Yang and Liu EURASIP Journal on Wireless Communications and Networking

(2019) 2019:120
https://doi.org/10.1186/s13638-019-1442-0

RESEARCH Open Access

An optimization routing protocol for

FANETs
Hua Yang* and Zhiyong Liu

Abstract
With the wide-ranging application of mobile ad hoc networks, flying ad hoc networks (FANETs) have received more
and more attention from the industry. Routing technology is a key technology of ad hoc networks. The high-speed
mobility of nodes poses a greater challenge to FANET routing technology. Based on the Dynamic Source Routing
(DSR) protocol, the continuous Hopfield neural network is used to optimize the route to be adapted to the high-
speed movement of the FANET node. In a simulation using NS3, the result shows that the optimized DSR protocol
has greatly improved key indicators such as end-to-end average delay, throughput, and packet delivery ratio.
Keywords: FANET, MANET, Routing protocol, Neural networks

1 Introduction may easily lead to route breaks, which will trigger rout-
In the past, mobile ad hoc networks (MANETs) had ing updates frequently, generate a large amount of con-
been applied more and more, and have expanded into trol overhead in the network, cause route convergence
the air to form flying ad hoc networks (FANETs) [1]. As difficulties, increase data forwarding delay, and increase
an improved mobile ad hoc network, FANETs use the packet loss rate, and even the routing protocol is invalid.
aircraft as a node for transmitting, receiving, or forward- In order to realize reliable and effective networking of
ing over the air wireless communication. It can set up a high dynamic nodes, routing protocols are one of the
network at any time and any place without any fixed fa- key issues that need to be solved in FANETs. The same
cilities and realize a multi-aircraft system. The network as the MANET routing, the FANET routing protocol
layer is effective communication. The FANETs have the should also have the ability to self-organize and
commonality of traditional MANETs, such as no center, self-heal. At the same time, the FANET routing protocol
multi-hop, self-organization, and self-healing, but are also needs to adapt to frequent changes in the network
also given their specific design goals and network char- topology brought by the high-speed movement of nodes.
acteristics. The characteristics of the aircraft nodes are Compared with MANETs, nodes in FANETs have higher
fast, maneuverable, and complex in the flight environ- moving speed, and their routing is the most likely to
ment and the high degree of confrontation. It will lead break in the communication process. Although the trad-
to restricted bandwidth of the communication transmis- itional meaning of MANETs’ routing also considers the
sion link, frequent switching changes, and high error moving speed of nodes, it cannot be directly trans-
rate. The network topology has dedicated time-varying planted to FANETs with high moving speed. In the
and significant dynamic, so FANET routing design is meantime, it is necessary to ensure the corresponding
also more difficult and at the same time very improvements or redesigns in conjunction with the ac-
challenging. tual situation of FANETs.
Routing protocols [1–6] have always been the core
technology in various types of wired networks and wire-
less networks, and are also research hotspots. For the 2 Related work
highly dynamic FANETs, the strong mobility of a node 2.1 Overview of FANET routing protocol
At present, the industry does not have a proprietary
routing protocol for FANET, and its main routing proto-
* Correspondence: gl-yh@guat.edu.cn
Guangxi Colleges and Universities Key Laboratory of Robot and Welding, col still uses the routing protocol of MANETs. MANET
Guilin University of Aerospace Technology, Guilin, China routing protocols [7–10] can be divided into different
© The Author(s). 2019 Open Access This article is distributed under the terms of the Creative Commons Attribution 4.0
International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and
reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to
the Creative Commons license, and indicate if changes were made.
Yang and Liu EURASIP Journal on Wireless Communications and Networking (2019) 2019:120 Page 2 of 8

