Wireless Personal Communications (2022) 126:3693–3714

https://doi.org/10.1007/s11277-022-09886-6

Token Based k‑Mutual Exclusion for Multi‑UAV FANET

Ashish Singh Parihar1,2 · Swarnendu Kumar Chakraborty1

Accepted: 29 May 2022 / Published online: 18 June 2022

© The Author(s), under exclusive licence to Springer Science+Business Media, LLC, part of Springer Nature 2022

Abstract
Mutual exclusion is a well-studied topic in distributed systems. A distributed system with
the duplicate copy of resources can increase throughput by allowing many processes to
call their critical sections simultaneously through parallel execution. Because of the ongo-
ing development of wireless technologies and dynamic networks such as mobile ad hoc
networks, they are becoming increasingly complicated and varied in today’s world. Fly-
ing ad hoc network (FANET), as a special variant of mobile ad hoc network, has nodes in
which the cooperation and collaboration are highly dynamic and unpredictable in nature.
In FANET, resources are mounted on an unmanned aerial vehicle (UAV) which can oper-
ate remotely with a flying capacity in the air without any human personnel. Various pro-
tocols as distributed mutual exclusion algorithms have been proposed as a solution to the
mutual exclusion problem in a distributed system. To the best of our knowledge, such algo-
rithms in FANETs haven’t been exposed much and moreover through this paper, we pre-
sent a novel token-based k mutual exclusion algorithm as mutual exclusion algorithm for
flying network-multi UAV with its correctness proof and fault handling capabilities. To
improve overall throughput as compared to single-UAV FANET, our approach works in a
multi resource occupied flying environment and allows k duplicated resources mounted on
available UAVs that facilitate processes at most k to invoke their critical sections simulta-
neously. This solution is also fault-tolerant to UAV failure in the system, whereas in sin-
gle-UAV FANET, this considers a failure of the entire system. We have also presented sim-
ulation results for our algorithm with range of 5–20 resource-equipped UAVs architecture
support that indicate more efficient results for performance throughput metric.

Keywords Distributed systems · Flying ad hoc network · Mutual exclusion · Resource

allocation · Unmanned aerial vehicle

* Ashish Singh Parihar

[email protected]; [email protected]; [email protected]
Swarnendu Kumar Chakraborty
[email protected]
1
Department of Computer Science and Engineering, National Institute of Technology (NIT), Jote,
Arunachal Pradesh, India
2
Department of Computer Science, KIET Group of Institutions, Delhi-NCR, Ghaziabad,
Uttar Pradesh, India

13
Vol.:(0123456789)
3694 A. S. Parihar, S. K. Chakraborty

1 Introduction

Mutual exclusion is an important issue to consider in a distributed system while main-

taining node synchronization in order to maintain a consistent and consensual system
[1]. The existence of nodes must be abstracted as nodes can have local or partial infor-
mation about the entire network. Because there was no shared memory or common clock
in this network, communication between nodes could only be established by message
passing. As a result, during the information flow among the available nodes, the num-
ber of messages exchanged is required, and the performance of any distributed mutual
exclusion (DME) algorithm is highly dependent on this metric, claiming a minimum
number of message exchanges to obtain a piece of information. These DME algorithms
are divided into two categories: token-based [2–4] and non-token-based (also known as
permission-based) [5]. Message complexity, synchronization latency, throughput, and
response time are key efficiency evaluation metrics for such solutions.
As an infrastructureless classification of a wireless network, ad hoc networking is the
most suitable and acceptable network topology in the current era because of its dynamic
and mobile nature. The establishment of communication between the nodes in the ad
hoc network is complex in nature because of the unpredictability of the node movement.
Various variants of ad hoc networks are available as mobile ad hoc network (MANET),
vehicle ad hoc network (VANET) [6] and FANET [7]. FANET is also considered as
a special variant of MANET that has a larger communication range and higher mes-
sage transmission speed as compared to MANET and VANET. In FANET, resources
are mounted on the UAV or remotely piloted aircraft (RPA) or drones that can operate
remotely with flying capacity in the air without any human personnel and nodes within
the communication range of the UAV can communicate with each other as well as with
UAV [8]. Weather monitoring [9], ground target detection [10, 11], remote sensing [12],
remote sensing in agriculture [13], attacks in hostile areas [14], disaster monitoring
[15], wind estimation [16], relay for ad hoc networks [17–20], search and rescue opera-
tions [21, 22], wild/forest fire monitoring [23], blockchain [24], real-time surveillance
[25], goods delivery and transportation [26], and so on are just some of the applications
for UAVs.
The concept of DME in an ad hoc network was firstly introduced by Walter et al.[27]
and later on, various DME algorithms were proposed for MANET [28–34] and VANET
[35–37]. To the best of our knowledge, such algorithms in FANETs haven’t been exposed
much which gives us a motivation to fill this gap through our proposed study. Hence,
we present a novel token-based k mutual exclusion algorithm for multi-UAV FANET as
Mutual Exclusion Algorithm for Flying Network-Multi UAV (MEAFN-MUAV) with its
correctness proof and fault handling capabilities. To improve overall throughput as com-
pared to single-UAV support architecture, MEAFN-MUAV works in a multi-UAV FANET
environment and allows k duplicated resources mounted on available UAVs that facilitate
processes at most k to invoke their critical section (CS) simultaneously. Our proposed art-
of-study is also fault-tolerant to UAV failure in the system whereas, in single-UAV FANET,
this considers a failure of the entire system. We also have presented efficient simulation
results for our algorithm. Overall, the prime contributions of this suggested work are:

• Study of single vs. multi-UAV FANET architectures.

• Proposed a novel distributed mutual exclusion solution for the muti-UAV FANET
system.

13
Token Based k‑Mutual Exclusion for Multi‑UAV FANET 3695

• Simulation setup and execution for the suggested art-of-study and evaluation of various
performance measures such as message complexity, throughput, synchronization delay
and response time.

