Secure Authentication Solution for Cloud-based Smart City

Surveillance System

Yosra Ben Dhief, Yacine Djemaiel, Slim Rekhis and Noureddine Boudriga

Communication Networks and Security Research Lab, University of Carthage, Tunisia

Keywords: Smart City, Physical Unclonable Functions, Sensor Authentication, Cloud, Virtual Sensor.

Abstract: With the emergence of physical monitoring systems and their use for smart cities, new security concerns

arise due to the sensitive nature of the data produced by the sensor devices of these systems, which makes

the smart city applications a prime target for cyber attacks. However, securing these devices is very

challenging given the fact that they are resource-constrained, and deployed in physically unsecured

environments. In this paper, we propose a secure scheme for cloud-based smart city surveillance system

providing a lightweight sensor authentication protocol based on Physical Unclonable Functions and

securing the exchanged data through the different components of this infrastructure.

1 INTRODUCTION

With the emergence of smart cities over the world,

there is a great need for services that ensure the

protection of citizens and cities' assets. In fact, these

services run in a dynamic environment containing

sets of physical supervisory systems, which in turn

can be managed by different providers to serve

several actors. However, the inherent complexities

of such environments raise several challenges for the

implementation of smart city applications. These

challenges are mostly caused by the massive amount

of data that are generated by the sensors. The cloud

technology presents an attractive solution to face this

challenge since it provides powerful computing

resources and elastic storage capacities (Schleicher,

et al., 2016).

Despite the benefits of the cloud, the smart city

applications are still suffering from security threats

given the fact that the sensitive nature of sensor data

makes them a prime target for cyber attacks.

Therefore, implementing security solutions is very

crucial to ensure efficient and reliable services for

smart cities. In particular, authentication and

encryption services are required to provide both

device and data exchange security. However, the

sensor devices are endowed with limited processing,

memory, and energy resources which make them

unable to support complex security mechanisms

(Wallrabenstein, 2016). In this context, several

works considered Elliptic Curve Cryptography

(ECC) as an effective solution for resource-

constrained devices that requires lower

computational and storage capabilities. In (Kalra &

Sood, 2015), the authors presented an ECC based

mutual authentication protocol for secure

communication between embedded devices and

cloud servers. In (Hu, et al., 2017), the authors

proposed a secure cloud-based architecture for

health monitoring using IoT devices which achieves

authentication and data security using ECC digital

signature.

However, the aforecited schemes call for the

storage of the secret key inside the Non-Volatile

Memory (NVM) of devices, while the sensors can be

deployed in a hostile environment and be the target

of many physical attacks. To thwart these attacks,

the sensor devices can be equipped with tamper

resistant hardware. However, theses solutions can be

too expensive to be practical in many smart city

applications. Several researches were proposed (Lao,

et al., 2017), (Wallrabenstein, 2015), (Aman, et al.,

2017) and (Wallrabenstein, 2016) focusing on the

use of a low-cost solutions to protect sensor devices

from tampering. These solutions are based on the

use of a Physical Unclonable Function (PUF) which

consists in an on-demand extraction of secrets from

complex properties of hardware, rather than storing

them in a NVM. In fact, a PUF produces to a given

input (or challenge) an output (or response) which is

unique due to inter-chip variation that is difficult or

impossible to model. The challenge and response

502

Dhief, Y., Djemaiel, Y., Rekhis, S. and Boudriga, N.

Secure Authentication Solution for Cloud-based Smart City Surveillance System.

DOI: 10.5220/0006867005020507

In Proceedings of the 15th International Joint Conference on e-Business and Telecommunications (ICETE 2018) - Volume 2: SECRYPT, pages 502-507

ISBN: 978-989-758-319-3

pairs (CRPs) of a PUF are used to generate chip-

unique signatures for an authentication system (Lao,

et al., 2017).

