com (R.Z.)
Abstract: Underwater sensor networks (UWSNs) play a vital role in marine exploration
and environmental monitoring. However, due to the characteristics of underwater acous-
tic channels such as high delay, low bandwidth, and energy limitation, the design of an
underwater media access control (MAC) protocol has brought great challenges, and ex-
isting MAC protocol designs rarely consider the influence of channel interference factors
in networking. Therefore, this paper proposes a collision avoidance MAC protocol for
clustering underwater sensor networks. The protocol first classifies users by combining
the channel characteristics of underwater nodes and the distance measurement between
nodes. Then, based on the clustering network, according to the channel correlation distance
measurement between nodes and the communication range of the cluster head (CH), the
transmit power in clusters is controlled to reduce the lifetime of the network based on the
cumulative reduction in node power consumption. Finally, the cluster structure in each
cluster is used to schedule the transmission of member nodes in the cluster, and at the same
time, the energy consumption of nodes is reduced while multi-node collision-free transmis-
sion is realized. The simulation results show that the throughput of the proposed adaptive
power control clustering MAC protocol (APCC-MAC) is 26.5% and 19.5% higher than
that of packet-level slot scheduling (PLSS) algorithm and Cluster-Based Spatial–Temporal
Scheduling (CSS) algorithm, respectively, providing better communication performance
and stability for clustered underwater acoustic networks.
Keywords: media access control (MAC); power control; collision avoidance; clustering
algorithm; game theory
1. Introduction
In recent years, with the increase in marine activities and the rapid development
of marine industries, underwater acoustic sensor networks (UASNs) [
1
] have received
widespread attention and research in marine exploration, environmental monitoring,
and other fields [
2
,
3
]. However, marine issues such as ships, waves, ocean noise, and en-
vironmental pollution are becoming increasingly serious. Moreover, sound propagation
speed is relatively slow, at about 1500 m/s [
4
], which is five orders of magnitude slower
than radio signals [
5
–
7
]. This leads to long propagation delays due to the slow signal
propagation speed. In addition, the underwater communication bandwidth is limited to
less than 15 kHz, which restricts the amount of data that can be transmitted. Furthermore,
underwater sensors are powered by batteries that are difficult to replace, posing a critical
issue for limited underwater energy consumption.
Moreover, due to significant signal propagation delays underwater, underwater acous-
tic communication faces the issue of spatiotemporal uncertainty [
8
], where collisions can
J. Mar. Sci. Eng. 2025,13, 76 https://doi.org/10.3390/jmse13010076
J. Mar. Sci. Eng. 2025,13, 76 2 of 18
still occur even when data packets are sent in different time slots and locations. The severe
access conflicts significantly affect the performance of the network.
In large-scale networks, when the propagation distance is limited, a single node cannot
handle the scheduling of the network. Therefore, a clustering network structure is usually
adopted. For example, reference [
9
] constructs a clustered spatiotemporal conflict graph to
schedule nodes in the cluster to reduce conflicts. However, due to the high complexity of
the graph structure, the utilization of channel time slots is reduced.
Game theory provides an effective means to study the most reasonable energy-efficient
scheme among nodes in clusters through competition and cooperation. However, the exist-
ing game theory only focuses on the current energy consumption, but ignores the influence
of complex underwater channel environments on the energy state between nodes; hence, it
cannot accurately reflect the game scenario in actual underwater acoustic networks [
10
,
11
].
In addition, the energy consumption during data transmission is closely related to the
transmission distance [
12
]. Therefore, when designing MAC protocols, the transmission
power of nodes should be controlled to reduce the energy consumption of nodes, reduce
interference between nodes, and avoid collisions.
To address the above issues, this paper proposes a novel underwater MAC protocol.
This protocol adopts an adaptive clustering network based on game theory and improves
the performance of network concurrent transmission through a transmission scheduling
mechanism based on cluster head nodes. Meanwhile, nodes adjust their power to change
the communication distance, reducing network energy consumption and further reducing
data conflicts.
The main contributions of this paper are summarized as follows:
1.
A power control and collision avoidance MAC protocol for UASNs based on adap-
tive clustering is proposed. It optimizes the power control of nodes on the basis
of clustering, thus providing more residual energy for network clustering in the
next cycle. After several cycles, the whole network has a longer lifetime than other
clustering algorithms.
2.
The adaptive clustering algorithm proposed in this paper takes into account the
actual distance between nodes and channel correlation in the water-based multi-path
channel environment, and designs a penalty mechanism to encourage nodes to make
decisions that are more beneficial to the overall income.
3.
Based on the above cluster network structure, the protocol combines the channel
correlation distance measurement between nodes, and adjusts the transmitting power
of nodes inside and outside the cluster to reduce the data conflict and node power
consumption between nodes.
4.
On the basis of the above power control, the protocol schedules the data transmission
order according to the sending priority, considering the CH degree, distance measure,
and propagation delay of the nodes, so as to further avoid the collision between
the nodes.
