RAHE: A Robust Attribute-Based Aggregate Scheme Enhanced with

Homomorphic Encryption for 5G-Connected Delivery Drones

Aagii Mariam Thomas

and Sana Belguith

Department of Computer Science, The University of Bristol, U.K.

Keywords:

Drones, Attribute-Based Encryption, Homomorphic Encryption, Data Aggregation.

Abstract:

Unmanned Aerial Vehicles (UAVs), commonly known as drones, have become essential for transporting pack-

ages, food, medicines, and other goods due to the growing demand for fast and efﬁcient delivery services. The

implementation of 5G technology provides high-speed, low-latency, and reliable connectivity, which allows

drones to exchange mission-critical data effectively. However, drones utilizing 5G networks are susceptible

to security threats that could compromise essential security requirements such as conﬁdentiality, authentica-

tion, integrity, and availability. In this paper, we propose a robust communication framework designed for

secure interactions among 5G-connected delivery drones. Our framework relies on a novel Attribute-Based

Encryption with Aggregation that is composed by an enhanced multi-level Attribute-Based Encryption (ABE)

scheme with Homomorphic Encryption (HE). By integrating HE with the ABE scheme, the Ground Control

Station (GCS) and the parent drone can decrypt mission-critical messages as required. This ensures that only

authorized entities have access to sensitive data. Additionally, in scenarios that require data aggregation with-

out exposing the underlying content, the HE property within the ABE scheme facilitates this process. As a

result, encrypted data subsets can be aggregated anywhere in the network without the need for decryption,

thereby preserving data conﬁdentiality and enhancing both communication and computational efﬁciency. We

utilize a hierarchical Chain-Based Data Aggregation (CBDA) model for the structural organization of drones,

which enhances communication efﬁciency and reduces energy consumption. By integrating multi-level ABE

for ﬂexible and secure access control with HE, our framework effectively addresses major security challenges

faced by 5G-based drone networks, ensuring the security and efﬁcient management of mission-critical data.

1 INTRODUCTION

Unmanned Aerial Vehicles (UAV) a.k.a drones are

now used in many sectors like the military, aviation,

and civil ﬁelds, for purposes like surveillance, search

and rescue, and disaster relief. Recently, drones have

started to gain traction in the transportation sector be-

cause they can deliver goods quickly, even to places

that are difﬁcult to reach. With growing demand

for fast, efﬁcient deliveries, a drone delivery system

could be an ideal solution. Not only are drones faster,

but they also reduce road trafﬁc, which helps cut down

on C O

emissions, beneﬁting the environment.

To function properly, drones need reliable wire-

less connectivity to share information like images,

video, commands, and sensor data with a Ground

Control Station (GCS). A standardized cellular net-

https://orcid.org/0000-0001-8677-3968

https://orcid.org/0000-0003-0069-8552

work with wide coverage can help ensure high-quality

connectivity, which is especially important for drones

operating beyond the line of sight, which means they

are out of the operator’s direct view. This is where 5G

wireless technology comes in, as it can connect many

devices at once with faster data speeds, low latency,

and better reliability than older networks. With 5G,

drone delivery systems can handle their tasks effec-

tively and maintain stable, high-quality connections.

However, using 5G to connect drones for trans-

portation has its challenges, particularly around se-

curity. Drones often operate in both supervised ar-

eas and unsupervised, potentially hostile zones, which

makes them vulnerable to security threats like eaves-

dropping, data tampering, and even hijacking. For ex-

ample, if two drones are communicating and a mali-

cious attacker intercepts the data, they could gather

information to compromise part or all of the com-

munication. Because of this, securing the communi-

cation channels, between drones and between drones

400

Thomas, A. M. and Belguith, S.

RAHE: A Robust Attribute-Based Aggregate Scheme Enhanced with Homomorphic Encryption for 5G-Connected Delivery Drones.

DOI: 10.5220/0013521800003979

In Proceedings of the 22nd International Conference on Security and Cryptography (SECRYPT 2025), pages 400-411

ISBN: 978-989-758-760-3; ISSN: 2184-7711

and the GCS, is crucial to protect against these types

of attacks.

To ensure safe drone operations, especially in de-

livery, we need measures like encryption, access con-

trols, and secure communication protocols. With

these in place, drones can expand securely into de-

livery services, meeting the demand for fast, safe, and

environmentally friendly deliveries, even in challeng-

ing areas. Combining 5G’s high-speed, reliable con-

nectivity with strong security can unlock the potential

of drones in the logistics industry, setting them up as

essential tools for the future of delivery.

Several researchers have explored the various se-

curity threats and vulnerabilities that drone networks

face. Menouar et al. (Menouar et al., 2017) and Alwa-

teer et al. (Alwateer and Loke, 2020) have presented

works discussing security attacks that target drone

communication systems. A comprehensive look at

these security threats is provided by Ossamah et al. in

(Ossamah, 2020), where they detail the speciﬁc vul-

nerabilities and challenges drones face.

Additionally, Guvenc et al. (Guvenc et al., 2018)

conducted a survey on methods to detect and track

unauthorized UAVs, which pose a risk to drone oper-

ations through both physical and cyber-physical at-

tacks, such as hijacking, spooﬁng, snifﬁng, hack-

ing, and network exploitation. Although several re-

searchers have proposed solutions for securing drone

networks, most have focused on Blockchain as the

primary method (Han et al., 2021; Wu et al., 2021).

Encryption techniques, which ensure conﬁdential-

ity for data exchanges within a drone network, are

also widely regarded as highly suitable for protecting

drone communications (Ismael et al., 2021; Cheon

et al., 2018).

In this paper, we present RAHE novel Multi-

Authorisation Attribute Based Encryption (MA-

ABE) scheme with aggregation. This scheme uses the

Homomorphic Encryption properties to enable aggre-

gating ciphertexts without any need for decryption.

To the best of our knowledge, this is the ﬁrst ABE

scheme that offers an aggregation functionality. We

apply and test the proposed RAHE in a real-world

scenario based on drone communications. This pa-

per’s key contributions are as follows:

• We extend the MA-ABE scheme proposed by

Kaaniche et al. (Kaaniche et al., 2018) to support

an aggregation functionality. This allows for ag-

gregating ciphertext without the need for decryp-

tion or re-encryption.

