Achieving Privacy, Security, and Interoperability among Biometric

Networks using Symmetric Encryption

Eduardo M. Lacerda Filho and Vinicius P. Gonc¸alves

Department of Electrical Engineering, University of Brasilia, Brazil

Keywords:

Privacy, Encryption, Biometric Security, Communication Protocol, Interoperability.

Abstract:

Privacy, security, and interoperability of biometrics systems are fundamental for any segment of a society that

uses it. In this work, we developed a network that uses a symmetric encryption scheme, to ensure the anony-

mous index data exchange and registration of a person, and an interoperability communication protocol to

process identiﬁcation requests between different biometric systems. Our main contribution is the construction

of a non-reversible encryption index that can safely traverse, without decrypting it, the network of connections

between different biometric systems with an interoperability and data integrity communication protocol. The

advantages of our work are the mitigation of known encryption and network attacks, the creation of a random

initialization vector, without sending it over the network, but feasible to be calculated for all the accredited

Biometric Service Providers, the increased security of biometric database, that not only relies about templates,

and the improvement of IEEE Biometric Open Protocol Standard. The security analysis of the scheme and the

results conﬁrm that the network holds anonymity of a person and that it is possible to interoperate this data

with an enhanced integrity protocol.

1 INTRODUCTION

Biometric Service Providers (BSP) are used in vari-

ous segments of society (Jain et al., 2016). For many

years the typical approach for the security and privacy

of a biometric system has been template protection

(Jain et al., 2016; Ngo et al., 2015; Campisi, 2013).

When it was proof that a vector minutiae representa-

tion of a ﬁngerprint could be retrieved (Ross et al.,

2007), many works have been developed to deal with

the security and privacy of biometric networks.

Enhanced schemes have been created. They are

based on cancelable biometric (feature transforma-

tions) (Kaur and Khanna, 2019) and biometric cryp-

tosystems (Toli and Preneel, 2018; Kumar and Ku-

mar, 2016; Li et al., 2015; Nasir and Perumal, 2013).

Those efforts over the years have some improve-

ments, but also some problems, which include link-

age attacks, brute force attacks, side-channel attacks,

the problem of level privacy against False Acceptance

Rate (FAR), and others (Natgunanathan et al., 2016;

Hirano et al., 2016; Quan et al., 2008; Kocher, 1996).

Furthermore, it was not shown how to use each of

these schemes on different biometric databases and

make them communicate with reasonable security ev-

idence.

From this exposition, one issue arises. How to en-

sure that the identity of one person is not revealed but

can be processed and used for different systems in-

teroperation, preserving privacy? This work resolves

this problem with a novel approach that does not fo-

cus on the biometric template. We build a scheme that

anonymizes the records in the BSP databases, enhanc-

ing security, holding privacy, and a set of communi-

cations techniques for interoperability and integrity.

The contribution of this paper can be summarized

as follows:

• A scheme which produces an encrypted register

based on AES-CBC algorithm (Dworkin, 2001),

with 256-bit symmetric secret key k, a random

and locally calculated initialization vector iv, and

on the one-way functions SHA-2 (Dang, 2015),

for checksum, calculated into Hardware Security

Module (HSM) (Wenqian Yu et al., 2016), embed-

ded in a safe and audit environment;

• A communication protocol based on HTTPS

(Rescorla, 2000), JSON (Bray, 2017), and

ANSI/NIST (Mangold, 2016) packages, that im-

proves the IEEE BOPS (Biometric Open Protocol

Standard) (IEEE, 2019) method.

The cryptographic scheme proposed turns a record

based on a unique biographical identiﬁcation of a per-

son, such as the social number SN into an IDN anony-

mous register. Anonymous register means that no bio-

graphical index data is stored or exchanged, not either

Filho, E. and Gonçalves, V.

Achieving Privacy, Security, and Interoperability among Biometric Networks using Symmetric Encryption.

DOI: 10.5220/0008961504810489

In Proceedings of the 6th International Conference on Information Systems Security and Privacy (ICISSP 2020), pages 481-489

ISBN: 978-989-758-399-5; ISSN: 2184-4356

481

the biometric raw images. So, one IDN represents

only one owner associated with SN. All the BSP in

the network can calculate locally the same IDN string

representing one person of that biometrics unequivo-

cally, without exchange iv. The goals of this scheme

are: 1. guarantee that it is impossible to use known

attacks to link IDN to the biographical SN identity; 2.

security and privacy of a person in the network it is

not maintained only by using biometric templates; 3.

usage of an anonymous register to exchange biomet-

ric information of the same person among different

systems, without revealing who the person is.