The rest of the article is organized as follows: In Section 2, we will mainly introduce the
related research of the proposed MAC protocol. In Section 3, we will introduce the adaptive
clustering algorithm proposed in this paper in detail. In Section 4, we will introduce the
proposed APCC-MAC protocol in detail. Then, in Section 5, we will conduct a simulation
experiment, and in Section 6, we will provide a prospect and summary.
2. Related Work
Currently, traditional MAC protocols include three main categories: scheduling-based
MAC protocols, competition-based MAC protocols, and a combination of both.
J. Mar. Sci. Eng. 2025,13, 76 3 of 18
The ALOHA protocol is the earliest contention-based MAC protocol [
13
], where
nodes choose to send immediately whenever they have data to transmit, leading to severe
collisions as the network load increases. Reference [
14
] analyzed an improved the slotted
ALOHA protocol with time slot constraints. Although slotted ALOHA reduces collisions by
allocating time slots, collisions still increase with growing traffic load. Therefore, researchers
began studying handshake-based contention MAC protocols. Reference [
15
] proposed
a collision avoidance MAC protocol based on game theory under the premise of a node
handshake mechanism. This protocol uses non-cooperative games between nodes, allowing
the node with the shortest underwater delay to win the competition with the highest
probability before starting data transmission, thereby increasing network throughput.
However, the protocol adopts a one-handshake-one-send mechanism, which prolongs the
overall transmission time of all data in the harsh underwater environment with long delays,
reducing the overall efficiency of data transmission.
Scheduling-based MAC protocols are mainly divided into three types: Time Divi-
sion Multiple Access (TDMA), Frequency Division (OFDM), and Code Division (CDMA).
TDMA achieves conflict-free transmission by scheduling nodes to send data in different
time slots. Reference [
16
] designed a TDMA protocol that allocates time slots to different
user groups based on slot grouping levels. However, such protocols sacrifice time resources
and increase the delay as nodes must wait for their time slots to be allocated. To over-
come the above-mentioned latency issues, some experts utilize physical layer frequency
division multiplexing modulation technology to improve data transmission rates [
17
].
Reference [
18
] proposes an MAC protocol that continuously adjusts transmission power
and OFDM subcarriers based on dynamic changes in traffic load, effectively reducing data
concurrency conflicts. Reference [
19
] presents an MAC protocol that combines CDMA
and OFDM to achieve concurrent communication in wide-area networks. While these
protocols address the long latency issues underwater, they result in sacrificing and wasting
spectrum resources.
To better overcome the drawbacks of scheduling-based MAC protocols, large-scale
node scheduling and clustering algorithms have gradually become a research focus for
improving network performance [
20
,
21
]. The classification and hierarchical ideas in clus-
tering algorithms reduce the complexity of underwater networks and the number of data
transmissions for user nodes. Among them, the most representative clustering algorithm
is the LEACH protocol [
22
], which achieves energy balance by periodically re-clustering.
However, the setting of random numbers in the algorithm eventually leads to uneven
energy consumption.
The authors of [
23
] proposed an underwater adaptive energy-saving clustering al-
gorithm based on a multi-dimensional game. By establishing a multi-dimensional game
model, it increases the chance for historical CH nodes to compete with CH again, and fur-
ther optimizes the CH election strategy. However, this algorithm rarely considers the state
and environmental factors of the nodes themselves, so it is not highly adaptable in real
scenarios. Based on clustering, more efficient MAC protocols have been proposed, such
as the TDMA-based enhanced MAC protocol in reference [
24
] and the handshake-based
power control MAC protocol in reference [
12
]. These protocols all use intra-cluster or
inter-cluster hierarchical scheduling to achieve network efficiency. However, these algo-
rithms still have many flaws; for example, the clustering process pays little attention to
the conditions of the underwater channel, and the overall goals of clustering and MAC
protocols are not consistent.
Therefore, an adaptive clustering underwater acoustic network MAC protocol is
proposed in this paper. This protocol clusters underwater users while fully considering
channel correlation, and adjusts the transmission power range of nodes and the priority
J. Mar. Sci. Eng. 2025,13, 76 4 of 18
of scheduling users to send data within and between clusters after clustering, in order to
combine clustering and MAC protocols to jointly achieve the goal of reducing data conflicts.
3. Adaptive Clustering Algorithm Based on Game Theory
As shown in Figure 1, nodes are randomly distributed in a certain sea area, clustered
using the clustering algorithm proposed in this paper to form a single-hop network from
member nodes to cluster heads. After ordinary member nodes collect ocean data, they send
data packets to the cluster head, which then transmits the received data to the surface base
station through multi-hop communication.
Figure 1. System model.
