A secure lightweight fuzzy embedder based user authentication scheme for internet of medical things applications

Abstract

The Internet of Medical Things (IoMT) is a network of medical devices, hardware infrastructure, and software that allows healthcare information technology to be communicated over the web. The IoMT sensors communicate medical data to server for the quick diagnosis. As, it handles private and confidential information of a user, security is the primary objective. The existing IoT authentication schemes either using two-factor(Username, password) or multi-factor (username, password, biometric) to authenticate a user. Typically the structural characteristics-based biometric trait like Face, Iris, Palm print or finger print is used as a additional factor. There are chances that these biometrics can be fabricated. Thus, these structural biometrics based authentication schemes are fail to provide privacy, security, authenticity, and integrity. The biodynamic-based bioacoustics signals are gained attention in the era of human-computer interactions to authenticate a user as it is a unique feature to each user. So, we use a frequency domain based bio-acoustics as a biometric input. Thus, this work propose a Secure Lightweight Bioacoustics based User Authentication Scheme using fuzzy embedder for the Internet of Medical Things applications. Also, the IoT sensors tends to join and leave the network dynamically, the proposed scheme adopts chinese remainder technique for generate a group secret key to protect the network from the attacks of former sensor nodes. The proposed scheme’s security is validated using the formal verification tool AVISPA(Automated Validation of Internet Security Protocols and Applications). The system’s performance is measured by comparing the proposed scheme to existing systems in terms of security features, computation and communication costs. It demonstrates that the proposed system outperforms existing systems.

Keywords

e-Healthcare internet of medical things security remote patient monitoring user authentication bioacoustics fuzzy embedder

1 Introduction

The development of the Internet and technologies causes people to access countless numbers of services via the Internet. People utilized their electric gadgets like notebooks, tablets, and mobiles to contact the remote server and access various information. The remote access concept implemented in health industries and most countries concentrates on telemedicine and e-health services to provide the easy and high availability of medical services [1]. The e-health services include the eCW health care portal [2] (eClinical Works, Georgia), Manage MyHealth [3] (NewZealand), Boynton Health [4](Minnesota University), and National Health Portal [5] (Government of India). The low-cost internet services cause most people to access health services remotely. Rapid development of new innovative technology in e-healthcare services, people access several healthcare assistance via the Internet that reduce the physical contact time and complexities [6]. This IoMT network requires a stronger security infrastructure than the other domains like Industrial IOT, Vehicular Adhoc NETworks [7], Mobile Ad-hoc NETworks [8] and others.Due to the impact of Covid-19, various self-medical devices available in the market such as insulin pumps, cardiac monitoring devices, blood pressure monitors, blood glucose monitors, respiratory devices, multipara meter monitors, Imaging systems, anesthesia machines, patient monitoring, neurological devices, implantable cardioverter defibrillators, fetal monitoring devices, ventilators, infusion pump, implantable cardiac monitors which are continuously monitoring patient health. The digital data from the sensor are stored as an Electronic Health Records(EHR) and transferred to the third-party medical server for the clinical analysis[9]. These remote sensors are reduce the unnecessary hospital visit, hospital stays, and overload hospital capacity.

The IoMT has become an emerging field in the domain of IoT nowadays. As per the report of Precedence Research, global healthcare market size is projected to reach around USD 960.2 billion by 2030 with a compound annual growth rate (CAGR) of 20.41% during the forecast period 2022 to 2030. The Remote Patient Monitoring(RPM) system is the primary application of IoMT. As in the RPM application layered architecture shown in Fig. 1, the perception layer sensors with the patient body monitors the change in a remote patient body signal and sends to the medical server via the Gateway Node. The data is shared using the wireless networks like bluetooth or Wi-Fi. As the data transferred are private, personal and sensitive in nature, it is important to secure the communication from the venerable attacks and store safely [10]. The key challenge for the medical IoT market is to establish the Secure wireless connectivity between the remote patient and the doctor in the RPM system. The mutual authentication schemes are helping to address this challenge [53].

Fig. 1

IoMT RPM application Layered Architecture.