types according to different classification criteria. Ac- node, it is possible to obtain more topology information,
cording to the route discovery strategy, it is divided into and more and more network topology information is
reactive routing, proactive routing, and hybrid routing. stored in the cache to improve the cache hit rate of the
route lookup. So, it can reduce the frequency of routing
2.1.1 Reactive routing discovery process and save network bandwidth. DSR
This type of routing protocol creates a route only when mainly includes route discovery and route maintenance.
the source node needs to communicate, and does not
need to maintain the routing information of all nodes in 2.2.1 Route discovery
real time like a table-driven routing protocol. Therefore, When the communication node needs to transmit
such routes are known as on-demand routing protocols. packet data, the route discovery process checks whether
This type of routing protocol has poor stability in highly the node itself contains the destination and routing in-
dynamic networks and is also less secure, for example, formation for the initialization data. If it has, the route is
Dynamic Source Routing (DSR) [11–13] and ad hoc initiated to establish a connection. Else, the route is initi-
on-demand distance vector (AODV) [11, 14, 15]. ated to establish a flood routing the request to deter-
mine whether the destination node exists. If it exists, the
2.1.2 Proactive routing routing response is sent. The source node receives the
According to the characteristics of MANET, based on the response to establish a connection. If there is no destin-
wired network routing protocol, all nodes maintain a rout- ation node, it determines whether the routing informa-
ing table that contains routing information from this node tion to the destination node is cached. If sure, routing
to other nodes. When the network topology changes, the response is sent, and the source node receives the re-
node maintains the routing table updated by exchanging sponse to establish a connection. Else, it determines
routing information, so the routing table can accurately re- whether the maximum hop count reached restrict, and
flect the network topology information. Because of FANETs, then return to search again. If the limit reached, the
high-speed movement of nodes causes network topology to route fails to establish a connection. When the routing
change very frequently, such routes are not suitable for use connection is interrupted, the node will send an error
in FANETs, for example, Optimized Link State Routing packet to the source and the notice in reverse. At this
Protocol (OLSR) [11, 16, 17] and Destination-Sequenced point, the route discovery process is completed.
Distance Vector (DSDV) [11, 18].
2.2.2 Route maintenance
2.1.3 Hybrid routing After the route is established using the route discovery,
This type of routing algorithm uses both active routing a routing table will exist in the source node, and the data
and on-demand routing. It divides the network into do- transmission will be performed according to the routing
mains and uses active routing and reactive routing pol- table. During the process of transmitting the data, the
icies within and between domains, for example, Zone established route has to be maintained to be continu-
Routing Protocol (ZRP) [11, 19]. ously available. By maintaining the routing table, the
In the above main routing protocols, the DSR has low source node can monitor the network topology and dis-
overhead and rapid response to network topology cover changes in the network topology in time to deter-
changes and can provide fast response services to ensure mine whether the route to the destination node is
that data packets can successfully reach the destination continuously available. When a node in the network detects
node. Therefore, it is more suitable for FANETs with fast a route break, the node will send a route error packet, and
node movement and strong topological changes. Based the node who receives the error packet then caches it and
on the DSR, it is optimized based on CHNN, and the ex- deletes the route information. If the source node in a com-
pected route is more suitable for FANETs, thereby im- munication receives a route error packet, it deletes the
proving the performance of FANETs. route and the route discovery process is started.

2.2 DSR protocol 2.3 Hopfield neural networks

DSR [12] is a topology-based reactive routing protocol The Hopfield neural networks [20–24] is a recurrent neural
that uses active caching strategies and extracts topology network, proposed by John Hopfield in 1982. It is a neural
information from source routes. The node can obtain network that combines storage systems and binary systems.
the route to all the downstream nodes from the data- The Hopfield neural network is called the Hopfield neural
gram header and can also derive more topology informa- network model, also known as the Hopfield model is a cyclic
tion by combining multiple pieces of information. In neural network with feedback connection from the output
addition, in the network interface mode of the setup to the input. According to the different activation function
node, by listening to the route used by the neighbor selection, it can be divided into continuous Hopfield neural
Yang and Liu EURASIP Journal on Wireless Communications and Networking (2019) 2019:120 Page 3 of 8