The communication protocol uses HTTPS and

JSON protocols, and ANSI/NIST packages. It is pos-

sible to achieve interoperability and integrity between

the biometric systems that handle the encrypted IDN

without decrypting it. As an improvement to the IEEE

BOPS, we implemented one ﬂow of identiﬁcation

(1:n) of a record, with biometric base integrity assur-

ance procedures. The goal of this protocol is that any

biometric network can use this method to address any

request for identiﬁcation without tampering with the

recognition technology. A network is created where

different system technologies have a hub and direc-

tory services to authenticate and fulﬁll the requests

for identiﬁcation between biometric systems.

This paper is divided into the following parts. In

Section 2, we will discuss the related works, high-

lighting the technical issues involved. We introduce

the proposed scheme of our contributions, in section

3. In Section 4 and 5, the security analysis and the

results of a running instance of the proposed frame-

work evaluation are reported, respectively. Finally, in

Section 6, we will conclude indicating future works.

2 RELATED WORK

2.1 Privacy and Security Concerns

Achieving data privacy and security is a challenge.

The Dwork’s differential privacy work (Dwork, 2006)

shows that there is much auxiliary information that

an adversary (A ) can obtain without accessing the

database. Therefore, it is essential to delimit what

a method intends to establish security and privacy

(Campisi, 2013). For biometric systems security and

privacy, many cryptography techniques are used.

The work of Nassir and Perumal (Nasir and Peru-

mal, 2013) uses symmetric and RSA algorithms. The

work encrypts and signs the user ID and password

with the biometric data extracted into one package.

Some performance evaluation is done without secu-

rity analysis.

Li et al. (Li et al., 2015) describe a new security

analysis, proposing a multibiometric construction. By

combining information-theory and security, the work

uses features extraction and two levels of encryption,

one with hash functions and fuzzy vaults to bind the

transformed ﬁngerprint template, and the other use

Shamir’s secret sharing scheme to split and store the

hash values. A decision-level fused obtains the iden-

tity of a sample.

The paper from Kumar and Kumar (Kumar and

Kumar, 2016) proposed a multimodal biometric cryp-

tosystem based on feature-mode and decision-mode.

The construction consists of three phases, i.e., a Bose

Chaudhuri Hocquenghem (BCH) applied in the bio-

metrics, creating parity-code, a locking stage hash

code computation performed on the biometric modal-

ities, and an unlock stage where the parity-code is re-

generated using XORCoding. Experimental analysis

conﬁrms the superiority of multimodal cryptosystems

and decision-level fusion.

The Toli and Preneel (Toli and Preneel, 2018)

work uses a pseudo-identity authentication recorder

of a bank’s client. With a client’s PIN code, the pack-

age is encrypted and stored in the device, being dis-

carded the biometrics and the PIN. For secure require-

ments, the proposed use the ISO biometric, ﬁnancial,

and cryptographic device standards.

Kaur and Khanna (Kaur and Khanna, 2019) pro-

posed a random distance method. Considering mul-

timodal cancelable biometric template approach, it

generates a discriminative and privacy-preserving re-

vocable pseudo-biometric identities. The security

analysis shows some resistant to some attacks.

2.2 Interoperability Network

Interoperability between biometric data is discussed

in Tolosana et al. and Mason et al. (Tolosana et al.,

2015; Mason et al., 2014). However, regarding sys-

tem interoperability, the ANSI/NIST proposal (Man-

gold, 2016) and IEEE BOPS (IEEE, 2019) should be

considered relevant references. For this paper, we

will depart from the ANSI/NIST package and will im-

prove the IEEE BOPS method.

The IEEE BOPS is a standard that enables interop-

erability independent of the underlying system. The

proposed architecture is built using neutral languages.

It is based on a client/server authentication with soft-

ware running on a mobile system. There are some

formats, including JSON requests and responses, that

addresses the interoperability among those devices.

The fact that the IEEE BOPS mechanism is not

complete about checking communication integrity is

a problem. As IEEE BOPS, we will show that our sys-

ICISSP 2020 - 6th International Conference on Information Systems Security and Privacy

482

tem has better security characteristics, and a complete

workﬂow to address different biometric systems. That

includes dealing with a time-out or bad connection

among systems, check the status for the identiﬁcation

of biometric procedures, and message acknowledg-

ments between the systems.

3 THE PROPOSED SCHEME

In this section, we describe our work. We will explain

the IDN encrypted scheme, the biometric operating

network running two different technologies, and the

communication protocol. This network is currently

operational in a real government system.

3.1 IDN: The Encrypted Index Register

for Biometric Databases

The IDN generation scheme uses the SN, the k stored

in a non-exportable HSM slot (|k| = 256 bits), the

AES-256 algorithm in CBC mode, with a random,

and locally calculated iv, and the SHA256 hash func-

tions, for checksum, as follows:

• The SN is expanded by concatenation by itself un-

til it gets a 256-bit length; we concatenate k and

SN expanded, resulting 512-bit string (y); then, we

get x = SHA256(y), a 256-bit string; after, we di-

vide x in two halves, letting A be the ﬁrst 128 bits

and B the trailing ones; and ﬁnally iv = A⊕B;

• The SN padded with 128 bits is AES-256-CBC

encrypted using k, feasible to be locally calcu-

lated for the BSP, as the iv is not transmitted over

the network. This yields z, an encrypted 128-bit

string;

• Then, SHA256(z) is calculated and concatenated

with SHA256(SHA256(z)), for checksum; the

previous result is BASE64-encoded (Josefsson,

2006) to generate the IDN.