Furthermore, this article is organized as follows: The FANET environment, including

its characteristics and applications, is discussed in Sect. 2. The related work is presented in
Sect. 3. Section 4 explains the proposed algorithm’s system concept, as well as its descrip-
tion and fault-tolerant capabilities. The proposed algorithm’s correctness has been demon-
strated in Sect. 5. Section 6 evaluates the proposed algorithm’s simulation setup and analy-
sis using several performance measures. Finally, in Sect. 7, we have come to a conclusion
with recommendations for future research.

2 Flying Ad Hoc Network (FANET)

In this section, FANET [7] has been discussed. MANET is a special variant of ad hoc net-
work and FANET is classified as a sub-variant of MANET in which resources are on UAV.
UAV has the capacity to operate remotely and fly in the air autonomously without any
human personnel. FANET based on multi-UAV systems, has an advantage over the sin-
gle-UAV systems in terms of cost, scalability, survivability, and speed-up. An instance of
single-UAV FANET and Multi-UAV FANET can be seen in Fig. 1, in which circular parts
indicate the communication range of UAV. Although due to the highly dynamic behavior
of nodes in the FANET, the communication between the nodes is complex for the estab-
lishment of collaboration and cooperation between them in both single-UAV and multi-
UAV FANETs. FANET is considered as in-infrastructure if all the UAVs are connected to
any ground base and can be applicable in single-UAV FANET. Such FANET has a restric-
tion like limited range, bounded speed and dependency on ground base over the UAV-UAV
communication-based FANET. A detailed characteristic comparison between various vari-
ants of ad hoc networks is presented in Table 1.
Nowadays due to the versatile nature of UAVs, they have a huge application in the field
of defence, monitoring, transportation, spying and civilian. The communication coverage

Fig. 1  Sample snapshot of single vs. multi-UAV FANET Architecture

13
3696 A. S. Parihar, S. K. Chakraborty

Table 1  Characteristics Network characteristics MANET VANET FANET

comparison between various
variants of ad hoc network
Mobility of nodes Low Moderate High
Density of nodes Low Moderate Low
Topological change Low Moderate High
Power consumption Moderate Moderate Low
Computational power Low Moderate High
Scalability Moderate Low High
Payload capacity Moderate Moderate High
Transmission speed Moderate Moderate High
Propagation delay Low Low Moderate
Communication complexity Moderate Moderate High

of FANET based on multi-UAV has a higher range as compared to single-UAV due to

UAV-UAV communication rather than UAV-ground base. Such multi-UAV FANET is
much more effective on defence-based applications where the environment and condi-
tions are highly dynamic where in case of one UAV failure, others can handle the situ-
ation accordingly. Various monitoring-based applications like weather prediction, tem-
perature check, animal health condition, and human density in an area are also effectively
handled by the multi-UAV FANET, where data can be received on the ground base trave-
ling through UAV-UAV communication. In the current era, e-commerce companies are
using this multi-UAV FANET for their faster delivery of goods to their customers. Overall,
FANET based on multi-UAV nodes has a huge potential in various domains due to its high
mobility nature and fault tolerance capability which is most suited to today’s rapid environ-
mental change.

3 Literature Survey

Mutual exclusion in a distributed system is a highly researched area. Walter et al. [27] pre-
sented the first token-based DME algorithm as a reverse link (RL) in which the location
of system nodes is mobile. Communication among nodes is possible directly or through
some intermediate nodes where transmission range is the deciding factor by considering a
directed acyclic graph (DAG) as an underlying network topology. They used bidirectional
links for signal passing and eliminated the concept of node failure and network partition in
their discussion. They claimed better results as compared to Raymond [38]. Another token-
asking algorithm as ad hoc distributed mutual exclusion (AHDME) algorithm, is presented
by Zheng et al. [39], in which they incorporated concepts to finalize the CS request priori-
ties of nodes. The concepts of distributed and centralized DME schemes are integrated by
Bharti et al. [41] for mobile networks. The subdivision of these networks is formed into
the number of clusters in such a way that the overall system can be realized as inter-cluster
(Ie_C) and each sub-cluster is observed as intra-cluster (Ia_C). Ia_C of the network used
the centralized token passing scheme, while the distributed scheme took place in Ie_C.
Various fault-tolerant capabilities like- loss of token and failure of nodes are discussed
through their methodology. To serve any CS request of node, their approach used the token
circulation scheme and maintained a unique token at the Ie_C level. Each Ia_C had its local
token. Therefore solution can also be treated as a dual token-based approach. Cluster head

13
Token Based k‑Mutual Exclusion for Multi‑UAV FANET 3697

(CH) had been elected at every level of Ie_C and Ia_C through which transmission of the
token occurred. By considering K number of cluster heads and M number of nodes, their
proposed work has a message complexity as O(K) and O(1/M) in low and high load situa-
tions in the network.
As a mobile resource mutual exclusion (MRME) approach, Khanna et al. [42] presented
the first token-based system to enable resource sharing in FANET. They demonstrated the
success of their method by displaying the message complexity as O(N) and the synchro-
nization delay as T, as well as its fault tolerance. Each node has a unique identifier and
can be in one of four states: reminder (REM), critical section (CS), requesting for token
(REQ), and HOLDING the token (HI), according to their algorithm methodology. At first,
the node with the lowest identifier is the token-holder node, broadcasting token informa-
tion to its neighbors (nodes that are in the range of UAV). Khanna et al. [43] developed the
local mutual exclusion (LME) idea in FANETs as request collector local mutual exclusion
(RCLME) with fault-tolerant token loss capability. In LME, more than one neighboring
node is not permitted to enter its CS. By presenting the LME notion in FANETs, they
claimed their work was novel. RCLME is primarily a solution based on leader election, in
which a leader is chosen using fuzzy logic that considers a variety of system criteria such
as link quality and speed among nodes, shared resource distance, and so on (Table 2).