network (CHNN) and discrete Hopfield neural network iterated throughout the network. No longer
(DHNN). changing in the process.
CHNN mainly solves the combinatorial optimization (3) The state of the network changes with time, and
problem; Hopfield and Tank realize that with this basic the output state of each neuron at time t is related
neural network organization form, by selecting the to its own t-1 time.
weights and external inputs that can properly represent (4) Introduce the concept of energy function to judge
the function to be minimized and the desired state, it is the stability of network iteration, that is, the
possible to calculate the specific solution and the convergence of the network, that is, the energy
method of optimizing the problem. Updating neurons function reaches a minimum value.
based on differential equations ensures that energy func- (5) The solution of the network, that is, the state set of
tions and optimization problems can be minimized at each neuron when the network runs to stability.
the same time. The analog nature of the neurons and
the hardware implementation of the update process can The circuit topology structure of CHNN is shown in
be combined to create a fast and powerful solution. The Fig. 1, where Ui represents the input voltage of the elec-
key to solving various combinatorial optimization prob- tronic amplifying component, Vi represents the output
lems with neural networks is to map the problem into a voltage, and operational amplifier i represents the ith
neural network dynamic system and to write the corre- neuron.
sponding energy function expressions and dynamic Its state equation is
equations. They should satisfy the constraints of the
problem. Finally, the neural network is studied. The dy-
dui X n Xn
namic process ensures that the steady-state output of Ci ¼ T ij v j − T ij u j −T i0 ui þ I i ð1Þ
the network corresponds to the minimum value of the dt j¼1 j¼1
energy function and the solution to the combinatorial
dui X X
optimization problem. n n
The features of the Hopfield neural network are as Ci ¼ T ij v j − T ij u j −T i0 ui þ I i ð2Þ
dt
follows: j¼1 j¼1

(1) All neurons are both an input and an output, where the input voltage vi satisfies the nonlinear map-
forming a single-layer fully connected recursive ping rule f:
network.
(2) The synaptic weight of the network is different
from other neural networks obtained through vi ¼ f i ðui Þ ð3Þ
supervised or unsupervised iterative learning, but is
calculated according to certain rules when building CHNN uses the energy function to measure the stabil-
the network, and the weight of the network is ity of the network, which is defined as:

Fig. 1 CHNN circuit topology. Hopfield neural network equivalent to amplifying electronic circuits. Input information of each electronic
component (neuron), including constant external current input, and feedback connection with other electronic components
Yang and Liu EURASIP Journal on Wireless Communications and Networking (2019) 2019:120 Page 4 of 8

1X n X n Xn model. The wireless propagation model reveals the rela-

E¼ T ij vxi vxj þ vi I i tionship between radio wave transmit power, received
2 i¼1 j¼1 j¼1
Xn Z power, and propagation distance. When the two wireless
1 ui −1 nodes are in a very close position, the Friis model is
þ f ðvi Þd vi ð4Þ
R
i¼1 i 0 closer to the actual situation. The two-ray model is more
applicable when the two wireless nodes are farther away.
In general, the nodes of wireless communication are far
3 Optimization method apart, so the two-ray model is used.
3.1 Routing metric
Routing performance directly determines the perform- P i G i G r ð hi hr Þ 2
Pr ¼ ð6Þ
ance of FANETs, and routing metrics directly determine D4
the performance of the route. In ad hoc networks, due
In the formula, Pi represents the transmission power,
to frequent changes in the topology, the routing protocol
Pr represents the received power, Gi represents the
usually considers that there are routes for communica-
transmit antenna gain, Gr represents the receive antenna
tion, and does not consider the quality of the route.
gain, hi represents the transmit antenna height, and hr
Therefore, the minimum hop becomes the main route
represents the receive antenna height. It has been set
metric. However, the hop count does not directly reflect
that Pi is a constant, and assuming that the node uses a
the quality of the route. The result is that the hop count
directional antenna, GiG(rhih)2 is a constant expressed in
is the smallest, but the performance of the route is the
k1. Thus, D can be obtained by the transformation of
worst.
Eq. (6):
In order to make the design of routing metric reflect
the quality of routing, in the design of routing metric, sffiffiffiffiffiffiffiffiffi
k 1 Pi
we cannot rely solely on hop, link capacity, link stability, D¼4 ð7Þ
node energy, node performance, interference, node Pr
power, and other link and node factors; all need to be in-
Let the lowest received power of the node be P1, and
cluded. If the communication route only contains the
the corresponding distance is the node effective commu-
hop count and does not consider other factors such as
nication range D0, then k1Pi = P1D04 can be obtained
link stability, it is easy to cause the route to be broken
from Eq. 3, and the calculated distance D can be ob-
due to frequent changes in the topology, causing the
tained by substituting into Eq. (7):
network to generate a high retransmission rate and caus-
rffiffiffiffiffi
ing high delay of the network, resulting in some nodes. P1
The energy is consumed unnecessarily, which in turn af- D¼4 D0 ð8Þ
Pr
fects the performance and longevity of the network. A
reliable route is often determined by the stability of the Finally, substituting Eq. (8) into Eq. (5) yields the cal-
link in ad hoc networks, so the design of the routing culated link stability Sij.
metric need to fully consider the stability of the link. 8
< 0; D≥ D0
rffiffiffiffiffi
Sij ¼ P1 ð9Þ
3.2 Definition of link stability : 1−4 D0 ; 0 ≤ D < D0
Let i and j be any two adjacent nodes in the network, Pr
and D is its communication distance. Assuming that the
nodes are of the same type, the transmit power is the 3.3 Optimization method
same, which also determines the effective communica- Whether in MANETs or FANETs, routing finds an opti-
tion range (set to D0). Let the stability of the link be- mized source-to-destination communication path in a
tween nodes be represented by Sij, which is related to set of wireless nodes {N1, N2,...Nn}. In order to map rout-
the effective range of communication and the distance ing problems to the dynamic evolution of a neural net-
between nodes, as defined below. work, we must first find a suitable representation. The
( FANET network node is mapped to the CHNN, and the
0; D ≥D0
S ij ¼ D ð5Þ optimal route is searched through the dynamic evolution
1− ; 0 ≤ D < D0 of the neural network state. When the network state
D0
tends to be stable, the obtained route is the optimal
When D is greater than or equal to D0, the nodes can- route or the suboptimal route. The search time of the
not communicate directly, and Sij is 0. Conversely, nodes neural network is the time when the state of the neural
can communicate directly, with Sij between 0 and 1. Link network reaches a stable state, and the calculation
stability prediction is based on the wireless propagation process is also the dynamic process of the neural
Yang and Liu EURASIP Journal on Wireless Communications and Networking (2019) 2019:120 Page 5 of 8