Any BSP, with k, can reach the same IDN for

only one SN, but the IDN becomes irreversible as

soon it leaves the HSM. The premise of using the

non-exportable key attribute, within HSM leased in

an audit secure environment, makes it easy to iden-

tify any misuse or key compromised. There is also

internal protection of the information, which beneﬁts

segments that have issues in sharing their client’s in-

formation.

3.1.1 Export and Import the Secret Key

For this work, we generate a 256-bit random symmet-

ric secret key k into a ofﬂine HSM. The exporting of

k is done using an OpenSSL library

. We export k

by wrapping it with the public keys of the BSP HSM

in the network, producing one cryptographic enve-

lope per each HSM, containing k. In this way, only

the HSM that owns the respective private key can un-

wrap the envelope. A local audit ceremony imports

the secret key into the HSM with the “no export” fea-

ture that guarantees it cannot be taken or copied from

the slot anymore, just used. For this purpose we use

the public key encryption RSA-2048-OAEP padding

(Moriarty et al., 2016; Bellare and Rogaway, 1995) to

export k.

3.2 The Biometric Network

The ﬁrst step of Figure 1 is the enrollment process,

which uses ﬁngerprint and face sensors. The enroll-

ment process prepares an ANSI/NIST Package 1, with

a secure proprietary biometric template and the bio-

graphical information of the person. The Package 1 is

sent through a mutually authenticated TLS/SSL chan-

nel for the system Network 1.

At the system Network 1, in the client/server in-

terface, the biographical SN is replaced into the HSM

by the encrypted index IDN. An ANSI/NIST Package

2 is built and includes a Transaction Code Number

(TCN) based on Universal Time Coordinated (UTC).

In the core of the system Network 1 (Bio API and En-

gine) begins the local identiﬁcation process. In the di-

rectory service, it is checked if the IDN is already in

the base, without decrypting it. If IDN exists, the Bio

API and Engine checks if there is only one owner or

other. If it is only one, it performs the 1:1 veriﬁcation

process (if positive, ends the process with a ”veriﬁ-

cation ok”; if not, performs the following); if not the

only owner, it must inform the enrollment processes

that there is something to be treated, possibly a fraud.

The Bio API and Engine initiates the identiﬁcation

1:n process in case no IDN found. If there is some ex-

ception (biometric found associated with other IDN),

it reports for the enrollment process that there is some-

thing to be treated, possibly a fraud. If not, the

HUB service prepares a JSON message, attaching the

encrypted ANSI/NIST IDE packet with the BSP re-

ceiver’s public key and the IDN. The ﬁngerprint and

face images are not a proprietary biometric template

anymore, but a wrap RSA-OAEP-2048-bit encrypted

bits that can only be opened by the receiving BSP.

Through a mutually authenticated SSL/TLS channel

between the BSP, this encrypted packet is sent to Net-

work 2, which initiates the same local biometric iden-

tiﬁcation 1:n. If some irregularity is found, the re-

ceiving BSP sends back a JSON message with an

https://www.openssl.org/

Achieving Privacy, Security, and Interoperability among Biometric Networks using Symmetric Encryption

483

Figure 1: Biometric Service Providers Network Diagram.

ANSI/NIST VRE with M value, informing that there

is something to be veriﬁed, possibly a fraud. If not, a

JSON message with an ANSI/NIST VRE with X value

is sent, informing that everything is “ok” and the IDN

can be stored.

Shortly after receiving the “ok” information, the

issuing BSP sends an acknowledgment JSON mes-

sage change_status to the network, informing that

the IDN can be stored. Also, the network have the

pending_operation mechanism, that checks if one

BSP has any process to be made, but for somehow

was not able to performer it on-line.

3.3 The Communication Protocol

3.3.1 Network and HTTPS Messages

Requests for HUB services must follow the asyn-

chronous pattern. All responses must be returned by

the HUB that received the request when it has the

available information. The requests must use the POST

method, containing only the ANSI/NIST ﬁle in the re-

quest body.

Requests for directory services must follow the

synchronous pattern. All responses must be returned

in the same request/response. The requests must use

the POST method.

3.3.2 JSON Messages Standards and Formats

The JSON messages will be described.

• JSON IDN query operation: Checks if IDN code

is registered in the network. The response for

the registered IDN is a JSON package, ﬁgur-

ing out which ﬁngerprints and face are registered

(TRUE|FALSE), along with the related IDN.