4 System Model and Proposed Algorithm

Multi-UAV FANET based distributed system containing N number of mobile nodes with
unique identifiers has been shown in Fig. 2 in which the communication range of UAV is
represented by a circular area as neighborhood [45] of UAV. Total k replicas of a resource
are available in the system and mounted on each UAV. Initial assumptions are as follows:

Assumption 1 The number of UAVs, resource replicas and tokens are k in the case of
k-UAV FANET.

Assumption 2 Each k replica of a resource is mounted on each of k UAV.

Assumption 3 Nodes available within the neighborhood of any UAV can access the
shared resource mounted on the same UAV, but not from the other UAVs.

Assumption 4 In each neighborhood of UAV, a unique token exit and the node that have
the token can’t leave the neighborhood i.e., the token can’t go outside of the UAV neigh-
borhood with token.

Assumption 5 Message passing is the only way of communication between node–node

and node-UAV. Also, the delivery of the message is guaranteed via the pre-assumed trusted
FANET routing protocol.

Assumption 6 The token holder node can invoke its CS and at the same time, more than
one token holder nodes can invoke their CS belonging to a different UAV neighborhood.

Assumption 7 Nodes available in UAV’s neighborhood can communicate with each

other as well as with UAV and can participate in DME.

13
 3698

13
Table 2  Characteristics comparison between various variants of ad hoc network
DME algorithm Network topology Publisher Message Complexity Synchroni- Fault-tolerant capabilities
zation delay
Worst Case Best Case Token Loss Node Loss Network
Link
Loss

[27] MANET Wireless Networks < O(log N) 0 – ✗ ✗ ✓

[29] MANET Journal of Parallel and Distributed Computing – – T ✗ ✗ ✗
[30] MANET Mobile and Wireless Communication Networks – – T ✗ ✗ ✗
[31] MANET Mobile Ad-Hoc and Sensor Networks – – – ✓ ✗ ✗
[33] MANET Pervasive and Mobile Computing N 3 N/2 1 ✗ ✓ ✓
[35] VANET Advances in Parallel Distributed Computing O(1/N) – T ✓ ✗ ✗
[36] ON Pervasive and Mobile Computing ∝ (𝜔) – – ✗ ✓ ✗
[42] FANET Computers and Electrical Engineering O(N) – T ✓ ✗ ✗
[43] FANET Computer Communications < O(N) – T ✓ ✗ ✗
[44] VANET IEEE Access O(N) – – ✗ ✗ ✓

Notation
N: Total number of nodes in model
T: message propagation delay
α(ω): Time unit in average for next CS request generation by a node
ON: Opportunistic networks
A. S. Parihar, S. K. Chakraborty
Token Based k‑Mutual Exclusion for Multi‑UAV FANET 3699

Fig. 2  System model for MEAFN-MUAV for multi-UAV FANET

4.1 Proposed Algorithm

Here in this section, we present our token-based k mutual exclusion algorithm for multi-
UAV FANET as mutual exclusion algorithm for flying network-multi-UAV (MEAFN-
MUAV). This solution works in a multi-UAV FANET environment and allows k dupli-
cated resources mounted on available UAVs that facilitate processes at most k to invoke
their CS simultaneously. Firstly, we provide the data structures and various messages
used by MEAFN-MUAV and then we discuss its corresponding working algorithms
(Sect. 4.1.3: Algorithm 1–7).

4.1.1 Data structures used in MEAFN‑MUAV (At each ­Nodei)

• Token_Holderi Contains the token holder identifier corresponding to its UAV neigh-
borhood.
• Previousely_Token_Holderi Contains the previous token holder identifier corre-
sponding to its UAV neighborhood.
• CS_Request_Queuei List of pending CS requests of nodes in terms of their identi-
fiers.
• CS_Executingi Boolean variable as T or F indicating the node’s CS executing state.

13
3700 A. S. Parihar, S. K. Chakraborty

4.1.2 Various Messages used in MEAFN‑MUAV

• Have_Token(i) Token holder node (say, Nodei) broadcasts this message to its UAV
neighborhood nodes to inform them that it has the token. While receiving this mes-
sage, each node updates its Token_Holder variable.
• Ask_CS(i) Message generated by Nodei for asking a token and sent to token holder node
using its Token_Holder variable.
• Previousely_Have_Token(i) Token holder node (say, Nodei) broadcasts this message to
its UAV neighborhood nodes just before sending the token to the next token asking
node. While receiving this message, each node updates its Previousely_Token_Holder
variable.
• Update_Your_Queue(CS_Request_Queuei) The message is sent from the current token
node to the next token asking node and contains the CS_Request_Queue of the current
token holder node. While receiving this message, the node updates its CS_Request_
Queue variable with the incoming CS_Request_Queue along with the message param-
eter.
• Hello_Message() The formal message is sent between two nodes to ensure whether the
receiving node is alive or in a range of UAV neighborhoods.
• Create_Token(i) Message generated and sent by any arbitrary node in the system ask-
ing for CS to the previous token holder node. While receiving this message, the node
creates a copy of the token from the token’s replica and becomes the new token holder.
• Get_Me_Info(i): Newly joined Nodei to any UAV neighborhood broadcast this message
to the rest of the nodes. Only the token holder node responds to this message along
with its Token_Holder and Previousely_Token_Holder data to the originator node of
this message. Nodei updates its Token_Holder and Previousely_Token_Holder based on
the information received from the token holder node.