network. So the key to solving the problem is to con- Using RSD to represent the integrated distance from the
struct a suitable neural network energy function. node S to node D, the Fig. 2 total length of the route is
In order to optimize the CHNN routing of the FANET,
the appropriate expression of the route must be found RSD ¼ RS4 þ R42 þ R23 þ R31 þ R15 þ R5D ð11Þ
first. A route is an ordered arrangement, which has a
source node, some of the intermediate nodes, and the des- For FANETs, routing cannot pass through each inter-
tination node; the intermediate nodes can be represented mediate node. The size of the matrix must be set to an
in the form of n-dimensional vectors. In FANET, there upper limit. When there are too many intermediate nodes,
have n intermediate nodes, and the required vector di- it is considered unreachable. Referring to the RIP routing
mension is n − 2, such as there is a matrix representation protocol in the wired network, set to n = 16, the energy is
of 5 intermediate nodes (Fig. 2). In Fig. 2, the only route is the third function of the function is described as:
shown: Source - > N4 - > N2 - > N3 - > N1 - > N5 - > Destin-
!2
ation. For every possible route, every intermediate node T XX
can only reach once, so each row and column in this hðV Þ ¼ V xi −n ð12Þ
2 x i
square can only have one element with 1 and the
remaining elements with 0. This is the constraint of the
routing problem. Every route with the above constraints In order to contain the effective solution route length
can be represented by a transposition matrix. So, the first information, the fourth term of the energy function is
two terms of the energy function are set to
PX XX  
f ðV Þ ¼ S mn V mi V nðiþ1Þ þ V nði−1Þ
QX XX 2 m n i
g ðV Þ ¼ V mi V ni m≠n
2 m n i
m≠n ð13Þ
SXXX
þ V mi V mj ð10Þ
2 m j i Therefore the total energy function can be expressed
i≠ j using the following equation:

E ¼ f ðV Þ þ hðV Þ þ g ðV Þ ð14Þ

The parameters Q, S, T, and P are called the weight,

the first three items are the constraints that satisfy the
route replacement matrix, and the last one contains the
optimization objective function item.
In order to simplify the calculation, there is a need to
ignore the integral subitem of the energy function, and
then the neural network connection weight can be
expressed as:
 
ωmi;ni ¼ −Sδ mn 1−δ ij −Qδ ij ð1−δ mn Þ−T ð15Þ
−PS mn δ jðiþ1Þ þ δ jði−1Þ

The random item added is δij ∈ {−1, 1} .