In this work, we propose a security scheme for

cloud-based smart city surveillance architecture

leveraging an extension of the PUF-based

authentication system proposed in (Wallrabenstein,

2016). The architecture was presented by our work

in (Yosra, et al., 2018) that provides global

monitoring services () for smart cities built on

virtual sensors (), which are created by different

brokers. A  provides an indirect measurements by

processing data collected by physical sensors ()

deployed by different physical supervisory systems

(). The main goal of this paper is to design a

lightweight scheme involving more than two actors

(i.e., sensor devices, broker and provider) to perform

secure authentication and key management for

sensors in cloud-based environment, in addition to

guaranteeing sensors’ anonymity. The proposal

exploits, firstly, the ECC to enable the sensor device

to generate pair of asymmetric key and the PUF

hardware to secure the private key from tampering,

and then the Zero knowledge Proof (ZKP) to prove

the authenticity of the generated public key. The

generated asymmetric keys are used to encrypt and

sign the exchanged data.

The main contribution of this paper is four fold:

) A lightweight authentication scheme for sensor

devices that preserves their anonymity in cloud

environment; ) A low cost tamper resistance

solution for sensor devices based on the use of

unclonable function; ) The proof of the real

involvement of sensors in data collection; and )

The integrity and confidentiality of the exchanged

data in a cloud sensor based architecture.

The remaining part of the paper is organized as

follows. Section  discusses the security

requirements for cloud-based smart city surveillance

system. Section  details the proposed architecture

and the functions ensured by its components.

Section  presents our proposed authentication and

key distribution scheme. The next section provides a

security analysis and validation of the proposed

scheme. Section  concludes the paper.

2 SECURITY REQUREMENTS

FOR CLOUD-BASED SMART

CITY SURVEILLANCE

In this section, we detail the security requirement to

design sensor authentication scheme in cloud-based

smart city surveillance system.

Anonymous Sensor Authentication. The

virtualization enables GMS to make use of different

 without knowing their identity. However, it is

required to prove that the used data are collected by

a trusted sensor. Thus, sensor devices should be

authenticated anonymously.

Low Computational and Energy Operation. The

computation power and the battery capacity of

sensor are limited and may be insufficient for the

processing of security algorithms. Therefore, a

security scheme for sensor devices should involve

low resource and energy consumption.

Tamper Resistance. Given the fact that the sensor

devices can be deployed in remote unattended

places, security solution should be able to detect any

attempt to tamper with them.

Freshness and Integrity of Sensing Data. The

surveillance applications requires firstly, the

freshness of the used data to guarantee that it is very

much recent and no old data have been replayed.

Secondly, the integrity of the used data to guarantee

that it has not been altered in transit.

3 CLOUD-BASED SMART CITY

SURVEILLANCE SYSTEM

ARHITECTURE

The proposed cloud-based smart city surveillance

architecture is composed of three actors: ) Physical

Supervisory System () /provider ) A broker of

 and ) A Logical Supervisory System (). An

architectural map of the proposed architecture is

shown in Figure . In the following we detail the

role of each actor.

Physical Supervisory System / Provider

A typical  is a centralized system that ensures

real-time monitoring of an infrastructure process

using a set of heterogeneous sensors. The  acts

as a provider that allows the reuse of its  by

different external systems with different

configurations. It also interacts with a trusted

Authentication System () to enroll and to

authenticate its . The  is composed of

enrollment servers that ensure the authentication of

 and the generation of shared keys to be used to

secure the data exchange.

Secure Authentication Solution for Cloud-based Smart City Surveillance System

503

Broker of virtual sensors

Local

directory

Control

server

Local

sensing data

repository

Provisoning

server

Broker of virtual sensors

Logical Supervisory System

Proxy server

Front end

server

Template

repository

PSS/Provider

Control center

PSS/Provider

Controller

Services data

repository

Active services

database

Controller

Physical

sensors

Local

directory

Control

server

Local

sensing data

repository

Controller

Proxy Provisoning

server

Brokering

directory

Sensing data

repository

Brokering

database

Active VS

database

Provisoning

server

Proxy Provisoning

server

Brokering

directory

Sensing data

repository

Brokering

database

Active VS

database

Local

directory

Control

server

Local

sensing data

repository

Provisoning

server

PSS/Provider

Enrollment servers

Authentciation system

Figure 1: Cloud-based smart city surveillance architecture.