4.1.3 MEAFN‑MUAV Algorithm

We start our discussion with a formal description of the proposed algorithm, namely as
MEAFN-MUAV. In the very first phase of its Initialization (Algorithm 1–2), our solution
felicitates initial values to various data structures at each node and broadcasts token-
holder node information to every UAV neighborhood as one unique token-holder node
exists in each UAV neighborhood (Assumption 4). Each node updates its Token_Holder
with the token-holder node’s identifier belonging to the corresponding neighborhood
of UAV in the system. In the second phase as CS_Asking (Algorithm 3), whenever any
node, say Nodek demands its CS to invoke, it sends a Hello_Message() message to the
token-holder node say Nodel, that is unique to its neighborhood to confirm its aliveness.
After receiving an acknowledgment from Nodel, Nodek sends a Ask_CS(k) to Nodel. If
the token is idle at Nodel, Nodel sends the token to Nodek after creating a replica of the
token and becomes the previous token-holder node for the corresponding UAV neigh-
borhood in case to handle token loss, add the incoming request into its CS_Request_
Queuel otherwise. The third phase as CS_Release (Algorithm 4) explains the working
scenario of CS release by token-holder node. When the token-holder node releases its
CS then it fetches the next CS asking request from its CS_Request_Queue and sends the
token towards that node by confirming its existence using Hello_Message() along with
Update_Your_Queue(CS_Request_Queue) message. In case the CS_Request_Queue of

13
Token Based k‑Mutual Exclusion for Multi‑UAV FANET 3701

the token-holder node is empty, then the algorithm remains idle. Any node, after receiv-
ing the token, broadcasts its information to the rest of the nodes belonging to the same
UAV neighborhood (Algorithm 5). Since in FANET, the movements of nodes are highly
unpredictable, algorithm effectively handles such cases where any node leaves or join
any UAV neighborhood (Algorithm 6–7).

13
3702 A. S. Parihar, S. K. Chakraborty

4.2 MEAFN‑MUAV Fault Tolerant Capabilities

Theorem 4.2.1 Our solution is fault-tolerant to node failure in the system.

Proof Let’s take the following considerations in case of a node failure,

13
Token Based k‑Mutual Exclusion for Multi‑UAV FANET 3703

Case 1 (Node doesn’t belong to any UAV neighborhood) Node failure in such case has
no impact on the system because the node doesn’t belong to any UAV neighborhood that
can’t participate in DME (Assumption 7) and also, such node failure can’t be responsible
for the token loss as per Assumption 4.
Case 2 (Node belong to any of the UAV neighborhood)-
Case i (Node is not a token holder node)
Case a (Node has the pending CS request in token holder node queue): In this case,
whenever the node turn comes for getting the token from the token holder node,
the token holder node simply identifies this node as dead using the Hello_Mes-
sage() and ignore the CS request.
Case b (Node doesn’t have any pending CS request in token holder node queue):
Such a case has no impact on the system; hence no action is required.
Case ii (Node is a token holder node): In such a case, MEAFN-MUAV becomes a
self-stabilizing algorithm whenever any new CS request comes into the system (in a
UAV neighborhood) and generates a new token within the UAV neighborhood.
Theorem 4.2.2 MEAFN-MUAV handles token loss.

Proof Loss of token in the system occurs either in the failure of the token holder node
within the UAV neighborhood or the token holder node about to leave the UAV neighbor-
hood. In both cases, the proposed solution behaves as a self-stabilizing algorithm. (Theo-
rem 4.2.1, Case 2 (ii)).

Theorem 4.2.3 MEAFN-MUAV handles UAV failure.

Proof Any UAV failure in the system also destroys the corresponding resource and token
(Assumption 4). Such failure of UAV only impacts the corresponding neighborhood nodes
participating in DME till the time they haven’t joined the new UAV neighborhood and
doesn’t have an impact on the rest of the system.

5 Correctness Proof

Theorem 5.1 Mutual exclusion is guaranteed in every UAV neighborhood and at most k
nodes are allowed to invoke their CS in the entire system i.e., MEAFN-MUAV supports
safety.

Proof By contradiction, to support such a statement, two or more nodes must be in their
CS and there must be more than one token exit in every UAV neighborhood. In such a
case before invoking the CS, nodes send Ask_CS() message to their Token_Holder vari-
able which contains the node identifier of token holder node. Having more than one token
within a UAV neighborhood implies more than one entry in Token_Holder variable. Now
since Token_Holder variable is a single-valued attribute that is updated by broadcasting
Have_Token() message by token holder node (Sect. 4.1.3, MEAFN-MUAV Initialization
(For each Nodei)). Observing a contradiction exists that proves our initial assumption that
two or more nodes must be in their CS and there must be more than one token exit in every
UAV neighborhood is false. Also, our Assumption 4 supports unique token existence in

13
3704 A. S. Parihar, S. K. Chakraborty

every UAV neighborhood. Extending the same to the entire k-UAV FANET with k dupli-
cate resources mounted on each UAV, at most k nodes belonging to different UAV neigh-
borhoods can invoke their CS simultaneously.

Theorem 5.2 Each CS request will be served in a finite time i.e., MEAFN-MUAV is free
from starvation.

Proof, Case 1 (No node failure scenario).

All CS requests generated by nodes are eventually appended in CS_Request_Queue
of token holder node belongs to their UAV neighborhood and according gets a chance to
receive the token for invoking their CS (Sect. 4.1.3, Algorithm 4: Step 5–8).
Case 2 (Node failure scenario) In case the failure node has the token, Theorem 4.2.1,
Case 2(ii) holds true and the system observes partial starvation till the time a new CS
request has not been generated by any node within the same UAV neighborhood. After-
ward, MEAFN-MUAV declares the previous token holder as a new token holder and starts
granting the CS requests for the nodes available in CS_Request_Queue of the new token
holder.