• We leverage a chain-based data aggregation

method to improve communication efﬁciency in

drone networks by organizing the drones into par-

ent and children. This design minimizes energy

consumption and supports scalable, structured

communication between multiple drones and the

GCS, which is crucial in resource-constrained

drone networks.

• Unlike most ABE techniques, the size of the re-

sulting ciphertext does not depend on the number

of attributes involved in the access policy, which

makes our scheme more suitable for bandwidth-

limited applications such as drone communica-

tions.

• By employing a multi-level authorization ABE

scheme, this framework efﬁciently minimizes

the communication overhead between the GCS

and the drones. Instead of communicating di-

rectly with multiple child drones, the GCS sends

mission-critical data only to the parent drone.

This signiﬁcantly reduces the communication

load, as the parent drone is responsible solely

for relaying the encrypted information to its con-

nected child drones. Child drones can selectively

decrypt the relevant data based on their access

permissions, ensuring data conﬁdentiality without

imposing a computational burden on the parent

drone.

• We prove that RAHE scheme is secure under stan-

dard assumptions and achieves Indistinguisha-

bility under Chosen Ciphertext Attack (IND-

CCA). RAHE can defeat the main security at-

tacks against drone transportation systems such as

eavesdropping and injection attacks thanks to the

use of encryption, as well as Man in the middle

attack thanks to the use of CBDA.

• We provide a detailed theoretical and experimen-

tal analysis of the proposed scheme that proves its

efﬁciency and reduced costs making it suitable for

drones applications.

The remainder of this paper is organized as follows:

In section 2, a detailed literature review is presented.

In section 3, a brief description of the background is

presented. In section 4, we introduce the framework

speciﬁcations, which include the system and threat

models as well as the security requirements and se-

curity model. In section 5 a detailed description of

the proposed secure communication framework is ex-

plained. We perform a security analysis in Section

6 based on the deﬁned security requirements before

presenting a performance analysis in Section 7. And

in section 8 an experimental analysis of the proposed

scheme is presented. Finally, we conclude the paper

in Section 9.

RAHE: A Robust Attribute-Based Aggregate Scheme Enhanced with Homomorphic Encryption for 5G-Connected Delivery Drones

401

2 LITERATURE REVIEW

Security in 5G-connected drone networks has been

widely studied, with existing works primarily focus-

ing on blockchain, traditional encryption, and authen-

tication mechanisms. However, these approaches of-

ten fail to balance security, computational efﬁciency,

and scalability in resource-constrained drone environ-

ments. Several studies have identiﬁed critical secu-

rity threats affecting UAV communication. Guvenc et

al. (Guvenc et al., 2018) introduced threat detection

methodologies for unauthorized UAVs but did not

propose encryption-based countermeasures. While

this paper highlights security challenges, it lacks con-

crete cryptographic solutions for securing real-time

drone communication. Blockchain has been explored

as a security solution for UAV networks. Ismael et

al. (Ismael et al., 2021) introduced a lightweight au-

thentication mechanism using blockchain for secure

drone communication. Although blockchain ensures

tamper-proof data integrity, its high computational

and communication overhead limits applicability in

drone networks, where low latency and energy efﬁ-

ciency are critical.

Recent research efforts have addressed emerging

security concerns and proposed novel solutions for

UAV security and privacy. L. He et al. (He et al.,

2025) introduced an efﬁcient threshold attribute-

based proxy signature scheme for UAV networks, im-

proving authentication and integrity in multi-entity

drone systems. Kaushik et al. (Kaushik et al.,

2025) proposed an end-to-end homomorphically en-

crypted UAV navigation system, leveraging a fu-

sion of reinforcement learning and deep neural net-

works to ensure secure autonomous navigation. Meng

et al. (Zhang et al., 2025) designed a lightweight

acoustic ﬁngerprint-based drone authentication sys-

tem, providing an alternative non-cryptographic se-

curity mechanism that enhances UAV identity veri-

ﬁcation. Feng et al. (Feng et al., 2021) introduced

blockchain-integrated ABE for drone authentication,

focusing solely on authentication rather than secure

mission-critical data transmission.

Efﬁcient data aggregation is essential for optimiz-

ing communication in UAV networks. Traditional

ABE schemes lack support for secure data aggre-

gation, necessitating their combination with homo-

morphic encryption. Homomorphic Encryption (HE)

has been employed for data aggregation in various

ﬁelds, including IoT and Wireless Sensor Networks

(WSNs). Loukil et al. (Loukil et al., 2021) com-

bined blockchain and HE for secure IoT data aggrega-

tion, demonstrating how HE can facilitate data com-

putations without decryption. Othman et al. (Othman

et al., 2015) proposed a homomorphic encryption-

based aggregation scheme ensuring message integrity

and ﬂexibility in computational cost and security as-

sumptions. Li et al. (Li et al., 2022) extended HE

for UAV-assisted WSNs, enhancing data aggregation

efﬁciency through blockchain integration. However,

these studies primarily focus on IoT and healthcare

applications, failing to address drone-speciﬁc chal-

lenges such as real-time encrypted communication

and dynamic network topology changes.

These recent studies collectively reinforce the

need for a uniﬁed security framework integrating

ABE, HE, and CBDA to address access control, com-

putational efﬁciency, and secure data aggregation in

5G-enabled UAV networks. Our proposed framework

seeks to bridge these gaps by providing a scalable and

efﬁcient cryptographic model for securing UAV com-

munications in 5G and beyond.

Hu et al. (Hu et al., 2019) introduced Chain-

Based Data Aggregation (CBDA) to enhance energy

efﬁciency in WSNs by structuring nodes in a chain

topology. Wan et al. (Wan et al., 2020) further re-

ﬁned energy-efﬁcient chain communication for peri-

odic data collection in WSNs. While CBDA enhances

structured communication, its integration with ABE

and HE for secure aggregation in UAV networks re-

mains unexplored.

3 BACKGROUND

3.1 Attribute Based Encryption

Attribute-Based Encryption (ABE) (Belguith et al.,

2016) is an encrypted access control mechanism

which is being widely used in multiple applications

such as IoT, cloud computing, vehicular networks,

etc. In ABE, the data owner sets an access policy to

deﬁne the users who can access the data. The data is

encrypted with respect to the access policy which is

embedded in the ciphertext in order to enforce access

control policies. A user is able to decrypt a ciphertext

only if there is a match between their attributes and

the access policy embedded in the ciphertext. ABE

has a signiﬁcant advantage over the traditional public

key encryption primitives as it achieves ﬂexible one-

to-many encryption instead of one-to-one.