Broker of Virtual Sensors

The broker of  is the intermediary between the

/providers and the . According to the request

of , the broker selects providers to use their 

with the appropriate configurations. It creates a

enabling the processing of the data generated by

the selected . The broker also maintains an

archive of the sensed data collected from the

different PSSs.

Logical Supervisory System

The provides s for heterogeneous and

distributed infrastructures making the use of 

provided by different brokers of . The supervision

ensured by this  is based on the analysis and the

processing of data produced by a set of .

4 SECURE DATA EXCHANGE IN

CLOUD-BASED SMART CITY

SYSTEM

This section describes the proposed PUF-based

authentication and key distribution scheme for

cloud-based smart city surveillance, in addition to an

overview about the PUFs concept.

4.1 Physical Unclonable Function

Overview

The PUF performs a mapping from a challenge to a

unique response which depends on the unique

characteristics of the physical hardware on which it

is executed. Several authentication schemes have

been proposed based on the PUF (Aman, et al.,

2017). They use the fact that only a user and a

sensor device know a CRP. To authenticate a sensor,

the user has to tell the device in clear his challenge

so that it can get the response. Thus a man in the

middle can hear the challenge, get the response from

the PUF device and use it to spoof the PUF device.

The Controlled PUF (Dijk, et al., 2008) (CPUF),

which is a combination of a PUF and an Integrated

Circuit (IC) bound together such that an attacker has

no access to the communication between the PUF

and the IC. Any attempt to force them apart will

damage the PUF. The IC completely governs the

PUF input and output and reveals only indirect

information derived from the output.

4.2 PUF-based Authentication and Key

Distribution Scheme

The proposed scheme is based on ECC which is a

suitable solution for resource-constrained sensors.

We define an elliptic curve  over a field 



of

prime order , and a base point  of order , where

    . The proposed scheme is also based on

a CPUF implemented by the used sensor devices and

a hash function denoted by . We assume that the

 is a trusted entity that authenticates the actors of

SECRYPT 2018 - International Conference on Security and Cryptography

504

our architecture and that generates the secret keys to

be used during the data exchange. In addition, each

provider has a pair of keys 



 



 generated

by the  and to be used during the exchange of

data with sensors.

4.2.1 Enrollment Phase

This phase is executed in a secure environment by

the . The sensors are pre-configured with a virtual

identity 



. Once the enrollment phase is

finished, the device deletes 



from its memory.

Figure  shows the steps of this phase, which are

detailed in the following:

1. The  generates a Pre-challenge 



and sends

(



) to the sensor device.

2. The sensor selects rand from 



and calculates a

challenge     



 and  

     which is used as input to the

PUF to generate the response . Next, it

generates a public key    , where

the private key  is selected from 



 It applies an

Error Correction Code (ECC) over the private

key, and blinds this value with , in a way that

the final helper value      leaks

no information about the private key. The device

stores  and  in the NVM, and sends to the

the public key , the rand and its 



3. The ES saves the public key  and rand with the

correspondent 



and the identity of its

provider 



4.2.2 Sensor Authentication Phase

The device authentication is performed through a

ZKP protocol. In this phase, the verifies whether

the sensor can re-generate the private key involving

its PUF. Figure  shows the steps of this phase,

which are detailed in the following:

1. The  recovers the relevant public key , 



and  using 



. It generates a nonce 

and 



and calculates 



 and  









. Next, the ES

encrypts 



,



, ,  and  using a

secret key 



, which is shared between the ES

and the provider that owns the sensor device.

2. The PSS decrypts the received message with





and transfers to the relevant device 



, ,

 and .

3. The device reads from its memory  and . It

calculates 



      and verifies

the received . This enables the sensor device

to authenticate the , given that it is the only

entity that knows the 



and . Then, the

device calculates        and gets

its correspondent response  from its PUF. It

re-generates its private key  using an error

decoding,      It also calculates

its public key    .

Figure 2: Enrollment phase.

4. After re-generating the asymmetric key pair, the

device selects  from 



and calculates  

 . Then, It calculates  

    and      . It also

calculates 



 



 and

generates 



. Finally, it sends 



 



   to the

/Provider.

5. The  adds its identity 



and the 



the received message and encrypts them with





. It adds to the encrypted message in clear

its 



and sends them to the .