Calculate the increment of the input state using the
CHNN dynamic equation:
!2
X X XX
dU mi ¼ −S V mj −Q V nj −T V mj −16
j≠i m≠n m j
X   U mi
−P smn V nðiþ1Þ þ V nði−1Þ −
n
Rmi C mi
ð16Þ
Fig. 2 Use matrix to represent five nodes FANET routing. Use matrix
to represent five nodes FANET routing; the only route is shown:
Update the output state of the neural network at the
Source - > N4 - > N2 - > N3 - > N1 - > N5 - > Destination
next moment by the sigmoid function:
Yang and Liu EURASIP Journal on Wireless Communications and Networking (2019) 2019:120 Page 6 of 8

   waypoint mobility model. The simulation runs for 900 s.

1 U mi
V mi ¼ f ðU mi Þ ¼ 1 þ th ð17Þ See Table 1 for detailed settings.
2 U0

Penalty parameters S, Q, T, and P need to be fixed

values. Hopfield [20] uses the set of parameters S = Q = 4.3 Results and discussion
P = 500, U0 = 0.02 to balance these terms. In order to make the results more realistic, the simula-
tion is performed multiple times and then the method of
4 Simulation results and discussion averaging is performed. Figures 3, 4, and 5 show the
4.1 Simulation platform comparison of the average end-to-end delay, packet de-
In order to verify the effectiveness of CHNN-optimized livery ratio, and throughput of CHNN-DSR and DSR
routing, we simulated the FANET scenario to simulate when the speed of the node movement increased. The
the DSR and CHNN-DSR. We mainly analyze and com- results show that node mobility has a great impact on
pare the following parameters that determine routing the routing in FANETs. In the case of high-speed mobil-
performance [25, 26]: ity, the route interruption may occur very frequently.
Therefore, the use of optimized routing in the route dis-
(1) Average end-to-end delay: Defined as the average of covery process will effectively improve the stability of
the packet delays for all successful delivery of the the network.
entire network. Its calculation formula is From Fig. 3, we can see that as the node moves faster,
the average end-to-end delay increases. DSR uses source
routing and storage routing, but the topology changes
drastically cause source routing breaks and storage
1Xn−1 routes to failure. The risk is also increased, and CHNN-
Delay ¼ ðp ðT receive Þ−pi ðT send ÞÞ ð18Þ DSR uses optimized source routing, which has higher
n i¼0 i
stability itself, so the probability of re-initiating route
where pi represents the ith packet and n indicates the lookup is lower, and the communication quality is more
total number of successfully delivered packets. secure.
From Fig. 4, we can see that as the node moves faster,
(2) Packet delivery ratio: Defined as the number of the packet delivery ratio also decreases. The reason is
packets accepted by the destination node and the that the route through CHNN-DSR is optimized and its
number of packets transmitted by the constant bit stability is strong, so the packet delivery ratio can always
rate (CBR) source. Its calculation formula is be maintained at a higher ratio. It is worth noting that
although the speed increases and packet delivery ratio
decreases, the route delivery ratio of both routes is
above 90%, indicating that DSR is still a better choice for
P
Delivery Data FANETs.
Delivery Ratio ¼ P ð19Þ
TransmiteData
Table 1 Network parameter setting
Parameters Value
Number of nodes 30
(3) Throughput: Defined as the average of the amount
of data (kb/s) successfully delivered by each pair of Simulation range 1000 m × 1000 m
communicating parties in the network per unit MAC layer 802.11 DCF
time. Wireless channel bandwidth 11 Mbps
Transfer model WaveLan model
4.2 Parameter settings
Motion model Random waypoint mobility
The simulation software uses the NS3 simulation system
Node pause time Random
(https://www.nsnam.org/) [27]. The CHNN code is de-
veloped in C++ and can be directly applied to the NS3 Data transmision rate 1 Mbps
system. When 30 nodes in the network start to simulate, Data channel packet size 1024 bytes
they are evenly distributed in the area of 1000 m × 1000 Control channel packet size 128 bytes
m. The internode communication uses a CBR stream, Transmission range 0~80
and the data packet length is 1024 bytes. The MAC layer Max speeds 10~50 m/s
protocol uses IEEE 802.11 DCF. The transfer model uses
Simulation times 900 s
Lucent’s WaveLan. The motion model adopts a random
Yang and Liu EURASIP Journal on Wireless Communications and Networking (2019) 2019:120 Page 7 of 8

Fig. 3 Average end-to-end delay simulation results for DSR and Fig. 5 Throughput simulation results for DSR and CHNN-DSR. The
CHNN-DSR. The simulation results of the average end-to-end delay simulation results of the throughput for CHNN-DSR and DSR under
for CHNN-DSR and DSR under the change of node the change of node movement speed
movement speed

the route determines the stability of the network.