3.2 Homomorphic Encryption

Homomorphic encryption (HE) is an encryption

scheme that allows computation to be performed

over encrypted data without any need to decrypt it

(Shrestha and Kim, 2019). Just like other forms of

SECRYPT 2025 - 22nd International Conference on Security and Cryptography

402

encryption, homomorphic encryption uses a public

key to encrypt the data. It uses an algebraic system

to allow functions to be performed on the data while

it is still encrypted. Then, only the individual with

the matching private key can access the unencrypted

data after the functions and manipulation are com-

plete. This allows the data to be and remain secure

and private even when someone is using it. HE has

been applied in different ﬁelds because of its ability

to compute on encrypted data while providing secu-

rity and privacy to the users.

3.3 Chain Based Data Aggregation

Chain-Based Data Aggregation (CBDA) (Hu et al.,

2019) is a hierarchical method used to efﬁciently ag-

gregate data in wireless sensor networks by forming a

structured chain among nodes. This approach reduces

data trafﬁc and prevents eavesdropping by establish-

ing secure paths for data transmission. In the con-

text of multiple drones, CBDA enables a tree topology

where one drone functions as the parent node, while

others connect as leaf nodes. A chain is constructed

among drones based on their residual energy, prior-

itizing nodes with higher energy as leaf nodes. This

structure supports efﬁcient data aggregation by allow-

ing each node to relay information through the chain,

minimizing overall energy consumption and ensuring

secure, structured communication.

4 FRAMEWORK

SPECIFICATION

4.1 System and Threat Models

In our scheme, we consider a swarm of delivery

drones and a base station that sends the signal and

data to and from. In the proposed model (Figure 1),

there are four main components:

• Certiﬁcate authority (CA) – CA is responsible for

the registration and issue of keys to the Ground

Controller and the drones.

• Ground Control Station (GCS)- GCS sends the

information relating to the remote control com-

mands, telemetry information, and mission sensor

information to the drones. This infrastructure uses

wireless communication.

• Drones – They are responsible for delivering the

packages to the destination based on the com-

mands received from GCS.

• 5G network Base station – They provide a reliable

connection between drones and GCS to transfer

data between them

Figure 1.

The possible adversaries are:

1. Malicious Drone – They are drones that are not

part of the swarm delivery drones and try to inter-

fere with drone-to-drone communication as well

as drone-to-GCS.

2. Malicious ground Controller – They send false in-

formation to delivery drones.

3. A User Equipment – They try to listen to the in-

formation from the GCS and also try to give false

signals to UAVs and control them.

4.2 Security Requirements

In order to have proper secure communication be-

tween the drones and GCS we need to consider the

following requirements:

• Access Control: data exchanged between drones

and between GCS and drones should be secure

against unauthorised access.

• Data Conﬁdentiality: ensures the secrecy of the

exchanged data between drones and with the

GCS.

• Data Aggregation: maintains lower energy con-

sumption, removes data redundancy, improves

network overhead

• Low processing and storage overheads - the

framework should provide low communication

and processing overheads.

4.3 Security Model

To formally demonstrate the security of the proposed

scheme, we will prove that it satisﬁes Indistinguisha-

bility under Chosen Ciphertext Attack (IND-CCA)

RAHE: A Robust Attribute-Based Aggregate Scheme Enhanced with Homomorphic Encryption for 5G-Connected Delivery Drones

403

standards. IND-CCA security guarantees that an ad-

versary cannot differentiate between the encryptions

of any two chosen messages, even when they can

make adaptive decryption queries. To evaluate the

scheme’s resistance to chosen-ciphertext attacks, the

security model involves an interaction between an ad-

versary A and a challenger C , where A attempts to

distinguish between two challenge messages while

having access to adaptive decryption queries.

Game Initialization - The adversary A selects a set

of encryption attributes A

enc

, corresponding to the at-

tributes deﬁned in the access policy. These attributes

are associated with threshold values t

, where t

∈

,...t

}, where m is the number of threshold val-

ues. A sends A

enc

to the challenger C .

Setup - The challenger C runs the setup algorithm

and generates the public parameters pp and the ac-

cess policies, with associated thresholds t

for each

attribute set S and sends it to the adversary A.

Decryption Query Phase - During this phase,

A makes adaptive decryption and key generation

queries:

Key Generation: The adversary A requests a decryp-

tion key for each session i by selecting an encryption

attribute set A

corresponding to a threshold value t

where t

∈ {t

,...t

}. The challenger C processes

this request by generating the decryption key SK

and providing it to the adversary A, ensuring that the

key complies with the required threshold t

Decryption Query: The adversary A submits a re-

quest to decrypt a ciphertext C under a threshold t

while considering the encryption attribute set A

enc

The challenger C responds by generating the corre-

sponding secret key SK

enc

, ensuring that the attribute

set A

enc

satisﬁes the predeﬁned policy conditions. The

challenger then executes the decryption algorithm,

which results in either the original plaintext message

M if the decryption policy is met, or a rejection sym-

bol ⊥ if the access conditions are not satisﬁed.

Challenge Phase - In the Challenge Phase, adversary

A selects two plaintext messages of equal length and

submits them to challenger C , along with an attribute

set that has not been used in any decryption queries.

The challenger C randomly selects one of the two

messages, encrypts it according to the access policy

and returns the challenge ciphertext to A . The goal of

the adversary is to determine which of the two mes-

sages was encrypted, using any information gained

from previous queries.

Guess Phase - In the Guess Phase, A attempts to cor-

rectly identify the encrypted message by outputting a

guess bit. The security of the scheme is conﬁrmed

if the adversary’s probability of guessing correctly

remains close to random (50%), meaning they gain

no meaningful advantage. The adversary’s success is

mathematically deﬁned as:

Adv

[Exp

Con f

)] =



P(a = ˆa) −



Thus, the security model ensures that this advantage

remains negligible, proving that A cannot distinguish

between ciphertexts with probability greater than one-

half. This conﬁrms the framework’s resistance to

adaptive chosen-ciphertext attacks (IND-CCA).