Initially, the two factor authentication schemes are proposed to establish the secure session between the users. Various attacks are breaking these schemes and intrude the applications. To improve the security, additional factors are introduced in the authentication schemes like biometric, multimodal and multifactor. The majority of IoMT applications employ a two-factor or multi-factor authentication system based on structural characteristics-based (fingerprint, face recognition, iris recognition, or any other form of biometric trait) biometric authentication methods to resolve the privacy, authenticity, and integrity issues. However, these physical characteristics based biometrics can be manipulated.[11]. To address this research gap, researchers are proposing several biodynamic-based bioacoustics signal to authenticate a user in the era of human-computer interactions. These frequency-based bio-acoustic inputs as a lightweight signal to use in the IoMT environment. Bio-acoustics is the study of the production, dispersion, and reception of sound in human beings. The human body produces various bio-acoustic sounds like heart sounds, lung sounds, stomach sounds, ligament sounds, etc., It has been found that the bio-acoustic sounds are unique for each human being. Existing studies proved the classification accuracy of these bioacoustic signals using various machine learning and deep learning models [11]. The state-of-the-art protocols are proven to be vulnerable to various cyber threats. Even the most resource-efficient protocols have significant computation and communication overhead in an environment like the Internet of Things.

With this motivation, this work propose a secure authentication scheme with bio-acoustics input to authenticate a IoMT user and establish a secure session. Smartwatches can be used for sensing bio-acoustic sound vibrations with the help of the gyroscope and accelerometer sensor [12]. The users can register themselves with the help of smartwatches and the smartwatch can be used for authentication purposes as well. The Internet of Things(IoT) sensors are resource constrained and work on low-memory devices, which requires lightweight protocols. We propose a novel lightweight bio-acoustic based authentication scheme with fuzzy embedder technique for securing the IoMT RPM applications. The security goals considered to propose this scheme are to provide a Strong security that the attacker cannot successfully impersonate, preserve the user anonymity and identity, resistivity against password guessing and dictionary attacks and maintain the key Freshness that protects temporary sensitive data from information leakage attacks.

2 Literature survey

Authentication is the process of identifying whether a user or a program or a request is genuine or malign. It is the process of validating the identity of a program or a registered user when they try to login or access a protected resource. Authentication is very important as it ensures that only authorized users or processes are allowed to access protected resources. The data or the attribute which is used for authenticating a user is called an authentication factor. Based on the authentication factor, the authentication technique can be classified as Two-factor authentication, Three-factor authentication, Multi-factor authentication, Multimodal authentication.

2.1 Two-factor authentication

In a two-factor authentication system, two authentication factors(smartcard and password/password and fingerprint) are used for authenticating the user. The most commonly used two-factor authentication scheme is the smartcard and password authentication. But if the smartcard gets compromised the system will become prone to various attacks. Comprehensive and systematic metrics are not available for schemes to be assessed objectively. Wang et al. [13] proposed a Fuzzy extractor along with honeywords to overcome the problem of systematic metrics by providing a benchmark for the evaluation of current and future two-factor authentication systems. But the storage cost is higher when compared to other models.

Advanced hacking techniques produce a severe threat to the traditional authentication system like pin/password/fingerprint. Zhang et al. [14] proposed a usable two-factor authentication called MagAuth along with COTS wrist wearables to increase the security and usability of password-based authentication systems in mobile touchscreen devices. A major defect of this system is that the accuracy will be low while swiping fast. Most of the systems require extra user effort for authentication which seriously affects user interaction and delays authentication time. To avoid this problem, Yetong Cao et al. [15] proposed a novel two-factor authentication for mobile called PPGpass which uses unique PPG features as one factor of authentication, that is used to increase the security of existing mobile authentication. But the quality of the signal from the PPG sensor is low and the measurement from the sensor varies with positioning.

To overcome that issue Qingxuan Wang et al. [16] proposed Quantum 2FA, a practical quantum-resistant smartcard-based password authentication system that employs lattice-based key exchange and the “fuzzy-verifier + honeywords” technique which provides the best computational efficiency than the other two-factor authentication. But there is a chance of man-in-the-middle attacks, in which an attacker may attack communication between the server and authentication system. To strengthen the two factor authentication system, multifactor and multimodal systems are being introduced.

2.2 Three-factor authentication

The Qi Jiang et al. [17] proposed a three-factor authentication with a unified biometric privacy-preserving scheme. They provided security for remote access of aerial vehicles through a three-factor authentication system. The model provided a system that can provide authentication while preserving the privacy of the user. When the number of IoT devices increases, it is not possible for a single server architecture to provide high security. To solve this problem, Chien-Lung et al. [18] proposed a dynamic time-bound key distribution scheme with three-factor authentication. They provided a highly secure environment with multi-server-based architecture. This model provides high security and gives resistance to many attacks but the computational costs are higher when compared to other models. There are schemes[19 –24] proposed for graph based authentication codes generation using the mutually orthogonal graph squares.