3.1. Underwater Channel Model and Distance Measurement
There are two main losses when acoustic signals propagate underwater: absorption
loss and propagation loss. If
r(m)
is the transmission radius of the node signal within its
communication range,
a0(f)
(dB/km) is the absorption factor, and
f
(kHZ) is the trans-
mission frequency, then the attenuation formula of the acoustic signal can be expressed as
follows:
a0(f) = 0.011 f2
f2+1+4.4 f2
f2+4100 +2.75 ∗10−4f2+0.003 (1)
Furthermore, the transmission loss of the acoustic signal can be expressed as fol-
lows [25]:
A(r,f) = Lsp +Lab =10 logrka0(f)r×10−3=A0rka0(f)r×10−3(2)
In the above formula, the product of coefficient 10 and the
log
function are replaced
by the scaling constant A0. Then, the channel gain Hcan be further expressed as follows:
H=1
A0(A)(3)
J. Mar. Sci. Eng. 2025,13, 76 5 of 18
In the underwater channel, there are many types of noise, such as wind waves,
tides, ship navigation, etc. In this paper, the sum of underwater communication noise is
represented as σ2, and the signal-to-noise ratio can be calculated as follows:
SNRi=PiH
σ2(4)
By combining Equations (2)–(4), we obtain the following:
Pi= (SNR)σ21
A0hrka0(f)r×10−3i(5)
As can be seen from Formula (5), user i can use the transmission power required for
SNR communication, that is, the minimum power necessary to ensure normal communi-
cation between network nodes. If the transmission power is too large, interference and
communication conflict between communication nodes will be caused.
Therefore, according to the distance between nodes, the power of the control nodes
is in a range, which can improve the quality of underwater acoustic communication, and
avoid collisions between nodes.
The phenomena of reflection and refraction when sound waves propagate in the
seabed will cause the sound signal to form multi-path effects in the way of multi-path
superposition at the receiving end, resulting in inter-symbol interference, which seriously
affects the quality of underwater acoustic communication.
Therefore, when we consider the channel conditions of node clustering, we must
consider the multi-path channel information of node data transmission.
The distance measurement is as follows:
Suppose that the channel information of user
i
can be marked as follows:
hi=
[hi1
,
hi2
, ...
hil ]
, then the auto-correlation matrix of user i can be expressed as
Rii
= E{
hihhi
},
and through the feature decomposition of the auto-correlation matrix,
Rii
can be expressed
as follows:
Rii Si=ΛSh
i(6)
If the eigenvalues in Formula (6) are regarded as coefficients, then the multi-path
channel information can be represented by the eigenvector rias follows:
ri=
l
∑
i=1
λil Sil (7)
Next, the channel correlation between the cluster head and the common member node
i can be calculated as
Ei,CH =Enrh
irCH o
Erh
iri∥Erh
CH rC H (8)
Then, the distance metric that integrates the actual distance from the node to the CH
and the channel state is designed as follows:
Di,CH =βri,C H + (1−β)(−Ei,CH)(9)
where
β∈(
0, 1
)
is the weight factor.
ri,CH
is the actual direct distance between CH and
node i. According to the result obtained by Formula (9), any node in the network can be
assigned to the cluster with the lowest D value from the other nodes.
J. Mar. Sci. Eng. 2025,13, 76 6 of 18
3.2. Node Traffic Weight Analysis
Assume that the traffic of node i is x(i), considering the influence of each node on clus-
ter head nodes. Assuming that the weight coefficient of traffic is
θ∈
(0,1), and considering
the channel distance measurement between nodes, the comprehensive weight generated
by member node i on CH nodes is as follows:
yi=θx(i)
Di,CH
(10)
The weight of the cluster head can be taken as the weight of the cluster head by
summing the traffic weight generated by all nodes in the cluster to CH, which is denoted
as Z:
Z=
N
∑
i=1
θx(i)
Di,CH
(11)
3.3. Node Energy Consumption Model
According to the underwater channel model formula, the energy consumption for a
node to send l−bit data packets is as follows:
Esent (l, r) = P∗A(r, f)∗T (12)
where P is the received power of node i, T is the transmission delay of i, and
Zi
is the traffic
weight of i; then, the energy consumption of node i for receiving l−bit data is as follows:
Ereceived(l) = lZiE1(13)
The energy consumption of the converged l−bit data on node i is as follows:
Eintegrated(l) = lZiE2(14)
The energy consumption of data forwarding by CH node i can be expressed as follows:
Eforward (l) = ηlZiE3+εDE4(15)
Non-cluster head (NCH) node i consumes energy only when forwarding data, which
is as follows:
ENCH
forward (ηlZi,D)=ηlyiE3+εDE4(16)
3.4. Game Model Between Nodes
Define underwater nodes as a set of competing participants: N = {
N1
,
N2
,......,
Nn
}.
The strategy set of each node at the current time of game is as follows: S = {
S1
,
S2
,......,
Sn
}.
The revenue function set of game nodes is expressed as follows: U = {U1,U2,......,Un}.
The proposed adaptive clustering algorithm consists of three steps: Firstly, based on
the distance measurement of channel correlation and node traffic analysis, the game model
between nodes is established. Secondly, on the basis of setting the penalty factor, according
to the Nash equilibrium obtained by the non-cooperative game, a node i is determined to
become a CH node and form a cluster. Finally, after a period of cycles, CH nodes are rotated
according to the maximum remaining energy of nodes in the cluster to balance the energy
consumption. The overall flow chart of the algorithm is shown in Figure 2; as can be seen
from the figure, the number of times to judge the income of each node i is
n
times, and the
number of times to further judge whether the residual energy of the node is suitable for
CH is also ntimes. Therefore, the time complexity of the clustering algorithm is O(n2).