6 Simulation Results and Performance Metrics

6.1 Simulation Setup and Analysis Through The ONE Simulator

For the simulation purpose, we have used The opportunistic network environment (The
ONE) simulator [46], which can generate node movement with various supported move-
ment models and also provide a graphical view of messages passing between the nodes
in a real-time system. In our simulation model, we create a multi-UAV FANET with 5
interconnected UAVs with transmitting speed of 10–20 mbps and transmitting range of
10 km. Each UAV has a buffer size of 1 gb and follows the ShortestPathMapBasedMove-
ment movement model to support a fixed range of movement. Firstly, we collect the various
message stats during our finite simulation time of 3600 s with 25 nodes having a transmit-
ting range of 800 m with RandomWaypoint movement model. The same can be seen in
Fig. 3a–d with different node moving speeds. With a continuous random number of mes-
sage packets, nodes within the system start their communication with message passing and
these messages are relayed over multiple internal nodes in the network as shown in Fig. 3a.
Also in Fig. 3b, with delivery probability in the range of 0.73–0.93, an average number of
hops are captured via sharing a message among two different nodes. Latency and buffer
time count can be seen in Fig. 3c, d. Sample instances of our simulator during this experi-
mental setup are available in Fig. 4a, b.
Secondly, while keeping all the basic configurations the same as earlier, we have con-
tinuously increased the number of nodes in the system. Figure 5 contains the various stats
(discussed earlier) while performing on such a simulation setup. Figures 6a, b are the
sample snapshot with 75 nodes with a moving speed of 40–50 m/s. Under such a simula-
tion model, few observations have been made. Average latency to the system is getting
decreased in both the cases of a higher number of nodes and as well as the higher node
movement. Message delivery probability lies in the range of 0.70–0.85 while the overhead
ratio is getting increased with a higher number of nodes and larger node movements.

13
Token Based k‑Mutual Exclusion for Multi‑UAV FANET 3705

Fig. 3  Various message stats with 5-UAV FANET system with different node movement speeds using The
ONE simulator. a Message packet counts. b Number stats. c Latency count. d Buffer time in seconds

6.2 Performance Metrics of MEAFN‑MUAV

6.2.1 Message Complexity

Let us discuss the total number of message exchanges during a CS invocation by any pro-
cess in all the aspects of best, average and worst-case by our algorithm. MEAFN-MUAV
has been organized in the form of Algorithm 1–7 (Sect. 4.1.3). In the best case, the token
holder node with no pending CS requests asks for the CS to invoke, hence requires no mes-
sage exchange and the node can enter its CS. The same case can be applied to each UAV
neighborhood and overall, this solution requires no message exchange in the best case.
In worst case, all existing nodes (say N) in the system can come within the communica-
tion range or neighborhood of a particular UAV, say UAVx and ask for the CS invocation.
With N node multi-UAV FANET, in such case, token holder node of UAVx contains all CS
requests in its CS_Request_Queue and the last CS request can be served after N message
exchanges. Hence overall message complexity of suggested work in worst case is O(N).
Likewise, average message complexity is the case where nodes are equally distributed in
all UAV neighborhoods and with the total number of M-UAVs in the system, this defines
message complexity as O(N/M) in the average case. Figure 7a–c represents average mes-
sage exchanges per CS execution by MEAFN-MUAV through our simulation setup (dis-
cussed in Sect. 6.1).

13
3706 A. S. Parihar, S. K. Chakraborty

Fig. 4  The ONE simulator snapshot of 5-UAV FANET with 25 nodes. a At simulation time = 0 s. b At
simulation time = 233 s

13
Token Based k‑Mutual Exclusion for Multi‑UAV FANET 3707

Fig. 5  Various message stats with 5-UAV FANET system with different node movement speed using The
ONE simulator. a Message packet counts. b Number stats. c Latency count. d Buffer time in seconds

6.2.2 Throughput

Providing duplicate shared objects or resources can be considered as a strong feature of

a distributed system. MEAFN-MUAV works in a multi-UAV FANET environment and
allows k duplicated resources mounted on available UAVs that facilitate processes at most
k to invoke their CS simultaneously. Through this parallel execution of invoking multi-
ple CS by the nodes, the overall throughput of the system improves. Through our simula-
tion setup (discussed in Sect. 6.1), system throughput is measured on a different number of
UAVs available in multi-UAV FANET. This can be seen in Fig. 8a–c.

6.2.3 Synchronization Delay

Once the token holder node releases its CS, the total time required for the algorithm to
allow the next node to invoke its CS is the time required to propagate the message token
holder node to the next pending CS request node in CS_Request_Queue. Hence consider-
ing the message propagation delay as T defines synchronization delay in MEAFN-MUAV
also T.

7 Conclusion and Future Scope

This research paper presents a token-based k mutual exclusion algorithm for multi-
UAV FANET as mutual exclusion algorithm for flying network-multi UAV with its
correctness proof, performance metrics and fault handling capabilities. We began our

13
3708 A. S. Parihar, S. K. Chakraborty

Fig. 6  The ONE simulator snapshot of 5-UAV FANET with 75 nodes. a At simulation time = 0 s. b At
simulation time = 229.5 s

discussion with the background knowledge of mutual exclusion in a distributed system

along with its classification, basic requirements, and performance metrics. Thereafter,
we discussed the flying ad hoc network with its properties and applications along with

13
Token Based k‑Mutual Exclusion for Multi‑UAV FANET 3709

Fig. 7  Average message exchanges per CS. a With 5-UAV FANET. b With 10-UAV FANET. c With
20-UAV FANET

13
3710 A. S. Parihar, S. K. Chakraborty

Fig. 8  MEAFN-MUAV system throughput. a With 5-UAV FANET. b With 10-UAV FANET. c With
20-UAV FANET