5 THE PROPOSED

FRAMEWORK

The framework proposes a secure and efﬁcient com-

munication between drones and ground control sta-

tions (GCS), with GCS and a parent drone holding

full authority. In a network of multiple drones, by the

use of Chain Based Data Aggregation (CBDA), a tree

topology is formed, where one drone acts as the par-

ent node and others connect as leaf nodes (Hu et al.,

2019). The chain formation process is as follows.

• Tree Formation: Parent drone broadcasts

invite messages. Drones without a parent re-

ply with join and become children. Each child

repeats this to expand the tree. Drones receiving

multiple invites select a parent randomly.

• Chain Formation: Leaf drones form chains by

sequentially connecting to neighbors that reply to

chain messages, based on residual energy or ran-

dom choice.

When the Ground Control Station wants to send a

message to the drone or drone to GCS we are using

the ABE algorithm with the homomorphic encryption

scheme. Attribute-Based encryption is a good candi-

date for sending mission data to the drones. We as-

sume that the chain has been already formed and the

parent node is the responsible node for relaying the

information to children nodes. To this end, we use

a multi-level authorisation ABE scheme where the

parent node can decrypt all the information but also

forward the subset of data needed to children nodes

(Kaaniche et al., 2018). The GCS will ﬁrst deﬁne an

access policy composed by a set of attributes and a set

of thresholds.

For example, the set attributes can be as follows:

S = (Parent,Destination,Location), while the thresh-

olds are (t

) = (3,2). In this case, the GCS has set

two access policies (t

,S) and (t

,S) where only the

parent node can satisfy the ﬁrst one and therefore ac-

cess the related data and a child node can only match

the second access policy. Thus, only the parent node

SECRYPT 2025 - 22nd International Conference on Security and Cryptography

404

can decrypt and access the full encrypted data sent by

the GCS and a child node can access a restricted sub-

set of data. Here if more than one child drone sends

the data to the parent drone, then the messages can be

aggregated with the help of homomorphic encryption

scheme. This helps to add the encrypted messages

together without the need of decryption.

5.1 The RAHE Scheme

This scheme consists of ﬁve algorithms:

• Setup : Setup algorithm is run by the Certiﬁcate

Authority and it takes as input a security param-

eter. It outputs the public parameter (pp) and the

master secret key (MSK).

• KeyGen : Keygen algorithm is executed by the

CA to derive the secret keys for GCS and drones

based on the attributes.

• Encrypt : The encryption algorithm is performed

by the encrypting entity and it takes pp, multi

level threshold access policy and the message.

This algorithm outputs an encrypted message re-

ferred to as E.

• Aggregation : The aggregation algorithm is per-

formed when there is a need to add two encrypted

messages together even without decrypting the

message.

• Decrypt : The decryption algorithm is executed

by the user. It takes as inputs the pp, user’s secret

key, multi level threshold access policy, and the

encrypted message. If the algorithm is successful

then the message is obtained.

5.2 Concrete Construction of the

Framework

This scheme relies on ﬁve algorithms deﬁned as fol-

lows:

Setup. The certiﬁcate authority, responsible for se-

cure initialization between drones and GCS, selects

a bilinear group ˆe, p,G

,G of prime order p,such

that ˆe : G

× G

→ G. It selects random generator

g ∈ G

and a random value z ∈ G

along with a set of

generators {q

}{ j = 1,··· , m} such that m = |T |

is the cardinal of the threshold universe T , required

for drone-GCS communication. In addition, it deﬁnes

an encoding function ε such that ε : U → (Z/pZ)

∗

where |U| = n and U is an attribute universe. For

each attribute a ∈ U, corresponds to a property of ei-

ther a drone or the GCS. The encoded attribute values

ξ(a

) = x

are pairwise different, where i ∈ [1,n].

Then, the setup algorithm selects a set O =

,··· , o

n−1

} of n − 1 pairwise distinct elements of

(Z/pZ)

∗

, ensuring they differ from ξ(a

), for all a

∈

Let O

represent the subset o

,··· , o

. Thus, the

setup algorithm computes u deﬁned as u = gαγ and

outputs the global public parameters pp as follows:

pp = {G

,G, ˆe,u,{q

αγ

}

{i=0,···,2n−1; j=1,··· ,m}

O,ξ,z,{ ˆe(g

)}

{ j=1,···,m}

}

The Certiﬁcate authority holds the master key, de-

ﬁned as

msk = (g,α, γ)

where α,γ are two random values from (Z/pZ)

∗

Keygen. For any subset A

⊂U of attributes associ-

ated with a drone or the GCS, the certiﬁcate authority

selects a random value k

∈ (Z/pZ)

∗

and computes

the corresponding secret key:

sk = ({g

γ+ξ(a)

}a ∈ A

,{q

}

i=0,···,n−2, j=1,··· ,m

−1

}

j=1,···m

= (sk1,sk2,sk3)

Encrypt. Let ({t

}{ j = 1,· ·· , l}, S) be the access

policy where {t

} is the set of threshold values and

S ⊂ U is an attribute set that a drone or GCS must

process to access the message. The entity E(eg.,a

drone or GCS) encrypts a message D using a random

κ ∈ Z/pZ and a random blinding factor r

∈ Z/pZ,

producing ciphertext E = (E

1 j

2 j

3 j

){ j = 1,· ·· , l}

deﬁned as:

1 j

= g

−kαγ

2 j

= q

kα

∏

a∈S

(γ+ξ(a))

∏

o∈O

n+t

−1−s

(γ+o)

3 j

= d

ˆe(g

)

= d

The entity outputs the encrypted message D as E =

1 j

2 j

3 j

){ j = 1,· ·· , l}.

Aggregate. For two messages D

and D

from dif-

ferent drones, aggregation proceeds as follows :

For message D

11 j

= g

−k

αγ

21 j

= q

∏

a∈S

(γ+ξ(a))

∏

o∈O

n+t

−1−s

(γ+o)

RAHE: A Robust Attribute-Based Aggregate Scheme Enhanced with Homomorphic Encryption for 5G-Connected Delivery Drones

405

31 j

= d

ˆe(g

)

= d

For message D

12 j

= g

−k

αγ

22 j

= q

∏

a∈S

(γ+ξ(a))

∏

o∈O

n+t

−1−s

(γ+o)

32 j

= d

ˆe(g

)

= d

Now, the aggregated encrypted message :

1 j

= E

11 j

12 j

= g

−(k

)αγ

2 j

= E

3 j

= E

= (d

+ d

).K

= ˆe(g,q

)

α(k

)

Decrypt. In the decryption phase, there is two sub-

cases, the ﬁrst one is when the received message is

the encrypted message generated by the Encrypt algo-

rithm, and the second sub-case is when the aggregate

algorithm is applied:

The decrypting entity U, drone or GCS, with a set

of attributes A

that satisﬁes the threshold |A

∩ S|=

can decrypt the encrypted message d

under the ac-

cess policy ({t

}{ j = 1, ··· ,l},S)), at the threshold

level t

. This process allows only authorized drones

or the GCS with sufﬁcient attributes to access the in-

formation.