6. Using 



, the  retrieves 



and decrypts

the message. Then, it verifies 



and calculates

     ) and     

 . The  compares  and . If they

are equal, it sends an acknowledge to the

provider. Else, it refuses to authenticate the

device.

Figure 3: Device authentication phase.

Secure Authentication Solution for Cloud-based Smart City Surveillance System

505

Once a sensor is authenticated, the parameters

used for sensor authentication can be renewed.

4.2.3 Data Transfer Phase

The data collected by a sensor are transferred to the

broker without revealing the sensor identity.

However, the broker should make sure that the

received data are collected from an involved and an

authentic sensor. To this end, we proposed a

protocol that enables a broker to authenticate a

sensor through a digital signature without revealing

its identity. The sensor signs data with its private

key  that it re-generates using its PUF. In the other

hand, the broker verifies the sensor's signature using

a public key  provided by the . The validity of

the key pair ( ) is verified by the ES through the

sensor authentication described above.

The proposed signature protocol is based on the

ElGamal digital signature algorithm described by

(Wallrabenstein, 2016). Figure  shows the steps of

this phase, which are detailed in the following:

1. The device reads from its memory  and 

and re-generates its private key  and public 

as described at step  in the authentication phase.

It selects a random curve point from





  



  



 . Next, it sets

   



   



  where 



denotes

the  coordinate of the point . It also sets

  



    where

 is a sequence number which is initially

synchronized between the broker and the device.

The device creates the message 



starting from

  and data. It encypts the message 



and a

random value 



using the public key 



of the

provider. It adds to the encrypted message 



clear and sends them to its provider.

2. The provider decrypts the message its 



 and

adds  and its identity 



to the decrypted

message. Then, it encrypts them with the secret

key 



shared with the concerned broker. It

also adds in clear its identity 



. Finally, it

sends them to the broker.

3. Using 



, it selects the appropriate 



and

decrypts the message. Then, it retrieves the

public key  using 



and verifies the

signature by calculating  

 











 



 





5 SECURITY ANALYSIS AND

SYSTEM VALIDATION

In this section, we will analyze and validate the

robustness of the proposed scheme against a set of

attacks.

 Low Cost Tamper Resistant. The device uses

 and  with its PUF to regenerate its private

key  instead of stored it in the NVM. Hence, 

is physically obfuscated and only exists in

memory when needed for a cryptographic

operation. Besides, we use a CPUF that makes

an inseparable link between an IC and a PUF’s

input-output mapping, an attacker that attempts

to probe the circuit will irreversibly modify the

physical variations in the IC, which in turn

changes the PUF mapping and prevents

regeneration of the private key. Thus, PUF

circuits are a low-cost solution for tamper

resistance to resource-constrained devices.

Figure 4: PUF-based signature protocol.

 Sensor Anonymous Authentication. The

broker authenticates a sensor device through the

verification of its signature using the public key

 sent by the .  is generated based on a

private key which is only known by the sensor

device given that the regeneration of  requires

the involvement of its unique function PUF.

Based on this fact, the  authenticates the

sensor using ZKP protocol. Besides, the sensor

device is identified by a virtual identity 



which is only known by the provider and the

enrollment server. Thus, the broker

authenticates the sensor device without knowing

its identity and its physical specifications.

 Resistance to Replay Attack. An attacker can

replay previous messages exchanged by the

sensor device to the broker. But, the broker can

detect the invalidity of the message because

messages are signed using the private key  that

SECRYPT 2018 - International Conference on Security and Cryptography

506

can be generated only by e sensor using its

PUF and contain new sequence number

generated for new message. Besides, to prevent

replay attacks on the message exchanged

between actors, new nonces are generated to

guarantee the freshness of each session.

 Resistance to Impersonation Attack. The

proposed scheme implements ZKP protocol to

authenticate a sensor device which allows the

 to verify that the sensor knows the private

key  without disclosing it. A sensor can prove

that it knows the correct  by re-generating it

involving its implemented PUF. Given that the

PUF is unique for each device, an adversary

cannot re-generate  to impersonate the  .