J. Mar. Sci. Eng. 2025,13, 76 7 of 18
Figure 2. Adaptive clustering algorithm flow.
The detailed steps of the algorithm are as follows:
When node i selects strategy Si= CH, its return function is as follows:
Ui(NCH,S) = aEresidual
i−Ci(17)
where a> 0 is the positive correlation coefficient,
Ci
is the cost, and
Eresidual
is the remaining
energy of node I. According to the energy consumption model, cost
Ci
can be further
expressed as follows:
J. Mar. Sci. Eng. 2025,13, 76 8 of 18
Ci=lZiE1+lZiE2+Eforward (lZi,D)
=lZi((E1+E2)+ηlZiE3+εDE4
(18)
If the node selects NCH, since the non-cluster head node only has the energy con-
sumption of data forwarding, the income of node i can be expressed as follows:
Ui(NCH,S) = aEresidual
i−ENCH
forward (ηlZi,D)
=aEresidual
i−ηlyiE3−εDE4
(19)
Obviously, when the residual energy of node i is the same, the cost of choosing the
NCH strategy is far less than the cost of choosing the CH strategy, and the benefit is also
far greater. Therefore, it is necessary to impose certain penalties on nodes that choose the
NCH strategy, so as to promote more nodes to compete for the cluster head. The penalty
ri
can be expressed as follows:
ri=aCibEresidual
i
Emax
j
(20)
After penalizing the node, the payoff is as follows:
Ui(Si=NCH)=aEresidual
i−ηlyiE3−εDE4−ri(21)
where
Emax
j
is the energy of the node with the largest residual energy among the neighboring
nodes. It can be seen from Formula (20) that when the residual energy of node i is closer to
the maximum energy of the neighboring node with the largest residual energy, the ratio
Eresidual
i
/
Emax
j
of the two is closer to 1, meaning the penalty on node i increases. At this time,
node i is more inclined to choose the strategy of becoming a CH node in order to obtain
greater benefits for itself. In other words, the function of the penalty
ri
is to encourage the
nodes with the highest energy in the cluster, as much as possible, to elect the cluster head
with the greatest probability in the process of the game.
To sum up, in the course of a round of game, nodes with greater energy tend to become
cluster heads and thus obtain the maximum benefits. Nodes with lower energy receive
smaller penalty values
ri
during the course of the game, and Formula (20) can be used to
show that the strategy of choosing NCH has greater benefits. Ultimately, an equilibrium
state is reached during the game between the nodes. After a certain period of time, the node
that originally chose the CH strategy has a rapid decline in energy due to the excessive load,
while other ordinary nodes only send data, and the long-term low energy consumption
leads to the energy of an NCH node with the largest residual energy becoming larger than
the previous CH, thus replacing the original cluster head and becoming a new cluster head.
Through the continuous cycle of this process, finally, the energy consumption of the whole
cluster can be balanced.
4. APCC-MAC Protocol Design
4.1. Protocol Overview
The APCC-MAC protocol proposed in this paper adopts a large-scale underwater
network as the model, aiming to ultimately avoid packet collision conflicts, reduce node
latency, and reduce power consumption. The detailed flow chart of this protocol is shown
in Figure 3, with specific steps as follows:
J. Mar. Sci. Eng. 2025,13, 76 9 of 18
Figure 3. Flow chart of the APCC-MAC protocol.
1. Initial network: In this process, each node collects information about its neigh-
bors within a two-hop range and maintains a delay information table locally through
packet switching.
2. Members use the adaptive clustering algorithm based on game theory proposed in
this paper to cluster and calculate the CH degree of each node.
3. Determine whether the communication at this moment involves the data transmis-
sion of nodes in the cluster. If not, the data will be transmitted to the sink node through
multi-hop communication via CHs. If it does, the transmit power of nodes in the cluster will
be controlled according to the cooperative game algorithm proposed in the literature [
26
],
so as to reduce the power consumption of nodes and allocate optimal power resources for
the subsequent scheduling of the MAC protocol.
4. The packet structure of the algorithm in this paper adopts the data frame structure
shown in Figure 2in the literature [
27
]. After the CH node allocates transmission power
to all its members, the priority of transmission scheduling is determined according to the
J. Mar. Sci. Eng. 2025,13, 76 10 of 18
CH degree, distance measure, and propagation delay of the node, and the scheduling
information is encapsulated in a short packet for broadcast.
5. After the completion of scheduling, the data enter the transmission phase. Af-
ter the transmission is completed, CH broadcasts an ACK confirmation packet of the data
transmission results and continues to the next round of transmission.
4.2. Power Control in the Protocol
The power control of the protocol can be divided into the power control of the cluster
head node and the power control of the common member nodes; furthermore, the power
control of the cluster head nodes can be divided into intra-cluster communication and
inter-cluster communication.
As shown in Figure 4a,
H1
, B, and
H2
, C are two clusters, where
H1
and
H2
are cluster
heads. When
H1
sends signals to B and
H2
to C, respectively, collisions may occur between
nodes
H1
and
H2
, and between
H2
and B. If the transmitting power of cluster head nodes
H1
and
H2
is within a certain range, as shown in Figure 4b, collisions between nodes
H1
and
H2
, as well as between nodes
H2
and B, can be avoided, thus avoiding collisions between
clusters through power control.