From Fig. 5, we can see that as the node moves faster Based on CHNN, the DSR is optimized and named as
and the throughput performance is getting worse and CHNN-DSR, which obtains a more stable and stable
worse. The optimized CHNN-DSR performance is obvi- route, which improves the stability of FANET and the
ously better than the DSR. This is because the DSR does communication efficiency of the network. The simula-
not remove any mechanism of outdated invalid routing, tion results show that in the FANETs using CHNN-
and it is not determined. The freshness of each route is DSR, the key indicators such as packet delivery ratio,
used to select the correct routing mechanism after the end-to-end average delay, and throughput are better
communication route fails, and the CHNN-DSR has a than those directly using DSR. In the subsequent re-
lower probability of routing failure than the DSR be- search, the application of artificial intelligence algo-
cause of its strong stability. rithms such as neural network algorithms to the
design of FANET routing will be worthy of further
5 Conclusions research.
It is feasible to apply artificial intelligence algorithms
such as neural network algorithms to FANET routing Abbreviations
AODV: Ad hoc on-demand distance vector; CBR: Constant bit rate;
optimization. In highly dynamic FANETs, the stability of CHNN: Continuous Hopfield neural network; DHNN: Discrete Hopfield neural
network; DSDV: Destination-Sequenced Distance Vector; DSR: Dynamic
Source Routing; FANET: Flying ad hoc network; MANET: Mobile ad hoc
network; OLSR: Optimized Link State Routing; ZRP: Zone Routing Protocol

Acknowledgements
This work is supported by Guangxi Colleges and Universities Key Laboratory
of Robot & Welding.

Funding
The Science and Technology Research Project of Department of Education
of Guangxi (Grant:2019KY0815)

Availability of data and materials

Not applicable.

Authors’ contributions
This manuscript is mainly completed by YH and LZY. YH designed an
improved algorithm and simulated it. YH and LZY analyzed the simulation
results. YH wrote this manuscript, LZY checked and reviewed the manuscript.
Both authors read and approved the final manuscript.
Fig. 4 Packet delivery ratio simulation results for DSR and CHNN-
DSR. The simulation results of the packet delivery ratio for CHNN-
DSR and DSR under the change of node movement speed Competing interests
The authors declare that they have no competing interests.
Yang and Liu EURASIP Journal on Wireless Communications and Networking (2019) 2019:120 Page 8 of 8