13
Token Based k‑Mutual Exclusion for Multi‑UAV FANET 3711

the literature survey. We then present our proposed algorithm as MEAFN-MUAV with
its fault handling capabilities and correctness proof thereafter. For our simulation result
and analysis, we have used The opportunistic network environment simulator.
Flying networks have grown in popularity in recent years because of the wide range
of applications that UAVs can perform, including weather monitoring, ground target
detection, remote sensing, remote sensing in agriculture, attacks in hostile areas, disas-
ter monitoring, wind estimation, search and rescue operations, wild/forest fire monitor-
ing, real-time surveillance, goods delivery and transportation, and so on. UAVs are an
important part of flying networks since they can fly without human intervention and
can be controlled remotely. During operations, information sharing in such a network
is also critical. Distributed mutual exclusion ensures that nodes can only access shared
information in a mutually exclusive way. Such algorithms in FANETs have received
little attention to date, and in this work, we offer MEAFN-MUAV, a token-based k
mutual exclusion algorithm for multi-UAV FANETs. This approach operates in a multi-
UAV FANET environment, allowing k duplicated resources to be mounted on avail-
able UAVs, allowing up to k processes to invoke their CS at the same time, with more
efficient performance than a single-UAV FANET. This solution is also fault-tolerant to
UAV failure in the system, whereas with a single-UAV FANET, this is treated as a sys-
tem failure.
Finally, we believe that the research presented in this article will be valuable and can be
used as a quick reference for learning about distributed mutual exclusion in the flying net-
work context. Identifying and monitoring the pattern of node migrations in such dynamic
networks through trending machine understanding technologies could be a future focus of
this research. Apart from this, a tamper-proof transaction might be another future aspect of
this study through blockchain technologies.

Funding This study was not funded by any organization.

Data Availability Not applicable.

Code Availability Not applicable.

Declarations
Conflict of interest The authors have no a conflict of interest in relation to this work.

References
1. Tanenbaum, A. S., and Steen, M. V. (2013). Distributed Systems. Pearson Education
2. Parihar, A. S., & Chakraborty, S. K. (2021). Token-based approach in distributed mutual exclusion
algorithms: A review and direction to future research. The Journal of Supercomputing, 77, 14305–
14355. https://​doi.​org/​10.​1007/​s11227-​021-​03802-8
3. Parihar, A. S., & Chakraborty, S. K. (2022). Handling of resource allocation in flying ad hoc network
through dynamic graph modeling. Multimedia Tools and Applications, 81(13), 18641–18669. https://​
doi.​org/​10.​1007/​s11042-​022-​11950-z

13
3712 A. S. Parihar, S. K. Chakraborty

4. Parihar, A. S., & Chakraborty, S. K. (2022). A cross-sectional study on distributed mutual exclu-
sion algorithms for ad hoc networks. In D. Gupta, K. Sambyo, M. Prasad, & S. Agarwal. (Eds.),
Proceedings of international conference onadvanced machine intelligence and signal processing.
Pattern Recognition and Data Analysis with Applications. Singapore: Springer. https://​doi.​org/​10.​
1007/​978-​981-​19-​1520-8.
5. Parihar, A. S., & Chakraborty, S. K. (2022). A simple R-UAV permission-based distributed
mutual exclusion in FANET. Wireless Networks, 28(2), 779–795. https://​doi.​org/​10.​1007/​
s11276-​022-​02889-y
6. Jain, M., & Saxena, R. (2017). Overview of VANET: Requirements and its routing protocols. In:
2017 International Conference on Communication and Signal Processing (ICCSP). https://​doi.​org/​
10.​1109/​iccsp.​2017.​82867​42
7. Srivastava, A., & Prakash, J. (2021). Future FANET with application and enabling techniques:
Anatomization and sustainability issues. Computer Science Review, 39(2021), 100359. https://​doi.​
org/​10.​1016/j.​cosrev.​2020.​100359
8. Chriki, A., Touati, H., Snoussi, H., & Kamoun, F. (2019). FANET: communication, mobility mod-
els and security issues. Computer Networks. https://​doi.​org/​10.​1016/j.​comnet.​2019.​106877
9. Curry, J. A., Maslanik, J., Holland, G., et al. (2004). Applications of aerosondes in the Arctic.
Bulletin of the American Meteorological Society, 85(12), 1855–1861. https://​doi.​org/​10.​1175/​
BAMS-​85-​12-​1855
10. York, G., & Pack, D. J. (2012). Ground target detection using cooperative unmanned aerial sys-
tems. Journal of Intelligent and Robotic Systems, 65(1–4), 473–478. https://​doi.​org/​10.​1007/​
s10846-​011-​9590-4
11. Zhu, S., Wang, D., & Boon, L. C. (2013). Ground target tracking using UAV with input con-
straints. Journal of Intelligent and Robotic Systems, 69(1–4), 417–429. https://​doi.​org/​10.​1007/​
s10846-​012-​9737-y
12. Corrigan, C. E., Roberts, G. C., Ramana, M. V., et al. (2008). Capturing vertical profiles of aerosols
and black carbon over the Indian Ocean using autonomous unmanned aerial vehicles. Atmospheric
Chemistry and Physics, 8(3), 737–747. https://​doi.​org/​10.​5194/​acp-8-​737-​2008
13. Xiang, H., & Tian, L. (2011). Development of a low-cost agricultural remote sensing system based
on an autonomous unmanned aerial vehicle (UAV). Biosystems Engineering, 108(2), 174–190.
https://​doi.​org/​10.​1016/j.​biosy​stems​eng.​2010.​11.​010
14. Cevik, P., Kocaman, I., Akgul, A. S., et al. (2013). The small and silent force multiplier: A swarm
UAVelectronic attack. Journal of Intelligent and Robotic Systems, 70(1–4), 595–608. https://​doi.​
org/​10.​1007/​s10846-​012-​9698-1
15. Maza, I., Caballero, F., Capitn, J., et al. (2011). Experimental results in multi-UAV coordination
for disaster management and civil security applications. Journal of Intelligent and Robotic Systems,
61(1–4), 563–585. https://​doi.​org/​10.​1007/​s10846-​010-​9497-5
16. Cho, A., Kim, J., Lee, S., & Kee, C. (2011). Wind estimation and airspeed calibration using a UAV
with a single-antenna GPS receiver and pitot tube. IEEE Transactions on Aerospace and Electronic
Systems, 47(1), 109–117. https://​doi.​org/​10.​1109/​TAES.​2011.​57056​63
17. De Freitas, E. P., Heimfarth, T., Netto, I. F., Lino, C. E., Pereira, C. E., Ferreira, A. M., Wagner, F.
R., & Larsson, T. (2010). UAV relay network to support WSN connectivity. In Proc. IEEE int. con-
gress on ultra modern telecommunications and control systems and workshops (ICUMT), Moscow,
Russia (pp. 309–314). https://​doi.​org/​10.​1109/​ICUMT.​2010.​56766​21
18. Jiang, F., & Swindlehurst, A.L. (2010). Dynamic UAV relay positioning for the ground–to-air
uplink. In Proceedings of the IEEE GLOBECOM workshops (GC workshop), IEEE (pp. 1766–
1770). https://​doi.​org/​10.​1109/​GLOCO​MW.​2010.​57002​45
19. Sahingoz, O.K. (2013). Multi-level dynamic key management for scalable wireless sensor networks
with UAV. In Ubiquitous information technologies and applications, Springer, Dordrecht (pp.
11–19). https://​doi.​org/​10.​1007/​978-​94-​007-​5857-5_2
20. Sahingoz, O. K. (2013). Large scale wireless sensor networks with multi-level dynamic key man-
agement scheme. Journal of Systems Architecture, 59(9), 801–807. https://​doi.​org/​10.​1016/j.​sysarc.​
2013.​05.​022
21. Manathara, J. G., Sujit, P., & Beard, R. W. (2011). Multiple UAV coalitions for a search and pros-
ecute mission. Journal of Intelligent & Robotic Systems, 62(1), 125–158. https://​doi.​org/​10.​1007/​
s10846-​010-​9439-2
22. Asadpour, M., Giustiniano, D., Hummel, K. A., & Egli, S. (2013). UAV networks in rescue mis-
sions. In Proceedings of the 8th ACM international workshop on wireless network testbeds, experi-
mental evaluation & characterization - WiNTECH ’13. https://​doi.​org/​10.​1145/​25054​69.​25064​91