• JSON Pending operations listing operation: It re-

veals, in the event of any contingency with a bio-

metric system (time-out, bad connection or main-

tenance), a list of IDN that requires further pro-

cessing. Hourly, a JSON type is sent to ensure

that all processes have been executed, holding in-

tegrity through the network.

• JSON Change status notiﬁcation operation: This

operation notiﬁes BSP if a record, with its IDN,

was completed or there was some error.

3.4 Benchmarking

There are some works with security evidence and

anonymity of the index register into a biometric

database and network. As we showed, using cryp-

tographic techniques along with biometrics systems

is not new, but up to our best knowledge, shown in

Table 1, it had not been used to guarantee privacy

with security evidence and interoperability to the data.

Our main contribution is the construction of a non-

reversible cryptographic index that can safely traverse

the network of connections between biometric sys-

tems.

ICISSP 2020 - 6th International Conference on Information Systems Security and Privacy

484

Table 1: Benchmarking.

Works Security and

privacy bio-

metric data

approach

Security and

privacy bio-

graphical data

approach

Security evidence

against known

database or network

attacks

Interoperability

(Nasir and Perumal, 2013) YES YES NO NO

(Li et al., 2015) YES NO YES NO

(Kumar and Kumar, 2016) YES NO YES NO

(Toli and Preneel, 2018) YES YES NO NO

(Kaur and Khanna, 2019) YES NO YES NO

Our proposed scheme YES YES YES YES

4 SECURITY ANALYSIS

This section is divided into three areas of security

analysis (Katz and Lindell, 2007). The ﬁrst and the

second are focused on cryptoanalysis, mainly on the

randomness of the secret key and semantic security.

The third one is based on the operational security of

the network.

4.1 Randomness of the Secret Key

For the ﬁrst proof that the proposed scheme is secure,

we tested the randomness of keys that are generated

by HSM of the third reliable ofﬂine party. We use

the NIST Special Publication 800-90B - Recommen-

dation for Random Number Generation Using Deter-

ministic Random Bit Generators (Turan et al., 2018).

We compiled the “make iid” and “make non iid” tests

using the “libdivsufsort-dev / libbz2-dev” dependen-

cies, with a Ubuntu 18.04 operation system.

The results are:

NIST IID test

./ea_iid -i keys.bin

Calculating baseline statistics...

H_original: 7.886548

H_bitstring: 0.998301

min(H_original, 8 X H_bitstring):

7.886548

** Passed chi square tests

** Passed length of longest

repeated substring test

Beginning initial tests...

Beginning permutation tests... these

may take some time

** Passed IID permutation tests

NIST Non-IID test

./ea_non_iid -i keys.bin

Running non-IID tests...

Running Most Common Value Estimate...

Running Entropic Statistic Estimates

(bit strings only)...

Running Tuple Estimates...

Running Predictor Estimates...

H_original: 7.718814

H_bitstring: 0.932005

min(H_original, 8 X H_bitstring): 7.456043

The result demonstrates that the ofﬁcial NIST test

suite approves the randomness of the keys (k) that are

generated from the proposed scheme.

4.2 Semantic Security

Challenge: The BSP must ﬁnd a secure way to gen-

erate the same IDN for the same SN.

Supposing that we do not have any random input

of the block entrance, the birthday attack can be used

against non-Feistel ciphers (Schramm et al., 2004).

The encryption could be broken with not too much

effort (e.g., exhaustion of fewer than 65,000 tests -

15-bit exhaustion - would lead to its probability of

inferring any SN from the cipher to very high values).

To proof that the IDN scheme is secure enough

for any A , we must start by explaining the calcu-

lated iv created. First, we concatenate, for this re-

search, the 88-bit SN string, which have 4-bit entropy

for each octet block, until 256-bit length, with the

256-bit k, resulting in a 512-bit length. We use the en-

tropy of SHA-256 to one way 256-bit string, leading

a log

((1 − 1/e) ∗ 2

256

) outputs (Bellare and Kohno,

2004). A bitwise XOR-ed is used with equal parts of

the SHA-256 result, leading a 128-bit random iv. Be-

cause A cannot have control of the input bits calcu-

lated on the iv, the computational cost effort to ﬁnd a

collision is around 2

n/2

(Dobraunig et al., 2015), n =

output bits, leading to unfeasible known polynomial-

time attacks between the SN to IDN or IDN to SN.

Our 128-bit padding SN plaintext, it is XOR-ed

with a 128-bit random iv. Instead of rebooting the

encrypted AES-256-CBC with the previous outcome

and an initialization public vector (IV), we recalcu-

late the block entrance of the AES-256-CBC encryp-

Achieving Privacy, Security, and Interoperability among Biometric Networks using Symmetric Encryption

485

tion with a 128-bit random and local iv, derived from

known parameters (K and SN) only for the BSP. This

leads to an entropy effort of 256 + log

(1e9) bits.