Figure 4. Power control and collision avoidance between clusters.
The power control between nodes in the cluster is shown in Figure 5a, where H is the
cluster head, and member nodes
N1,1
and
N1,2
in the same cluster are all in the collision area
with each other. After controlling the power, as shown in Figure 5b, nodes
N1,1
and
N1,2
reduce their transmit power without affecting their normal communication with cluster
head node H. Thus, collisions with each other are avoided, and the power consumption of
nodes is also reduced.
Figure 5. Power control and collision avoidance in clusters.
When cluster head node H performs intra-cluster communication, combined with the
distance metric D, it is assumed that the minimum power that can normally communicate
with the farthest node in the cluster is
Pmin
; in this cluster, for more specific power control
J. Mar. Sci. Eng. 2025,13, 76 11 of 18
of each node, the power control algorithm in my recently published work [
26
] is further
adopted. When H performs inter-cluster communication, the maximum power that it can
use to communicate with the furthest cluster head is
Pmax
; then, the range of intra-cluster
and inter-cluster transmitting power should be adjusted as follows:
Pmin <Pth(1−β)<Pmax (22)
4.3. Node Scheduling
When the power of the nodes in the protocol is controlled, the protocol uses the
cluster head nodes in each cluster to complete the scheduling of its cluster member nodes.
Before scheduling, it is necessary to confirm the priority of the node to be scheduled. Here,
the three factors of CH degree, distance measurement, and traffic weight of the node are
mainly considered to determine the priority. The detailed process is as follows:
1. CH degree: As shown in Figure 6, nodes
N2,1
have collision relations with cluster
heads
H1
,
H2
, and
H3
of the three clusters, so their CH degree is 3; nodes
N1,1
, within their
propagation range, only have conflict relations with
H1
and
H2
, so its CH degree is 2.
Therefore, a node with a higher CH degree has the highest probability of collision and
should have a higher priority.
Figure 6. Calculation of the CH degree of nodes.
2. Distance measurement: If nodes have the same cluster head degree, it is necessary
to further determine the priority of nodes to be scheduled by using distance measurement
D in Formula (9) in Section 3. Because of the smaller
Di,CH
of the node i, which represents
the minimum compromise between channel correlation and the actual distance between
nodes and cluster heads, we hope that such nodes can obtain greater priority.
3. Node delay: After the network is clustered, each node collects delay information
within its propagation range and broadcasts its position and cluster head information
to the network. After receiving the information, the CH creates a general table of the
delay information of nodes in its cluster, adjusts its own transmission power based on
the farthest location of nodes in the cluster, and broadcasts the delay information table.
After receiving the broadcast, all member nodes in the cluster maintain the information
state table containing the delay information of all nodes in the cluster locally.
It can be seen from Figure 7that the CH node schedules the node transmission process
according to priority. Suppose that from Node1 to Node4, the CH degrees of the four
nodes are 1, 2, 1, and 1. The size relationship of the four nodes in the delay information
J. Mar. Sci. Eng. 2025,13, 76 12 of 18
table is
TNode2,h<TNode1,h<TNode4,h<TNode3,h
; the size relationship of the four distance
measures is
Dnode1,CH
=
Dnode2,CH <Dnode3,CH
=
Dnode4,CH
. At this time, as Node2 has the
largest cluster head degree, it obtains the highest sending priority. Node1 and Node4
have the same cluster head degree, but Node1 has a smaller distance metric than Node4.
Therefore, Node1 has a higher priority than Node4. Compared with Node3 and Node4,
although the cluster head and distance measurements are the same, Node4 is given higher
scheduling priority than Node3 because its delay is smaller than that of Node3. In summary,
the scheduling priorities of these nodes are in the following order: Node2, Node1, Node4,
and Node3. When the cluster head completes the scheduling sequence of nodes through
the control package, it enters the data transmission stage.
Figure 7. Scheduling priority of the nodes.
5. Simulation and Performance
In this chapter, we use the Aqua-sim-tg simulator based on ns-3, combined with real
ocean data, to simulate real ocean scenarios. The specific scenario settings are shown in
Figure 8. A random and uniform deployment of nodes is carried out within a 3D cubic
area measuring 5*5*6, with the network consisting of a total of 30 to 50 nodes. Each node
is equipped with its own location information and a unique id identifier. An energy-
unconstrained sink node is strategically deployed on the surface of this area. The weight
factor
α
for the channel correlation distance metric is assumed to be 0.6, the maximum
power
Pmax
of inter-node communication is 90w, and the node traffic weight is adjustable
within the range of
θ∈
(0,1). The length of control packets is set at 180 bits, while data
packets are 1024 bits in length, and the scheduling period of the cluster head node is 0.07 s.
Transmission power can be adjusted to meet the requirements of communication distance.
The duration of a single simulation run is established at 12,000 s. The complete parameter
settings are presented in Table 1.
Table 1. Simulation parameter.