13
Token Based k‑Mutual Exclusion for Multi‑UAV FANET 3713

23. Barrado, C., Messeguer, R., & L´opez, J., Pastor, E., Santamaria, E. and P. Royo,. (2010). Wildfire
monitoring using a mixed air-ground mobile network. IEEE Pervasive Computing, 9(4), 24–32.
https://​doi.​org/​10.​1109/​MPRV.​2010.​54
24. Parihar, A.S., Prasad, D., Gautam, A.S. & Chakraborty, S.K. (2021). Proposed end-to-end auto-
mated e-voting through blockchain technology to increase voter’s turnout. In: Prateek, M., Singh,
T.P., Choudhury, T., Pandey, H.M. & Gia Nhu, N. (eds) Proceedings of international conference on
machine intelligence and data science applications. Algorithms for intelligent systems. Springer, Sin-
gapore. https://​doi.​org/​10.​1007/​978-​981-​33-​4087-9_5
25. Semsch, E., Jakob, M., Pavlicek, D. & M. Pechoucek, (2009). Autonomous UAV surveillance in com-
plex urban environments. In Proc. IEEE/WIC/ACM int. joint conf. on web intelligence and intelligent
agent technologies (WI-IAT), Washington, DC, USA (vol. 2, pp. 82–85). https://​doi.​org/​10.​1109/​WI-​
IAT.​2009.​132
26. Kerr, S. (2014). UAE to develop fleet of drones to deliver public services. The Financial Times, World
News, Volume 12.
27. Walter, J. E., Welch, J. L., & Vaidya, N. H. (2001). A mutual exclusion algorithm for ad hoc mobile
networks. Wireless Networks, 7, 585–600. https://​doi.​org/​10.​1023/A:​10123​63200​403
28. Benchaïba, M., Bouabdallah, A., Badache, N., & Ahmed-Nacer, M. (2004). Distributed mutual exclu-
sion algorithms in mobile ad hoc networks. ACM SIGOPS Operating Systems Review, 38(1), 74–89.
https://​doi.​org/​10.​1145/​974104.​974111
29. Chen, Y., & Welch, J. L. (2005). Self-stabilizing dynamic mutual exclusion for mobile ad hoc net-
works. Journal of Parallel and Distributed Computing, 65(9), 1072–1089. https://​doi.​org/​10.​1016/j.​
jpdc.​2005.​03.​009
30. Mellier R., Myoupo JF., & Ravelomanana V. (2005). A non-token-based-distributed mutual exclusion
algorithm for single-hop mobile ad hoc networks. In: Belding-Royer E.M., Al Agha K., Pujolle G.
(eds) Mobile and wireless communication networks. MWCN 2004. IFIP international federation for
information processing, Volume 162. Springer, Boston, MA. https://​doi.​org/​10.​1007/0-​387-​23150-1_​
25
31. Wu W., Cao J., & Raynal M. (2007). A dual-token-based fault tolerant mutual exclusion algorithm for
MANETs. In: Zhang H., Olariu S., Cao J., Johnson D.B. (eds) mobile ad-hoc and sensor networks.
MSN 2007. Lecture notes in computer science, Volume 4864. Springer, Berlin, Heidelberg. https://​doi.​
org/​10.​1007/​978-3-​540-​77024-4_​52
32. Thiare O., Naimi M., & Gueroui M. (2007). A group mutual exclusion algorithm for mobile ad hoc
networks. In: Sobh T. (eds) Innovations and advanced techniques in computer and information sci-
ences and engineering. Springer, Dordrecht. https://​doi.​org/​10.​1007/​978-1-​4020-​6268-1_​67
33. Wu, W., Cao, J., & Yang, J. (2008). A fault tolerant mutual exclusion algorithm for mobile ad hoc net-
works. Pervasive and Mobile Computing, 4(1), 139–160. https://​doi.​org/​10.​1016/j.​pmcj.​2007.​08.​001
34. Thiare O. (2012). A token-based group mutual exclusion algorithm for MANETs. In: Kim T., Ko D.,
Vasilakos T., Stoica A., Abawajy J. (eds) Computer applications for communication, networking, and
digital contents. FGCN. Communications in computer and information science, Volume 350. Springer,
Berlin, Heidelberg. https://​doi.​org/​10.​1007/​978-3-​642-​35594-3_​34
35. Sharma B., Bhatia R.S., & Singh A.K. (2011). An O(1/n) protocol for supporting distributed mutual
exclusion in vehicular ad hoc networks. In: Nagamalai D., Renault E., Dhanuskodi M. (eds) Advances
in parallel distributed computing. PDCTA 2011. Communications in computer and information sci-
ence, Volume 203. Springer, Berlin, Heidelberg. https://​doi.​org/​10.​1007/​978-3-​642-​24037-9_​14
36. Tamhane, S. A., & Kumar, M. (2012). A token based distributed algorithm for supporting mutual
exclusion in opportunistic networks. Pervasive and Mobile Computing, 8(5), 795–809. https://​doi.​org/​
10.​1016/j.​pmcj.​2011.​08.​002
37. Wu, W., Zhang, J., Luo, A., & Cao, J. (2015). Distributed mutual exclusion algorithms for intersection
traffic control. IEEE Transactions on Parallel and Distributed Systems, 26(1), 65–74. https://​doi.​org/​
10.​1109/​tpds.​2013.​22970​97
38. Raymond, K. (1989). A tree-based algorithm for distributed mutual exclusion. ACM Transactions on
Computer Systems, 7(1), 61–77. https://​doi.​org/​10.​1145/​58564.​59295
39. Zheng, W., Song, L. X., & Mei’an, L. (2007). Ad hoc distributed mutual exclusion algorithm based
on token-asking. Journal of Systems Engineering and Electronics, 18(2), 398–406. https://​doi.​org/​10.​
1016/​s1004-​4132(07)​60104-2
40. Lamport, L. (1978). Time, clocks, and the ordering of events in a distributed system. Communications
of the ACM, 21(7), 558–565. https://​doi.​org/​10.​1145/​359545.​359563
41. Sharma, B., Bhatia, R. S., & Singh, A. (2014). A token based protocol for mutual exclusion in mobile
ad hoc networks. Journal of Information Processing Systems, 10, 36–54. https://​doi.​org/​10.​3745/​JIPS.​
2014.​10.1.​036