For this purpose, for all a ∈ A

,U, the decrypting

entity U (drone or GCS) ﬁrst summates its attributes,

with respect to the threshold t

satisﬁed by its certiﬁed

attributes, as follows:

A = aggreg({g

γ+ξ(a)

,ξ(a)}

a∈A

) = g

∏

a∈A

(γ+ξ(a))

Subcase 1:

For the encrypt case, U uses the summated secret key

A and the ciphertext element E

2, j

, associated with the

satisﬁed threshold t

, to compute:

= ˆe(g

∏

a∈A

(γ+ξ(a))

2 j

)

= ˆe(g,q

)

κα

∏

a∈S\A

(γ+ξ(a))

∏

o∈O

n+t

−1−s

(γ+o)

Then, U deﬁnes the polynomial P

,S)

(γ) to help

with attribute aggregation:

,S)

(γ) =

(

∏

a∈S∪O

n+t

−1−s

(γ + ξ(a))

−

∏

a∈S∪O

n+t

−1−s

ξ(a))

Afterwards, U uses the summated secret key A and

the sk

key elements to compute:

[ ˆe(E

1 j

,A)

(γ)

]

∏

a∈S∪O

n+t

−1−s

ξ(a)

= e(g,q

)

kακ

Next, the decrypting entity U uses its secret key

element sk

, related to t

to deduce the deciphering

key K

as follows:

= ˆe(E

1 j

,sk

) ˆe(g,q

)

kκ

where,

ˆe(E

1 j

,sk

) = ˆe(g

−kαγ

−1

)

Finally, combining terms as needed:

ˆe(g

−kαγ

−1

) ˆe(g

−1

)e(z

−1

) ˆe(g,q

)

kκ

Since the ﬁrst and last terms contain the encryp-

tion randomness, and the second and third terms in-

volves the message d and the random factor z, these

terms enable the recovery of K

= ˆe(g,q

)

ακ

Finally, U recovers the sub-message d

, with re-

spect to related access level t

, by computing

3 j

ˆe(g,q

)

ακ

Thus, the message d

is decrypted and obtained,

allowing drone or GCS to retrieve the necessary in-

formation securely.

Subcase 2:

In this aggregate case, U uses the summated secret

key A and the ciphertext element E

2 j

, associated with

the satisﬁed threshold t

, to compute:

= ˆe(g

∏

a∈A

(γ+ξ(a))

2 j

)

= ˆe(g,q

)

)α

∏

a∈S\A

(γ+ξ(a))

∏

o∈O

n+t

−1−s

(γ+o)

Then, U deﬁnes the polynomial P

,S)

(γ) to

help with attribute aggregation:

,S)

(γ) =

(

∏

a∈S∪O

n+t

−1−s

(γ + ξ(a))

−

∏

a∈S∪O

n+t

−1−s

ξ(a))

Afterwards, U uses the summated secret key A and

the sk

key elements to compute:

SECRYPT 2025 - 22nd International Conference on Security and Cryptography

406

[ ˆe(E

1 j

,A)

(γ)

]

∏

a∈S∪O

n+t

−1−s

ξ(a)

= e(g,q

)

kακ

Next, the decrypting entity U uses its secret key

element sk

, related to t

to deduce the deciphering

key K

as follows:

= ˆe(E

1 j

,sk

) ˆe(g,h

)

)κ

ˆe(E

1 j

,sk

) = ˆe(g

(d1+d2)

−(k

)αγ

)

−1

)

This K

equation simpliﬁes to :

= ˆe(g,q

)

α(k

)

Finally, U recovers the sub-message d

, with re-

spect to related access level t

, by computing

3 j

where d

is d1 + d2 in this case. Thus,

d1 + d2 =

3 j

(d1 + d2).K

Where the aggregated message is recovered.

6 SECURITY ANALYSIS

In this section, we are considering two main cases

to demonstrate the behaviour of the framework under

different conditions.

Case 1: Unauthorized Drone Attempting De-

cryption

In this case, we are considering an unauthorised

drone without sufﬁcient attributes to decrypt a mes-

sage encrypted under the access policy. Let us as-

sume that the access policy threshold for decryption is

t = 3 with attributes S = {ID,Location,Destination}.

Let an unauthorised drone D have an attribute set

= {Location}, which does not satisfy the required

threshold policy t = 3. To prove case 1; First, the CA

generates a secret key for D based on the attribute set

. Which can produce a key term like :

sk = {g

γ+ξ(a)

}a ∈ A

Since D has only one attribute, it lacks the com-

plete key structure needed to satisfy the threshold pol-

icy. Now when trying to decrypt the ciphertext com-

ponent.

2 j

= q

kα

∏

a∈S

(γ+ξ(a))

∏

o∈O

n+t

−1−s

(γ+o)

It needs to compute a pairing term that aggregates

all the attributes in S to satisfy the threshold. With-

out the rest of the attributes, D cannot construct the

polynomial that is required for decryption. Thus, the

unauthorised drone attempts to decrypt the message

and fails which ensures that the framework is secure

against unauthorised access.

Case 2: Aggregation of Two Messages

In this case, when we are trying to add two en-

crypted messages, the decrypted result will yield the

sum of the original message. Let us consider two

messages M1 = 3 and M2 = 5 and are encrypted

as E1 and E2. Now when GCS is encrypting the

messages for the parent drone with the attribute S =

{ID, Location, Destination} and t = 3. The aggrega-

tion of M1 and M2 results as:

1 j

= E

11 j

12 j

= g

M1+M2

−(k

)αγ

2 j

= E

21 j

22 j

3 j

= E

31 j

32 j

= (M1 + M2).K

= ˆe(g,q

)

α(k

)

Now, for decrypting the message the decrypt al-

gorithm as explained in section 5.1 is called and exe-

cuted and results in a simpliﬁed equation as :

= ˆe(g,q

)

α(k

)

Now recovering the sum of the message M1 + M2

M1 + M2 =

3 j

(M1 + M2).K

Which results in the sum of encrypted messages.