Publisher’s Note 24. S. Zhang, Y. Yu, Q. Wang, Stability analysis of fractional-order Hopfield
Springer Nature remains neutral with regard to jurisdictional claims in neural networks with discontinuous activation functions. Neurocomputing
published maps and institutional affiliations. 171, 1075–1084 (2016)
25. S. J. Lee, W. Su, M. Gerla, (2002). On-demand multicast routing protocol in
Received: 13 February 2019 Accepted: 18 April 2019 multihop wireless mobile networks. Mobile networks and applications, 7(6),
441-453.
26. G. Singal, V. Laxmi, M.S. Gaur, V. Rao, Moralism: mobility prediction with link
stability based multicast routing protocol in MANETs. Wirel. Netw 23(3),
References 663–679 (2017)
1. I. Bekmezci, O.K. Sahingoz, Ş. Temel, Flying ad-hoc networks (FANETs): a 27. Y. Mai, Y. Bai, N. Wang, Performance comparison and evaluation of the
survey. Ad Hoc Netw.. 11(3), 1254–1270 (2013) routing protocols for MANETs using NS3. J. Electr. Eng. 5, 187–195 (2017)
2. V. Sharma, R. Kumar, G-FANET: an ambient network formation between
ground and flying ad hoc networks. Telecommun. Syst. 65(1), 31–54 (2017)
3. A.V. Leonov, Application of bee colony algorithm for FANET routing. In:
2016 17th International Conference of Young Specialists on Micro/
Nanotechnologies and Electron Devices (EDM), Erlagol, Russia, pp. 124–132
(2016). https://doi.org/10.1109/edm.2016.7538709.
4. M. A. Khan, A. Safi, I. M. Qureshi, I. U. Khan, 2017. Flying ad-hoc networks
(FANETs): A review of communication architectures, and routing protocols.
In 2017 First International Conference on Latest trends in Electrical
Engineering and Computing Technologies (INTELLECT). 1-9. https://doi.org/
10.1109/INTELLECT.2017.8277614.
5. W. Zafar, B.M. Khan, A reliable, delay bounded and less complex communication
protocol for multicluster FANETs. Digit Commun Netw 3(1), 30–38 (2017)
6. A. Bujari, C.E. Palazzi, D. Ronzani, in Proceedings of the 3rd Workshop on
Micro Aerial Vehicle Networks, Systems, and Applications. FANET application
scenarios and mobility models (ACM, 2017), pp. 43–46. https://doi.org/10.
1145/3086439.3086440.
7. J. Macker, Mobile Ad Hoc Networking (MANET): Routing Protocol Performance
Issues and Evaluation Considerations (1999)
8. F. Bai, N. Sadagopan, A. Helmy, Important: a Framework to Systematically
Analyze the Impact of Mobility on Performance of Routing Protocols for
Adhoc Networks. In: Proceedings of INFOCOM 2003, San Francisco, CA
(2003)
9. M. Abolhasan, T. Wysocki, E. Dutkiewicz, A review of routing protocols for
mobile ad hoc networks. Ad Hoc Netw. 2(1), 1–22 (2004)
10. F. Bai, N. Sadagopan, A. Helmy, The IMPORTANT framework for analyzing
the Impact of Mobility on Performance Of RouTing protocols for Adhoc
NeTworks. Ad Hoc Netw. 1(4), 383–403 (2003)
11. A. Singh, G. Singh, M. Singh, in Intelligent Communication, Control and
Devices. Comparative study of OLSR, DSDV, AODV, DSR and ZRP routing
protocols under blackhole attack in mobile ad hoc network (Springer,
Singapore, 2018), pp. 443–453
12. D.B. Johnson, D.A. Maltz, in Mobile computing. Dynamic source routing in ad
hoc wireless networks (Springer, Boston, 1996), pp. 153–181
13. H. Yang, Z. Li, Z. Liu(2019). A method of routing optimization using CHNN
in MANET. Journal of Ambient Intelligence and Humanized
Computing, 10(5), 1759-1768.
14. C. Perkins, E. Royer, Ad hoc on-demand distance vector routing, in: IEEE
WMCSA'99(1999).
15. H. Yang, Z. Li, Z. Liu, Neural networks for MANET AODV: an optimization
approach. Clust. Comput. 20(4), 3369–3377 (2017)
16. T. Clausen, P. Jacquet, Optimized link state routing protocol (OLSR) (No. RFC
3626) (2003)
17. Y. Wu, L. Xu, X. Lin, J. Fang, in International Conference on Internet of Vehicles.
A new routing protocol based on OLSR designed for UANET maritime
search and rescue (Springer, Cham, 2017), pp. 79–91
18. C. Perkins, P. Bhagwat, Highly dynamic destination-sequenced distance-
vector routing (DSDV) for mobile computers, in: ACM SIG-COMM
Symposium on Communication, Architectures and Applications (1994).
19. Z.J. Haas, M.R. Pearlman, P. Samar, The Zone Routing Protocol (ZRP) for Ad
Hoc Networks (2002)
20. J.J. Hopfield, Neural networks and physical systems with emergent collective
computational abilities. Proc. Natl. Acad. Sci. 79(8), 2554–2558 (1982)
21. J. Zhang, X. Jin, Global stability analysis in delayed Hopfield neural network
models. Neural Netw. 13(7), 745–753 (2000)
22. G. Rajchakit, R. Saravanakumar, C.K. Ahn, H.R. Karimi, Improved exponential
convergence result for generalized neural networks including interval time-
varying delayed signals. Neural Netw. 86, 10–17 (2017)
23. C. Zhong, C. Luo, Z. Chu, W. Gan, (2017) A continuous hopfield neural
network based on dynamic step for the traveling salesman problem. Neural
Networks (IJCNN), 2017 International Joint Conference (pp 3318–3323). IEEE