Parameter Value
Network size 6000 ×6000 ×6000 (m)
Sink location (2500, 2500, 6000)
Node size 30 to 50
Data packet size 1024 bits
Control packet size 180 bits
Modulation mode bpsk
Transmission rate 1 kbps
Transmission bandwidth 10 kHZ
J. Mar. Sci. Eng. 2025,13, 76 13 of 18
Table 1. Cont.
Parameter Value
Carrier frequency 20 kHZ
E180 nJ/bit
Transmitter power 2 to 90 w
E230 nJ/bit
η0.65
α0.6
θ0 to 1
Simulation time 12,000 s
Figure 8. Network topology scenario.
5.1. Comparison of User Clustering Algorithms
Here, we first verify the effectiveness of the proposed clustering algorithm in terms of
performance indicators such as throughput and power consumption, and compare and an-
alyze three algorithms: elbow method clustering (EC) [
28
] based on distance measurement,
the random clustering algorithm (RC) [
29
], and the APCC clustering algorithm proposed
in this paper. The throughput in Figures 9and 10 below is normalized.
As shown in Figure 9, at the beginning, due to the very low traffic in the network,
the three protocols show almost the same throughput. However, as the network load
increases, the throughput of the EC algorithm gradually exceeds that of the RC algorithm.
This is because the RC algorithm only clusters randomly without considering the impact
of the underwater channel environment on the nodes between clusters, requiring higher
transmission power to avoid interference caused by the lack of correlation in channel classi-
fication. The EC algorithm classifies user nodes using distance measurement, effectively
J. Mar. Sci. Eng. 2025,13, 76 14 of 18
reducing intra-cluster interference. However, the EC algorithm does not further control the
transmission power of the nodes in the cluster, so the communication quality needs to be
improved by sacrificing the transmission rate to avoid the interference between nodes.
Figure 9. Throughput of users under different clustering protocols.
In the APCC algorithm, the node energy consumption of the whole network is bal-
anced by clustering through an inter-node game. Based on the clustering network and the
distance measurement of channel correlation, the cooperative game algorithm proposed in
the literature [
26
] is used in each cluster to control the power of the member nodes in the
cluster, thus further reducing the power consumption of the nodes. Therefore, as shown
in Figures 9and 10, with the increase in network traffic, the performance of the clustering
algorithm proposed in this paper, such as throughput and energy consumption, is superior
to the other two algorithms.
Figure 10. Energy consumption of users under different clustering protocols.
J. Mar. Sci. Eng. 2025,13, 76 15 of 18
5.2. Throughput Comparison with Other MAC Protocols
Then, we combine the proposed APCC-MAC protocol with a new time slot ALOHA
(SA) [
14
], the PLSS algorithm [
16
], and the CSS algorithm [
9
]. Three advanced MAC
protocols are compared. As can be seen from Figure 11, the throughput of the SA protocol
shows a trend of first increasing and then decreasing. This is because when the network
load is low, nodes can efficiently and freely compete to send data, thereby obtaining
increasing throughput. However, when the network load increases, more and more nodes
compete under the time slot mechanism. This can still lead to increased data conflicts and
decreased throughput. PLSS performs better than SA in the process of increasing load due
to the time difference field added to the packet. Through the construction of a space–time
conflict graph, CSS achieves a throughput slightly higher than the PLSS protocol; however,
it does not consider the interference caused by the power size to the node like the APCC
protocol, so its throughput is much smaller than the APCC protocol.
Figure 11. Throughput of different protocols with different number of nodes.
5.3. End-to-End Delay
As shown in Figure 12, PLSS is essentially a scheduling MAC protocol with fixed slot
allocation, and the delay increases with the increase in network load. When the network
traffic is small, the delay of the SA protocol is also small, but with the increase in network
traffic, its delay increases faster than that of the PLSS protocol. This is due to its competition-
based mechanism, which causes sending conflicts between packets to increase dramatically
as the network load increases. The end-to-end delay changes in the APCC protocol and
the CSS protocol are similar, both of which decrease with the increase in the number of
nodes. This is because both of them adopt the method of intra-cluster node scheduling
and transmission, which results in a decrease in transmission time as network data packets
increase. In the case of a small traffic load, the CSS protocol can make every node in the
network obtain a conflict-free continuous allocation scheme by constructing a space–time
conflict graph to fix the time slot, so its transmission delay is slightly lower than that of
APCC-MAC proposed in this paper. However, with the increase in network load, the CSS
protocol can reduce the traffic load. This fixed allocation scheme, which has a high degree
of correlation with node location, cannot adapt to the dynamic network topology, so its
delay is gradually longer than that of the proposed protocol. On the other hand, the priority
scheduling algorithm adopted by the APCC-MAC protocol can prioritize the scheduling
J. Mar. Sci. Eng. 2025,13, 76 16 of 18
and transmission of nodes that are more prone to conflicts on the basis of controlling power,
thus reducing the collision of nodes in the cluster to a greater extent, and thus obtaining
better performance than the CSS protocol.
Figure 12. End-to-end delay changes under different loads.
6. Conclusions
UWSNs have limited bandwidth and long delays, which bring great challenges to
underwater acoustic networks. To solve the problems of energy limitation and node conflict
in underwater acoustic networks, a new APCC-MAC protocol is proposed in this paper.