When originally the two messages are added, the ex-

act result of the encryption as obtained above is re-

ceived. Thus, we prove the correctness of this case

that homomorphic addition enables data aggregation

for encrypted data without requiring intermediate de-

cryption.

Based on these cases the security analysis for the

framework based on the requirements and models as

mentioned in 4.3 and 4.2 is as follows:

Conﬁdentiality. The scheme guarantees conﬁden-

tiality based on the IND-CCA model. This means that

an adversary without the correct decryption key can-

not identify any differences between the encryptions

of two chosen messages. Even if the adversary selects

two distinct messages, the encrypted outputs remain

indistinguishable.

Proof of Conﬁdentiality:

RAHE: A Robust Attribute-Based Aggregate Scheme Enhanced with Homomorphic Encryption for 5G-Connected Delivery Drones

407

• Setup : Let (G

,G, ˆe) be bilinear groups of

prime order p and a bilinear map ˆe : G

×G

→ G.

Let α,γ, r

∈ Z

∗

be random values chosen by the

certiﬁcate authority (CA).

• Encryption : For a message D to be encrypted, we

generate a ciphertext E containing:

1 j

= g

−kαγ

· g

· z

2 j

= q

kα

∏

a∈S

(γ+ξ(a))

∏

o∈O

n+t

−1−s

(γ+o)

where S is the set of attributes required for decryp-

tion, and z is a generator of G

• IND-CCA : Suppose an adversary A chooses two

messages D

and D

. The encryption scheme cre-

ates ciphertexts E

for these messages with

random r

and r

= g

−kαγ

· g

· z

, E

= g

−kαγ

· g

· z

Without knowledge of k,α,γ, or r

, A cannot dis-

tinguish between E

and E

due to the random-

ization by r

and the hardness of the Decisional

Bilinear Difﬁe-Hellman (DBDH) assumption.

Thus, the security relies on the Decisional Bilinear

Difﬁe-Hellman (DBDH) assumption. The Cipher-

text components include a random blinding factor r

ensuring that without the correct secret key, decryp-

tion is computationally infeasible. Which ensures the

scheme is IND-CCA secure, proving conﬁdentiality.

Access Control. The framework enforces access

control by requiring nodes to meet threshold-based

attributes to decrypt speciﬁc parts or all of a mes-

sage. The scheme enforces access control by allowing

only entities with attributes that satisfy the predeﬁned

threshold to decrypt speciﬁc parts of the message.

• Setup : Deﬁne an attribute set S = {a

,. .., a

}

and an access policy (t

,S) where t

is the thresh-

old required for decryption.

• Threshold Policy : For a drone U with an at-

tribute set A

⊂ S, let |A

∩ S| = t

. To de-

crypt, U requires a secret key element sk =



γ+ξ(a)

: a ∈ A



and computes:

= ˆe

∏

a∈A

(γ+ξ(a))

2 j

= ˆe(g,q

)

kα

∏

a∈S\A

(γ+ξ(a))

∏

o∈O

n+t

−1−s

(γ+o)

• Threshold Satisfaction: If |A

∩ S| < t

, then U

cannot compute the product terms necessary for

decryption, as

∏

a∈S\A

(γ + ξ(a)) ̸= 0. Therefore,

U cannot obtain L

, preventing unauthorized ac-

cess.

This proves that only drones meeting the threshold t

can decrypt the message, enforcing access control.

Key Management Security. The scheme relies on

a centralized certiﬁcate authority (CA) for secure key

management, ensuring unique keys and authorization.

1. Key Generation: The CA generates the master key

msk = (g,α, γ) and public parameters pp during

the setup phase, including a secret k

for each

user U with attributes A

2. Unique Secret Key: For each user U , the secret

key sk is generated as:

sk =



γ+ξ(a)

: a ∈ A



The random k

guarantees unique keys for each

user. Thus, keys are user-speciﬁc and cannot be

reused across drones.

3. Key Update and Revocation: The CA can issue

updated k

values to rekey drones or revoke ac-

cess by not reissuing keys. This ensures that only

authorized drones retain access to their data.

This proves that the framework maintains secure key

management through centralized control, unique key

generation, and ﬂexible rekeying.

Based on the above analysis, the proposed framework

can defeat the main security attacks against drone

transportation systems. The use of encryption can

stop eavesdropping attacks, for example, if a ma-

licious drone tries to access the data exchanged be-

tween drones or between drones and GCS, it will not

be able to decrypt the data as it is encrypted using

ABE with homomorphic encryption. A Man in the

middle attack cannot be successful due to the use

of CBDA among drones to prevent malicious drones

from joining the chain which prevents other nodes

from hearing the encrypted message. In addition, we

use an attribute-based encryption scheme with homo-

morphic encryption scheme where the messages are

encrypted and only the legitimate user can decrypt it,

as the third person will not be able to access it as

their attributes do not satisfy to the access structure.

Finally, the proposed scheme helps to prevent data

injection attack as we use a homomorphic encryp-

tion scheme within the ABE scheme. When a mes-

sage is sent between drones, the data is encrypted,

and a merging of data is only happening among the

chain; no adversary is able to inject data into the com-

munication. As a chain is being formed ﬁrst, it pre-

vents unauthorised drones or devices from accessing

the chain and injecting data to it.

SECRYPT 2025 - 22nd International Conference on Security and Cryptography

408

7 PERFORMANCE ANALYSIS

The multi-level Attribute-Based Encryption (ABE)

scheme used in this framework is speciﬁcally de-

signed for efﬁciency, resulting in smaller ciphertexts.

The size of the ciphertext primarily depends on the

number of thresholds deﬁned in the multi-level ac-

cess policy. In the context of communication between

Ground Control Stations (GCS) and drones, where the

typical setup consists of just two threshold levels, the

size of the ciphertext can be considered effectively

constant. This feature simpliﬁes performance met-

rics, as both encryption and decryption overheads are

affected only by these limited and manageable thresh-

old levels.