These random bits scattering on the input makes the

system unfeasible to known polynomial-time attacks,

holding anonymity.

There are others attacks in the literature that are

avoided:

• Padding Oracle Attacks (Kang et al., 2016): all

the SN padded block is XOR with 128-bit string,

local and random iv created, avoiding this attack;

• Chosen Ciphertext Attacks (Rogaway, 2011): the

proposed iv is not sent over the network, it works

only within the cryptography module of the HSM.

So, A(iv ⊕t,AES(SN)) cannot be enforcement;

• Chosen Plaintext Attacks (Rogaway, 2011): the

iv can not be predictable by A , i.e., the proposed

AES-256-CBC cannot be attacked in polynomial-

time, as shown. Because the calculations are done

into accredited HSM, this also avoids adaptive-

CPA (Ding et al., 2019).

• Timing Attacks (Kocher, 1996): the HSM of

the BSP implements cryptographic calculations in

a constant-time, accredited with FIPS test suite

(Schaffer, 2019).

For exporting k from the ofﬂine HSM, we use

RSA-2048-OAEP-wrap operation. Each BSP sends

its own HSM public key for these operations. By

the known literature, this RSA-2048-OAEP calcu-

lus has IND-CPA-security and IND-CCA1/CCA2-

security (Boldyreva and Fischlin, 2006), which makes

the calculation semantically secure. Using index cal-

culus NFS (Number Field Sieve), the published litera-

ture describes that the cost to break RSA-2048 is 2

112

(Bernstein and Lange, 2014), which turns out to be

unfeasible in polynomial-time for A . We use the same

wrap operation to the ANSI/NIST encrypted package

between BSP, for additional security.

4.3 Operation Security

After the cryptoanalysis, we face the attacks that

could be done in the network. It is essential to state

that the wrapped k is imported in a local ceremony at

the security environment of the BSP.

Accredited HSM (Schaffer, 2019) does not allow

any copy or misuse of the non-exportable k. Other se-

curity features of the HSM embedded in the network

are IDS (Intrusion Detection System), non-physical,

mechanical, chemical violability, and SQL injection

protection (Wenqian Yu et al., 2016). Within the bio-

metric network, we use a TLS/SSL (RSA-2048-bit)

mutually authenticated for all communication. A re-

liable entity informs for each BSP the certiﬁcates and

URL end-points. Further, dedicated ﬁrewalls are only

from the IPs of each BSP, also audit by a trusted re-

liable party. Each biometric ﬁle has the origin and

destination name of the BSP given on the certiﬁcate.

Thereby, Distributed DoS attacks are mitigated in

this network (Yan et al., 2016). In addition to every

network part mentioned, BSP set one random session

key per transaction and also have a timestamp for each

ﬁle sent. This scenario mitigates any replays attacks

(Ding et al., 2018).

5 RESULTS

The results were obtained using data acquired into the

operational BSP network. First, we show the IDN

scheme created, indicating the calculations needed to

generate an IDN. Second, we demonstrate the com-

munication protocol logs, working for an identiﬁca-

tion purpose (“IDE”) between two distinct networks

in a time interval.

5.1 The IDN Scheme

The IDN scheme:

Algorithm 1: IDN algorithm.

Data: SN, K

Result: IDN

SNEXT = “$SNhex$SNhex$SNhex

y = “$K$SNEXT

H(y) = x

x ∈ {0,1}

256

A ∈ (x

,...,x

127

)

B ∈ (x

128

129

,...,x

255

)

iv = A ⊕ B

DataHex = ${#SNhex}

BlockPadded = $(((32 − $DataHex)/2))

blockSN = “$SNhex

tmp = $(print f

%02x

$BlockPadded)

for ((i=0; i < $BlockPadded; i++ )) do

blockSN = “${blockSN}${tmp}”

end

AES-256-CBC(blockSN, K, iv) = ID

H(ID) = IDbase

H(IDbase) = IDCheck

IDNhex = ($IDbase$IDCheck)

IDN = Encode64(IDNhex)

Table 2 presents the IDN calculations, from Algo-

rithm 1. Those are made, step by step, buy using a

SN and k, generated from the HSM of the entity, for

experimental purpose. It is possible to conclude from

the calculations that only with k and SN it is possible

ICISSP 2020 - 6th International Conference on Information Systems Security and Privacy

486

Table 2: IDN calculations.