Firstly, an adaptive clustering algorithm based on a non-cooperative game is designed
to establish a clustering network that is more suitable for lower-channel communication.
Secondly, based on the clustering underwater acoustic network, the transmitting power
of nodes under different distance measures is controlled to reduce power consumption
and extend the lifetime of the whole network. Finally, on the basis of clustering and power
control, the cluster head node dispatches the member nodes to avoid data conflict and
reduce the energy consumption of the whole network. Simulation results show that the
proposed APCC-MAC protocol has higher throughput, lower energy consumption, and
lower latency than existing protocols. However, this paper only verifies the validity of the
proposed protocol in a simulation environment.
In future work, the efficiency of the protocol will be further verified by real underwater
environment experiments, and the low-level implementation of the protocols will be
explained in more detail. Furthermore, more specific presentations of node distance
proximity will be considered.
Author Contributions: Conceptualization, L.X.; methodology, L.X.; software, R.Z. and H.L.; val-
idation, R.Z. and H.L.; formal analysis, L.X.; investigation, L.X. and R.Z.; data curation, L.X.;
wr
iting—revie
w and editing, R.Z. and L.X.; supervision, H.L. All authors agree on the published
version of the manuscript.
Funding: This work was supported in part by the National Natural Science Foundation of China
under Grant 62163011, in part by the major science and technology plan of Hainan Province under
Grant ZDKJ2021052.
Institutional Review Board Statement: Not applicable.
J. Mar. Sci. Eng. 2025,13, 76 17 of 18
Informed Consent Statement: Not applicable.
Data Availability Statement: Data are contained within the article.
Conflicts of Interest: The authors declare no conflicts of interest.
References
1.
Zia, M.Y.I.; Poncela, J.; Otero, P. State-of-the-art underwater acoustic communication modems: Classifications, analyses and
design challenges. Wirel. Pers. Commun. 2021,116, 1325–1360. [CrossRef]
2.
Islam, K.Y.; Ahmad, I.; Habibi, D.; Waqar, A. A survey on energy efficiency in underwater wireless communications. J. Netw.
Comput. Appl. 2022,198, 103295. [CrossRef]
3.
Campagnaro, F.; Francescon, R.; Coccolo, E.; Montanari, A.; Zorzi, M. A software-defined underwater acoustic modem for
everyone: Design and evaluation. IEEE Internet Things Mag. 2023,6, 102–107. [CrossRef]
4.
Zhao, S.; Han, Y.; Liu, Q.; Huang, H. Detecting moving targets in active sonar echograph of harbor environment using high-order
time lacunarity. J. Acoust. Soc. Am. 2020,147, 2110–2120. [CrossRef] [PubMed]
5.
Zhou, J.; Wang, X.; Zhang, B. Dynamic timeslot MAC protocol for AUV underwater communication. In Proceedings of the 2020
39th Chinese Control Conference (CCC), Shenyang, China, 27–29 July 2020; IEEE: Piscataway, NJ, USA, 2020; pp. 5236–5240.
6.
Zhu, R.; Huang, X.; Huang, X.; Li, D.; Yang, Q. An on-site-based opportunistic routing protocol for scalable and energy-efficient
underwater acoustic sensor networks. Appl. Sci. 2022,12, 12482. [CrossRef]
7.
Zhu, R.; Jiang, Q.; Huang, X.; Li, D.; Yang, Q. A reinforcement-learning-based opportunistic routing protocol for energy-efficient
and Void-Avoided UASNs. IEEE Sens. J. 2022,22, 13589–13601. [CrossRef]
8.
Liu, X.; Du, X.; Li, M.; Wang, L.; Li, C. A MAC Protocol of Concurrent Scheduling Based on Spatial-Temporal Uncertainty for
Underwater Sensor Networks. J. Sens. 2021,2021, 5558078. [CrossRef]
9.
Yang, Q.; Chen, Y.; Dong, W.; Li, T.; Zhu, R.; Huang, X. Cluster-Based Spatial–Temporal MAC Scheduling Protocol for Underwater
Sensor Networks. IEEE Sens. J. 2023,23, 17690–17702. [CrossRef]
10.
Yan, X.; Huang, C.; Gan, J.; Wu, X. Game theory-based energy-efficient clustering algorithm for wireless sensor networks. Sensors
2022,22, 478. [CrossRef] [PubMed]
11.
Cai, L.; Huang, R.; Li, Z.; Luo, L.; Xiong, Z.; Chen, Y. A clustering election game-based and two-level management protocol for
wireless sensor networks. IEEE Internet Things J. 2023,11, 12058–12070. [CrossRef]
12.
Wang, S.; Zhao, D. A power control based handshake-competition MAC protocol for underwater acoustic networks. In Pro-
ceedings of the 2020 IEEE International Conference on Artificial Intelligence and Information Systems (ICAIIS), Dalian, China,
20–22 March 2020; IEEE: Piscataway, NJ, USA, 2020; pp. 665–669.
13.