Moreover, by employing a multi-level authoriza-

tion ABE scheme, this framework efﬁciently mini-

mizes the communication overhead between the GCS

and the drones. Instead of communicating directly

with multiple child drones, the GCS sends mission-

critical data only to the parent drone. This signiﬁ-

cantly reduces the communication load, as the parent

drone is responsible solely for relaying the encrypted

information to its connected child drones. In this

conﬁguration, the parent drone does not need to per-

form additional encryption; it simply forwards the en-

crypted data received from the GCS. Child drones can

selectively decrypt the relevant data based on their ac-

cess permissions, ensuring data conﬁdentiality with-

out imposing a computational burden on the parent

drone.

Without the multi-level ABE scheme, the parent

drone would need to decrypt the received message

from the GCS and re-encrypt data fragments for each

child drone. This process introduces signiﬁcant en-

cryption overhead and redundant operations, which

drastically reduce system efﬁciency. Thus, the en-

cryption scheme in our framework optimizes both

computational and communication costs as in table 1.

The Chain-Based Data Aggregation (CBDA)

method further enhances communication efﬁciency

by reducing interactions among drones as well as be-

tween drones and the GCS. By limiting communica-

tion to the parent drone, the volume of data exchanged

across the network is minimized, thereby enhancing

the overall efﬁciency of the communication system.

Additionally, incorporating homomorphic encryp-

tion enables efﬁcient data aggregation at the parent

drone. This allows the parent node to combine en-

crypted data from child drones without the need for

decryption, thereby avoiding the overhead associated

with traditional encryption schemes. This approach

ensures that the encryption and decryption load at

the parent node remains low, contributing to a more

streamlined and efﬁcient data aggregation process.

8 EXPERIMENTAL ANALYSIS

To evaluate the performance of the proposed scheme,

we implemented an experimental testbed using a

Raspberry Pi 4 Model B as a drone and a laptop run-

ning Ubuntu on a virtual machine (VM) as the Ground

Control Station (GCS). The Raspberry Pi acts as a

computational proxy for an autonomous drone, while

the laptop simulates the processing environment of a

ground control station in a real-world scenario.

The system utilizes a structured key distribution

process, where a Certiﬁcate Authority (CA) running

on the laptop generates the public key and the master

secret key. These keys are securely distributed to both

the drones and the GCS, ensuring that all participating

entities have the necessary cryptographic credentials

for secure encrypted data exchange and computation.

The experimental setup focuses on a drone-to-GCS

communication scenario, which is assessed under two

different conﬁgurations.

• Parent Drone to GCS: In this scenario, a single

Raspberry Pi serves as the parent drone, which

encrypts a message using the RAHE scheme be-

fore transmitting it to the Ground Control Station

(GCS). Upon receiving the encrypted message,

the GCS decrypts it based on the established ac-

cess policy. This scenario illustrates a basic im-

plementation of the scheme where the encryption

and decryption processes occur without any addi-

tional operations, such as data aggregation.

• Child Drone to GCS: In this scenario, we con-

sider a network of three Raspberry Pi devices,

with one acting as the parent drone and the other

two as child drones. The child drones encrypt and

send their messages to the parent drone. Instead

of forwarding each individual encrypted message,

the parent drone performs aggregation using our

RAHE scheme that combines ABE and HE in one

go. Therefore, the parent drone does not need

to re-encrypt the message but only aggregate re-

ceived encrypted fragments from each child drone

and forward it to the GCS. GCS, then decrypts

to retrieve the original data. This scenario as-

sesses the efﬁciency gains from integrating ABE

and HE, particularly in reducing redundant de-

cryption operations and optimizing the aggrega-

tion of encrypted data.

To analyze the performance of the proposed RAHE

scheme, we conducted experimental trials with 1 to 10

attributes, averaging each iteration over 10 runs. The

RAHE: A Robust Attribute-Based Aggregate Scheme Enhanced with Homomorphic Encryption for 5G-Connected Delivery Drones

409

Table 1: Computation and Storage Costs of Multi-Level Attribute-Based Encryption Scheme with Homomorphic Encryption.

Scheme Access Policy Multi-level Ciphertext Size E Computation Overhead U Computation Overhead

ABE with HE Threshold Yes 3n E

(2n + 1)+ nE

+ nEξ

+ 1)E

+ 4nξ

+ E(3n + 1)

E and U are the encrypting entity and the decrypting user. t, n, and m are the size of the threshold, the cardinal of the threshold universe and

the attribute universe, respectively. E

, E

, E represent exponentiation costs in G

, G

, while ξ

is the cost of a pairing operation.

below ﬁgures represent the analysis highlighting the

key ﬁndings regarding the delay, encryption and de-

cryption time, computational overhead and through-

put.

Delay: The graph in ﬁgure 2 compares the en-

cryption and decryption delays. Encryption delay in-

creases as attribute expansion introduces additional

computational overhead, exhibiting a polynomial-like

growth. Decryption delay also rises but remains sub-

stantially lower than encryption delay.

Figure 2.

Encryption vs. Decryption Time: The line graph

in ﬁgure 3 further highlights the encryption and de-

cryption times. Encryption time increases linearly as

the number of attributes increases. This demonstrates

that decryption is consistently faster, likely due to a

less complex computational process.

Figure 3.

Overhead: In Figure 4, we observe the computa-

tional overhead for both encryption and decryption.

The overhead for encryption increases as the number

of attributes rises, reaching a value of 0.82 with 10

attributes. In contrast, the decryption overhead grows

at a much slower rate, suggesting that the decryption

process is less complex.

Throughput: The ﬁnal graph in ﬁgure 5 illus-

trates the data processing capabilities of encryption

and decryption. Throughput decreases as the num-

Figure 4.

ber of attributes increases for both processes. Encryp-

tion throughput declines, while decryption throughput

shows a more gradual decrease. However, decryption

maintains a relatively higher throughput, as it requires

fewer computational steps than encryption, leading to

reduced complexity in access policy enforcement.

Figure 5.

Overall, these graphs indicate that encryption is

more computationally intensive, introducing higher

delays and lower throughput, whereas decryption is

signiﬁcantly faster and more efﬁcient. This insight

is crucial for optimizing the encryption-decryption

process in real-time or resource-constrained environ-

ments.

The experimental results conﬁrm that CP-ABE

with homomorphic encryption enhances secure en-

crypted data processing in drone networks. However,

the computational cost associated with encryption re-

mains a signiﬁcant factor. On the other hand, decryp-

tion and aggregation demonstrate more favourable

performance, making the scheme suitable for real-

time drone communication over 5G networks.