Parameter Values and Results

K ffd62502b336b5c5784813e310412438bb 91a45aade63aafff76154016d291fc

SN 00000000001

SHA256 (K||(SN:SN:SN)) b72c3caf4dfa7468b6db5752310939f0a5 67f084dd66c19b9e77e3ce3e76f43d

iv 124bcc2b909cb5f328acb49c0f7fcdcd

Input Block 30303030303030303030310505050505 (padding)

ID f0de00ed3a0c0569243d6d1924d8be25

IDbase 8N4A7ToMBWkkPW0ZJNi+JQ==

SHA256(IDbase) 4fc25d068b5661e7e5578ad4a20e3e3f ce0c6a86fc7b0772a3988ca20b0beebf

SHA256 (SHA256(IDbase)) 27d3f2d0113b96d039f96401a431f189 ce52c2268e12bc9c86b08f0b986f9ead

IDN T8JdBotWYeﬂV4rUog4+P84Maob8ewd yo5iMogsL7r8n0/LQETuW0Dn5Z

AGkMfGJzlLCJo4SvJyGsI8LmG+erQ==

to reach the IDN. In the next subsection, we will show

a real SN, shown as IDN, in this operation biometric

network.

5.2 The Operation Communication

Protocol

Based on the protocol created, shown in Section 3,

we present the network communications by following

a log trail for an IDE (identiﬁcation) purpose. This

communication is extracted from a real operational

network, between Network 1 and Network 2, with a

real and irreversible IDN.

5.2.1 Log Trail

We show the core of a communication protocol for a

pending_operation, an IDE, and a change_status

messages, suppressing some log messages to make it

clearer to explain. The encrypted ANSI/NIST bio-

metric package is attached, for identiﬁcation purpose.

psbio-19-09-2019-10.log.gz:[2019-09-19

21:08:10.209] 3.1.12 INFO 187

server.service.DirectoryService -

[DIR:RECEIVED] FROM=‘‘bsp2"; REQUEST=

{"requestType": "pending_operations"}

psbio-19-09-2019-10.log.gz:[2019-09-19

21:08:17.376] 3.1.12 INFO 187

server.service.DirectoryService

- [DIR:SENT] TO=‘‘bsp2"; RESULT=

{"pendingOperationsList":[{"operationType":

"1_n_queue","idnList":,"idnList":[{"idn"

:"RekIJzYwMUwEE9RMSThy7kwV1SR884EC1g29VdPNzD

jM5yjAffg8X+To5tlm4p/7D/9/6W2FVJTavAORlYkAo

w==","tcn":"77a538fd-8ebe-4bdf-8d39-0ea517b1

f491"}]}]}

The pending_operation mechanism ensures

that every service provider processes all data. By re-

quest, the issuing BSP receives from other providers

all the IDN/TCN that were not processed. This com-

mand allows the network to be integrated all the time.

[2019-09-18 17:17:04.559] - [HUB:SENT]

TO=‘‘bsp2";TRANSACTION={"transaction-type"

:"IDE","idn":"L3dipU0xsnNJ1qInrbh0QCcQH7QWDN

q6IquiMv1CBhSl5WsYwoyu3xe14TMojzwyJTNjH7xxI

ysA0ETyo6gW0g==","tcn":"9F925749-ADDE-408E-

B421-3C8374E33237","timestamp":"1568837824559"}

[2019-09-18 17:35:28.596] - [HUB:RECEIVED]

FROM=‘‘bsp2";TRANSACTION=

{"transaction-type":"VRE","idn":"L3dipU0xsn

NJ1qInrbh0QCcQH7QWDNq6IquiMv1CBhSl5WsYwoyu

3xe14TMojzwyJTNjH7xxIysA0ETyo6gW0g==","tcn":

"02FA798D-D81F-43DB-9E44-32489389C470",

"timestamp":"1568838928596","reference-tcn":

"9F925749-ADDE-408E-B421-3C8374E33237","srf"

:"X"}

[2019-09-18 18:02:32.830] - [HUB:SENT]

TO=‘‘bsp2"; REQUEST={"requestType":

"change_status","statusCode":"1","idn":

"L3dipU0xsnNJ1qInrbh0QCcQH7QWDNq6IquiMv1C

BhSl5WsYwoyu3xe14TMojzwyJTNjH7xxIysA0ETy

o6gW0g=="}

It is possible to interpret that the IDE request was

made for the encrypted IDN “L3dipU...”, which re-

sponse was a VRE with X value, i.e., no biometrics

were found in the biometric database. Then, the issu-

ing system sent a change_status acknowledge mes-

sage, completing the registration process.

6 CONCLUSIONS

In this work, we proposed a new approach scheme

that guarantees the anonymity of anyone within bio-

metric databases and still allows them to communi-

cate securely. Symmetric cryptography techniques

and known communication protocols were used to

Achieving Privacy, Security, and Interoperability among Biometric Networks using Symmetric Encryption

487

achieve privacy, with security evidence and interoper-

ability integrity between biometric networks. We suc-

cessfully showed that, for the same input data and se-

cret key among systems, we produced an anonymous

index register into all databases representing a person.