13
3714 A. S. Parihar, S. K. Chakraborty

42. Khanna, A., Rodrigues, J. J. P. C., Gupta, N., Swaroop, A., Gupta, D., Saleem, K., & De Albuquerque,
V. H. C. (2019). A mutual exclusion algorithm for flying ad hoc networks. Computers & Electrical
Engineering, 76, 82–93. https://​doi.​org/​10.​1016/j.​compe​leceng.​2019.​03.​005
43. Khanna, A., Rodrigues, J. J. P. C., Gupta, N., Swaroop, A., & Gupta, D. (2020). Local mutual exclu-
sion algorithm using fuzzy logic for flying ad hoc networks. Computer Communications, 156, 101–
111. https://​doi.​org/​10.​1016/j.​comcom.​2020.​03.​036
44. Shehu, H. A., Sharif, M. H., & Ramadan, R. A. (2020). Distributed mutual exclusion algorithms for
intersection traffic problems. IEEE Access, 8, 138277–138296. https://​doi.​org/​10.​1109/​ACCESS.​2020.​
30125​73
45. Attiya, H., Kogan, A., & Welch, J. L. (2008). Efficient and robust local mutual exclusion in mobile ad
hoc networks. In The 28th International Conference on Distributed Computing Systems, Beijing (pp.
321–328). https://​doi.​org/​10.​1109/​ICDCS.​2008.​82
46. Keranen, A., Ott, J., & Karkkainen, T. (2009). The ONE simulator for DTN protocol evaluation. In
Proc. 2nd international conference on simulation tools and techniques (pp. 55:1–55:10). https://​doi.​
org/​10.​4108/​ICST.​SIMUT​OOLS2​009.​5674

Publisher’s Note Springer Nature remains neutral with regard to jurisdictional claims in published maps and
institutional affiliations.

Ashish Singh Parihar received his Bachelor of Engineering (B.E.) in

the department of Computer Science and Engineering from Oriental
Institute of Science and Technology (India) and also completed his
Master of Technology (M.Tech) in the department of Computer Sci-
ence and Engineering from National Institute of Technology (NIT),
Arunachal Pradesh, Govt. of India. He served more than five years in
the industry as a senior developer. Currently during the presentation of
this work, he is an assistant professor in the KIET group of institu-
tions, Delhi-NCR, Ghaziabad (UP), India and also, pursuing his Ph.D.
under the supervision of Dr. Swarnendu Kumar Chakraborty. His
research areas are distributed systems, machine learning, blockchain
and big data. Some of his recent papers are published in most reputed
journals like, The Journal of Supercomputing (Springer), Wireless
Networks (Springer), Multimedia Tools and Applications (Springer)
etc.

Swarnendu Kumar Chakraborty is working as an Assistant Professor

in the department of Computer Science and Engineering at National
Institute of Technology (NIT), Arunachal Pradesh, Govt. of India. He
has more than twelve years of teaching experience. His research areas
are distributed systems, advanced error control, cryptography and
Information Security. He is the author of more than 30 peer-reviewed
publications in reputed journals.

13