Baccelli, F.; Blaszczyszyn, B.; Muhlethaler, P. An Aloha protocol for multihop mobile wireless networks. IEEE Trans. Inf. Theory
2006,52, 421–436. [CrossRef]
14.
Yavascan, O.T.; Uysal, E. Analysis of Slotted ALOHA with an Age Threshold; IEEE: Piscataway, NJ, USA, 2021; Volume 39,
pp. 1456–1470.
15.
Wei, D.; Qiuling, Y.; Yanxia, C.; Shijie, S.; Xiangdang, H. RHNE-MAC: Random Handshake MAC Protocol Based on Nash Equilibrium
for Underwater Wireless Sensor Networks; IEEE: Piscataway, NJ, USA, 2021; Volume 21, pp. 21090–21098.
16.
Liu, M.; Zhuo, X.; Wei, Y.; Wu, Y.; Qu, F. Packet-level slot scheduling MAC protocol in underwater acoustic sensor networks.
IEEE Internet Things J. 2021,8, 8990–9004. [CrossRef]
17.
Chen, W.; Tao, J.; Cao, H.; Jiang, M.; Qiao, Y. Vector approximate message passing based channel estimation for OFDM underwater
acoustic communications. In OCEANS 2021: San Diego—Porto; IEEE: Piscataway, NJ, USA, 2021; pp. 1–5.
18.
Su, Y.; Liu, X.; Han, G.; Fu, X. A Traffic Load-Aware OFDMA-Based MAC Protocol for Distributed Underwater Acoustic Sensor Networks;
IEEE: Piscataway, NJ, USA, 2021; Volume 70, pp. 10501–10513.
19.
Guo, J.; Song, S.; Liu, J.; Wan, L.; Zhao, Y.; Han, G. An Efficient Medium Access Control Scheme Based on MC-CDMA for Mobile
Underwater Acoustic Networks; IEEE: Piscataway, NJ, USA, 2022; Volume 36, pp. 167–173.
20.
Zhang, J.; Cai, M.; Han, G.; Qian, Y.; Shu, L. Cellular Clustering-Based Interference-Aware Data Transmission Protocol for Underwater
Acoustic Sensor Networks; IEEE: Piscataway, NJ, USA, 2020; Volume 69, pp. 3217–3230.
21.
Liu, Z.; Jin, X.; Yang, Y.; Ma, K.; Guan, X. Energy-Efficient Guiding-Network-Based Routing for Underwater Wireless Sensor Networks;
IEEE: Piscataway, NJ, USA, 2022; Volume 9, pp. 21702–21711.
22.
Heinzelman, W.R.; Chandrakasan, A.; Balakrishnan, H. Energy-efficient communication protocol for wireless microsensor
networks. In Proceedings of the 33rd Annual Hawaii International Conference on System Sciences, Maui, Hawaii, 4–7 January
2000; IEEE: Piscataway, NJ, USA, 2000; p. 10.
23.
Xie, W.; Shen, X.; Wang, C.; Sun, L.; Yan, Y.; Wang, H. Adaptive Energy-Efficient Clustering Mechanism for Underwater Wireless
Sensor Networks Based on Multi-Dimensional Game Theory. IEEE Sens. J. 2024,24, 26616–26629. [CrossRef]
J. Mar. Sci. Eng. 2025,13, 76 18 of 18
24.
Zhang, T.; Zhu, Q. EVC-TDMA: An enhanced TDMA based cooperative MAC protocol for vehicular networks. J. Commun. Netw.
2020,22, 316–325. [CrossRef]
25.
Qarabaqi, P.; Stojanovic, M. Statistical characterization and computationally efficient modeling of a class of underwater acoustic
communication channels. IEEE J. Ocean. Eng. 2013,38, 701–717. [CrossRef]
26.
Xue, L.; Cao, C. Research on Distributed Uwsn Power Control Algorithm Based on Cooperative Game. 2023. Available online:
https://ssrn.com/abstract=4654117 (accessed on 1 January 2025) .
27.
Zhu, R.; Liu, L.; Li, P.; Chen, N.; Feng, L.; Yang, Q. DC-MAC: A Delay-Aware and Collision-Free MAC Protocol Based on Game
Theory for Underwater Wireless Sensor Networks. IEEE Sensors J. 2024,24, 6930–6941. [CrossRef]
28.
Syakur, M.A.; Khotimah, B.K.; Rochman, E.; Satoto, B.D. Integration k-means clustering method and elbow method for
identification of the best customer profile cluster. In Proceedings of the IOP Conference Series: Materials Science and Engineering;
IOP Publishing: Bristol, UK, 2018; Volume 336, p. 012017.
29.
Guo, J.; Song, S.; Liu, J.; Chen, H.; Cui, J.H.; Han, G. A Hybrid NOMA-Based MAC Protocol for Underwater Acoustic Networks.
IEEE/ACM Trans. Netw. 2023,32, 1187–1200. [CrossRef]
Disclaimer/Publisher’s Note: The statements, opinions and data contained in all publications are solely those of the individual
author(s) and contributor(s) and not of MDPI and/or the editor(s). MDPI and/or the editor(s) disclaim responsibility for any injury to
people or property resulting from any ideas, methods, instructions or products referred to in the content.