Also, we improved the IEEE BOPS standard by con-

structing a framework for JSON messages between

systems, including a way of networks to maintain op-

erations regardless of the contingencies. As future

works, we could generate IDN for any government

database that needs privacy, security, and interoper-

ability. This future work is an immediate outcome

of the contributions we made to enhance the security,

and also, of the shared anonymous record index we

built.

REFERENCES

Bellare, M. and Kohno, T. (2004). Hash function balance

and its impact on birthday attacks. In Cachin, C. and

Camenisch, J. L., editors, Advances in Cryptology -

EUROCRYPT 2004, pages 401–418, Berlin, Heidel-

berg. Springer Berlin Heidelberg.

Bellare, M. and Rogaway, P. (1995). Optimal asymmet-

ric encryption. In De Santis, A., editor, Advances

in Cryptology — EUROCRYPT’94, pages 92–111,

Berlin, Heidelberg. Springer Berlin Heidelberg.

Bernstein, D. and Lange, T. (2014). Batch NFS. In Joux,

A. and Youssef, A., editors, Selected Areas in Cryp-

tography – SAC 2014: 21st International Conference,

Montreal, QC, Canada, August 14-15, 2014, Revised

Selected Papers, Lecture Notes in Computer Science,

pages 38–58, Germany. Springer.

Boldyreva, A. and Fischlin, M. (2006). On the security

of oaep. In Lai, X. and Chen, K., editors, Advances

in Cryptology – ASIACRYPT 2006, pages 210–225,

Berlin, Heidelberg. Springer Berlin Heidelberg.

Bray, T. (2017). The JavaScript Object Notation (JSON)

Data Interchange Format. RFC 8259.

Campisi, P. (2013). Security and Privacy in Biometrics.

Springer Publishing Company, Incorporated.

Dang, Q. (2015). Secure Hash Standard (SHS). Federal In-

formation Processing Standards Publication FIPS Pub

180-4, pub-NIST, pub-NIST:adr.

Ding, D., Han, Q.-L., Xiang, Y., Ge, X., and Zhang, X.-

M. (2018). A survey on security control and attack

detection for industrial cyber-physical systems. Neu-

rocomputing, 275:1674–1683.

Ding, Y., Shi, Y., Wang, A., Zheng, X., Wang, Z., and

Zhang, G. (2019). Adaptive Chosen-Plaintext Col-

lision Attack on Masked AES in Edge Computing.

IEEE Access, 7:63217–63229.

Dobraunig, C., Eichlseder, M., and Mendel, F. (2015).

Analysis of SHA-512/224 and SHA-512/256. In Pro-

ceedings, Part II, of the 21st International Confer-

ence on Advances in Cryptology — ASIACRYPT 2015

- Volume 9453, pages 612–630, Berlin, Heidelberg.

Springer-Verlag.

Dwork, C. (2006). Differential Privacy. In Proceedings

of the 33rd International Conference on Automata,

Languages and Programming - Volume Part II,

ICALP’06, pages 1–12, Berlin, Heidelberg. Springer-

Verlag.

Dworkin, M. J. (2001). SP 800-38A 2001 Edition. Rec-

ommendation for Block Cipher Modes of Operation:

Methods and Techniques. Technical report, Gaithers-

burg, MD, United States.

Hirano, T., Ito, T., Kawai, Y., Matsuda, N., Yamamoto,

T., and Munaka, T. (2016). A practical attack to

AINA2014’s countermeasure for cancelable biometric

authentication protocols. In 2016 International Sym-

posium on Information Theory and Its Applications

(ISITA), pages 315–319.

IEEE (2019). IEEE Standard for Biometric Open Protocol,

Redline. IEEE Std 2410-2019 (Revision of IEEE Std

2410-2017), Redline, pages 1–134.

Jain, A. K., Nandakumar, K., and Ross, A. (2016). 50

years of biometric research: Accomplishments, chal-

lenges, and opportunities. Pattern Recognition Let-

ters, 79:80–105.

Josefsson, S. (2006). The Base16, Base32, and Base64 Data

Encodings. RFC 4648.

Kang, H., Park, M., Moon, D., Lee, C., Kim, J., Kim, K.,

Kim, J., and Hong, S. (2016). New efﬁcient padding

methods secure against padding oracle attacks. In

Kwon, S. and Yun, A., editors, Information Security

and Cryptology - ICISC 2015, pages 329–342, Cham.

Springer International Publishing.

Katz, J. and Lindell, Y. (2007). Introduction to Modern

Cryptography (Chapman & Hall/Crc Cryptography

and Network Security Series). Chapman & Hall/CRC.

Kaur, H. and Khanna, P. (2019). Random Distance Method

for Generating Unimodal and Multimodal Cancelable

Biometric Features. IEEE Transactions on Informa-

tion Forensics and Security, 14:709–719.