 Resistance to Man In the Middle Attack. an

attacker cannot perform a MITM attack in the

communication between the and a sensor

because at each time each one checks that the

other knows the  and the  

 used to generate the challenge 

saved by the sensor. An attacker is not also able

to decrypt several messages because they are

encrypted by a secret key already calculated by

the . On top of that, our scheme is based on

CPUF which prevents an attacker, even though

he determined the challenge saved by the

sensor, to probe the device and get the response.

6 CONCLUSION

We proposed in this paper a secure authentication

scheme for cloud-based smart city surveillance

system. The proposal solution exploits, firstly, ECC

to enable the sensor device to generate a pair of

asymmetric key and the PUF hardware to secure the

private key from tampering, and then ZKP to prove

the authenticity of the generated public key.

Moreover, the generated asymmetric keys are

leveraged to provide encryption and signature

mechanism to protect exchanged data with sensors

while coping with their resources-limitations. In

addition, our proposed solution enables to

authenticate anonymously a sensor in the cloud

environment. In a future work, we will extend our

proposed scheme to support the authentication of

virtual sensors created starting from a dynamic set of

mobile and heterogeneous sensor devices.

REFERENCES

Aman, M. N., Chua, K. C. & Sikdar, B., 2017. Secure

Data Provenance for the Internet of Things. In

Proceedings of the 3rd ACM International Workshop

on IoT Privacy, Trust, and Security, IoTPTS'17, pp.

11-14, Abu Dhabi, United Arab Emirates.

Dijk, B. G. a. M. V. et al., 2008. Controlled Physical

Random Functions and Applications. ACM

Transactions on Information and System Security

(TISSEC), 10(4), pp. 3:1--3:22.

Dijk, B. G. a. M. V. et al., 2008. Controlled Physical

Random Functions and Applications. ACM

Transactions on Information and System Security

(TISSEC), 10(4).

Hofer, C. B. a. M., 2012. Physical Unclonable Functions

in Theory and Practice. 1 ed. New York, USA:

Springer Publishing Company.

Hu, J.-X., Chen, C.-L., Fan, C.-L. & Wang, K.-h., 2017.

An Intelligent and Secure Health Monitoring Scheme

Using IoT Sensor Based on Cloud Computing.

Sensors, Volume 2017, p. 11.

Kalra, S. & Sood, S. K., 2015. Secure authentication

scheme for IoT and cloud servers. Pervasive and

Mobile Computing, 24(C), pp. 210 - 223.

Lao, Y., Yuan, B., Kim, C. H. & Parhi, K. K., 2017.

Reliable PUF-Based Local Authentication With Self-

Correction. IEEE Transactions on Computer-Aided

Design of Integrated Circuits and Systems, 36(2), pp.

201-213.

Schleicher, M. V. a. J. M., Inzinger, C., Dustdar, S. &

Ranjan, R., 2016. Migrating Smart City Applications

to the Cloud. IEEE Cloud Computing, 3(2), pp. 72-79.

Suárez-Albela, M., Fernández-Caramés, T. M., Fraga-

Lamas, P. & Castedo, L., 2017. A Practical Evaluation

of a High-Security Energy-Efficient Gateway for IoT

Fog Computing Applications. Sensors, 17(9).

Wallrabenstein, J. R., 2015. Implementing Authentication

Systems Based on Physical Unclonable Functions. In

Proceedings of the 14th IEEE International

Conference On Trust, Security And Privacy In

Computing And Communications, TrustCom-15, pp.

790-796, Helsinki, Finland.

Wallrabenstein, J. R., 2016. Practical and Secure IoT

Device Authentication Using Physical Unclonable

Functions. In Proceedings of the IEEE 4th

International Conference on Future Internet of Things

and Cloud (FiCloud 2016), pp. 99-106, Vienna,

Austria.

Yosra, B. D., Yacine, D., Slim, R. & Noureddine, B.,

2018. Cloud-based Global Monitoring System for

Smart Cities. In Proccedings of the 32nd IEEE

International Conference on Advanced Information

Networking and Applications Workshops, WAINA

2018, Cracow, Poland.

Secure Authentication Solution for Cloud-based Smart City Surveillance System

507