9 CONCLUSION

In this paper, we present a robust and efﬁcient se-

cure communication framework for 5G-connected de-

SECRYPT 2025 - 22nd International Conference on Security and Cryptography

410

livery drones. Our approach utilizes an Attribute-

Based Encryption (ABE) scheme integrated with Ho-

momorphic Encryption (HE) to enhance both data

security and computational efﬁciency. By incorpo-

rating HE into the ABE structure, our framework

enables secure data aggregation, and encrypted ac-

cess control without the need for repeated decryption

operations, thereby preserving the conﬁdentiality of

mission-critical information.

Furthermore, we introduce a hierarchical Chain-

Based Data Aggregation (CBDA) model that op-

timizes communication between drones and the

Ground Control Station (GCS). This model reduces

energy consumption and ensures scalability. Our

comprehensive approach addresses critical security

challenges in drone networks, such as access control,

data conﬁdentiality, and low-latency communication,

making it well-suited for real-time 5G-based delivery

applications. As, a future work will focus on other

techniques such as edge computing and other optimiz-

ing libraries to reduce the encryption and decryption

overheads.

REFERENCES

Alwateer, M. and Loke, S. W. (2020). Emerging drone ser-

vices: Challenges and societal issues. IEEE Technol-

ogy and Society Magazine, 39(3):47–51.

Belguith, S., Kaaniche, N., Jemai, A., Laurent, M., and

Attia, R. (2016). Pabac: a privacy preserving at-

tribute based framework for ﬁne grained access con-

trol in clouds. In SECRYPT 2016: 13th International

Conference on Security and Cryptography, volume 4,

pages 133–146. Scitepress.

Cheon, J. H., Han, K., Hong, S.-M., Kim, H. J., Kim, J.,

Kim, S., Seo, H., Shim, H., and Song, Y. (2018). To-

ward a secure drone system: Flying with real-time ho-

momorphic authenticated encryption. IEEE access,

6:24325–24339.

Feng, C., Yu, K., Bashir, A. K., Al-Otaibi, Y. D., Lu, Y.,

Chen, S., and Zhang, D. (2021). Efﬁcient and se-

cure data sharing for 5g ﬂying drones: A blockchain-

enabled approach. IEEE Network, 35(1):130–137.

Guvenc, I., Koohifar, F., Singh, S., Sichitiu, M. L., and Ma-

tolak, D. (2018). Detection, tracking, and interdiction

for amateur drones. IEEE Communications Magazine,

56(4):75–81.

Han, T., Ribeiro, I. d. L., Magaia, N., Preto, J., Segundo,

A. H. F. N., de Mac

edo, A. R. L., Muhammad, K.,

and de Albuquerque, V. H. C. (2021). Emerging drone

trends for blockchain-based 5g networks: Open issues

and future perspectives. IEEE Network, 35(1):38–43.

He, L., Gan, Y., and Yin, Y. (2025). Efﬁcient threshold

attribute-based signature scheme for unmanned aerial

vehicle (uav) networks. Electronics, 14(2):339.

Hu, S., Liu, L., Fang, L., Zhou, F., and Ye, R. (2019). A

novel energy-efﬁcient and privacy-preserving data ag-

gregation for wsns. IEEE Access, 8:802–813.

Ismael, H. M. et al. (2021). Authentication and encryption

drone communication by using hight lightweight algo-

rithm. Turkish Journal of Computer and Mathematics

Education (TURCOMAT), 12(11):5891–5908.

Kaaniche, N., Belguith, S., and Russello, G. (2018). Ema-

lab: Efﬁcient multi authorisation level attribute based

access control. In Network and System Security: 12th

International Conference, NSS 2018, Hong Kong,

China, August 27-29, 2018, Proceedings, pages 187–

201. Springer.

Kaushik, A. R., Jutla, C., and Ratha, N. (2025). Towards

building secure uav navigation with fhe-aware knowl-

edge distillation. In International Conference on Pat-

tern Recognition, pages 373–388. Springer.

Li, G., He, B., Wang, Z., Cheng, X., and Chen, J. (2022).

Blockchain-enhanced spatiotemporal data aggrega-

tion for uav-assisted wireless sensor networks. IEEE

Transactions on Industrial Informatics, 18(7):4520–

4530.

Loukil, F., Ghedira-Guegan, C., Boukadi, K., and Ben-

harkat, A.-N. (2021). Privacy-preserving iot data ag-

gregation based on blockchain and homomorphic en-

cryption. Sensors, 21(7):2452.

Menouar, H., Guvenc, I., Akkaya, K., Uluagac, A. S.,

Kadri, A., and Tuncer, A. (2017). Uav-enabled intelli-

gent transportation systems for the smart city: Appli-

cations and challenges. IEEE Communications Mag-

azine, 55(3):22–28.

Ossamah, A. (2020). Blockchain as a solution to drone cy-

bersecurity. In 2020 IEEE 6th World Forum on Inter-

net of Things (WF-IoT), pages 1–9. IEEE.

Othman, S. B., Bahattab, A. A., Trad, A., and Youssef, H.

(2015). Conﬁdentiality and integrity for data aggrega-

tion in wsn using homomorphic encryption. Wireless

Personal Communications, 80(2):867–889.

Shrestha, R. and Kim, S. (2019). Integration of iot with

blockchain and homomorphic encryption: Challeng-

ing issues and opportunities. In Advances in Comput-

ers, volume 115, pages 293–331. Elsevier.

Wan, H., David, M., and Derigent, W. (2020). Energy-

efﬁcient chain-based data gathering applied to com-

municating concrete. International Journal of Dis-

tributed Sensor Networks, 16(8):1550147720939028.

Wu, Y., Dai, H.-N., Wang, H., and Choo, K.-K. R. (2021).

Blockchain-based privacy preservation for 5g-enabled

drone communications. IEEE Network, 35(1):50–56.

Zhang, M., Lu, L., Wu, Y., Yan, Z., Sun, J., Lin, F., and

Ren, K. (2025). Droneaudioid: A lightweight acous-

tic ﬁngerprint-based drone authentication system for

secure drone delivery. IEEE Transactions on Infor-

mation Forensics and Security.

RAHE: A Robust Attribute-Based Aggregate Scheme Enhanced with Homomorphic Encryption for 5G-Connected Delivery Drones

411