Kocher, P. C. (1996). Timing Attacks on Implementa-

tions of Difﬁe-Hellman, RSA, DSS, and Other Sys-

tems. In Proceedings of the 16th Annual Interna-

tional Cryptology Conference on Advances in Cryp-

tology, CRYPTO ’96, pages 104–113, London, UK,

UK. Springer-Verlag.

Kumar, A. and Kumar, A. (2016). A Cell-Array-Based

Multibiometric Cryptosystem. IEEE Access, 4:15–25.

Li, C., Hu, J., Pieprzyk, J., and Susilo, W. (2015). A New

Biocryptosystem-Oriented Security Analysis Frame-

work and Implementation of Multibiometric Cryp-

tosystems Based on Decision Level Fusion. IEEE

Transactions on Information Forensics and Security,

10(6):1193–1206.

Mangold, K. (2016). Data Format for the Interchange

of Fingerprint, Facial and Other Biometric Informa-

tion ANSI/NIST-ITL 1-2011 NIST Special Publica-

tion 500-290 Edition 3. Number 500-290e3.

Mason, S., Gashi, I., Lugini, L., Marasco, E., and Cukic,

B. (2014). Interoperability between Fingerprint Bio-

metric Systems: An Empirical Study. In 2014 44th

ICISSP 2020 - 6th International Conference on Information Systems Security and Privacy

488

Annual IEEE/IFIP International Conference on De-

pendable Systems and Networks, pages 586–597.

Moriarty, K., Kaliski, B., Jonsson, J., and Rusch, A. (2016).

PKCS #1: RSA Cryptography Speciﬁcations Version

2.2. RFC 8017.

Nasir, M. and Perumal, P. (2013). Implementation of Bio-

metric Security using Hybrid Combination of RSA

and Simple Symmetric Key Algorithm. Interna-

tional Journal of Innovative Research in Computer

and Communication Engineering, 3297.

Natgunanathan, I., Mehmood, A., Xiang, Y., Beliakov, G.,

and Yearwood, J. (2016). Protection of Privacy in Bio-

metric Data. IEEE Access, 4:880–892.

Ngo, D. C. L., Teoh, A. B. J., and Hu, J. (2015). Biomet-

ric Security. Cambridge Scholars Publishing, United

Kingdom.

Quan, F., Fei, S., Anni, C., and Feifei, Z. (2008). Crack-

ing Cancelable Fingerprint Template of Ratha. In

2008 International Symposium on Computer Science

and Computational Technology, volume 2, pages 572–

575.

Rescorla, E. (2000). HTTP Over TLS. RFC 2818.

Rogaway, P. (2011). Evaluation of Some Blockcipher

Modes of Operation. Evaluation carried out for the

Cryptography Research and Evaluation Committees

(CRYPTREC) for the Government of Japan.

Ross, A., Shah, J., and Jain, A. K. (2007). From Template

to Image: Reconstructing Fingerprints from Minutiae

Points. IEEE Transactions on Pattern Analysis and

Machine Intelligence, 29(4):544–560.

Schaffer, K. B. (2019). Security Requirements for Cryp-

tographic Modules. Federal Information Processing

Standards Publication FIPS Pub 140-3, pub-NIST,

pub-NIST:adr.

Schramm, K., Leander, G., Felke, P., and Paar, C. (2004).

A collision-attack on aes. In Joye, M. and Quisquater,

J.-J., editors, Cryptographic Hardware and Embedded

Systems - CHES 2004, pages 163–175, Berlin, Heidel-

berg. Springer Berlin Heidelberg.

Toli, C.-A. and Preneel, B. (2018). Privacy-preserving

biometric authentication model for e-ﬁnance applica-

tions. In ICISSP.

Tolosana, R., Vera-Rodriguez, R., Ortega-Garcia, J., and

Fierrez, J. (2015). Preprocessing and Feature Se-

lection for Improved Sensor Interoperability in On-

line Biometric Signature Veriﬁcation. IEEE Access,

3:478–489.

Turan, Meltem S.and Barker, E. B., Kelsey, J.and McKay,

K., Baish, M., and Boyle, M. (2018). SP 800-90B

Recommendation for the Entropy Sources Used for

Random Bit Generation. Technical report, Gaithers-

burg, MD, United States.

Wenqian Yu, Weigang Li, Junyuan Wang, and Changzheng

Wei (2016). A study of HSM based key protection in

encryption ﬁle system. In 2016 IEEE Conference on

Communications and Network Security (CNS), pages

352–353.

Yan, Q., Yu, F. R., Gong, Q., and Li, J. (2016). Software-

Deﬁned Networking (SDN) and Distributed Denial

of Service (DDoS) Attacks in Cloud Computing En-

vironments: A Survey, Some Research Issues, and

Challenges. IEEE Communications Surveys Tutorials,

18(1):602–622.

Achieving Privacy, Security, and Interoperability among Biometric Networks using Symmetric Encryption

489