The IoT devices are not capable of computing traditional authentication protocols, which require more memory & Computaional capacity. Hence, Shuming Qiu et al. [25] proposed a three-Factor Authentication Protocol Based on Extended Chaotic-Maps for Mobile Lightweight Devices. This model outperformed other models by meeting all 13 evaluation criteria regarding security. But this model is too complex and expensive. As long as data is collected and sent from less secure devices such as wearables or home appliances, it remains difficult to ensure the security of both data-at-rest and data-in-transit. Zhang et al. [26] proposed a Fast Authentication and Key Agreement based on three-factor authentication using biohash technique on iris.The Sahoo et al. [27] proposed the three factor authentication system for the health sector using the IoT devices, they use Elliptic Curve Cryptography(ECC) due to its small key size and high security and fuzzy extractor for the biometric feature extraction.

2.3 Multifactor authentication

In a multifactor authentication system, the security will be low when there is a leakage in one of the factors. Existing multifactor protocols do not provide much security and efficiency. To conquer, Rui Zhang et al. [28] proposed the MFAKE protocol which provides security higher than the previous model. But it requires more time for execution. There is rapid development in mobile technologies and the new challenges regarding multifactor authentication are also increasing. Dragan Korac et al. [29] proposed a model called the Fishbone model in the form of a universal authentication framework (UAF) which is more efficient in practice.

In Biometric-based authentication, performing user authentication depends on explicit user input and Internet connection. Tiantian Zhu et al. [30] proposed the system called RiskCog which does not require the user’s explicit input and has no necessity on the device placement. But here authentication is slow when compared to other models. Vehicle to Grid (V2G) networks must carefully manage shared keys due to security and privacy concerns. V2G networks have been proposed to employ several key agreement protocols in recent years. The existing lightweight schemes do not provide much security properties. So Dariush Abbasinezhad-Mood et al. [31] proposed an anonymous key agreement scheme with chaotic map cryptography which provides performance as well as security.

Nowadays, the scope of online healthcare services has increased enormously. One of the interfaces is Telecare Medicine Information System(TMIS).In this system, there is a connection between users from remote areas and hospitals to share the necessary information between them.The security of information that is transferred is very important. Dharminder Dharminder1 et al. [32] proposed a secure, efficient chaotic map-based authentication protocol that can be utilized in telemedicine information systems.

Multifactor authentication is mainly used in safeguarding high-value assets. Wenting Li et al. [33] proposed a protocol called T-MFAKE to bring back flexibility and usability without loss of security. But they use a fuzzy extractor in which if biometrics gets compromised then the security key will also get compromised. Multifactor authentication is slightly complicated as it requires extra factors for authentication from the user. Mangal Sainet et al.[34] discussed the evolution of the authentication process from single-factor authentication through two-factor and ending with multi-factor authentication.

2.4 Multimodal authentication

IoT is enormously developed in the modern world which interconnects several devices into smaller portable devices like smartwatches. Most smart devices like Smartwatches use explicit authentication like biometric in one kind and the same application is accessed in the internet via another mode of login like pin, password or OTP[35]. But due to the small display, users don’t feel comfortable typing in the pin, password, etc. To overcome this problem, Sudip Vhaduri et al. [36] proposed an implicit authentication system that uses behavioral (step counts), physiological (heart rate), and hybrid (calorie burn and metabolic equivalent of task). But in heartbeat authentication, the accuracy will be low.

Mikel Labayen et al. [37] proposed different biometric technologies to authenticate and proctor the students’ activities in online courses. They used biometric recognition and digital signal processing algorithms to authenticate the students. And also, they used AI algorithms to monitor the students’ activities automatically. The proposed model wasn’t able to verify the student’s identity due to variance in face pose, dim light, and sound conditions. L. Nisha Evangelin et al. [38] proposed to protect the multimodal biometric images using cryptography techniques. They used the Elliptic Curve Cryptography method and the Visual Shadow Creation Process to create a secret image from shadows by encryption and decoding process. This model takes more computational time. To overcome a spoof attack in a unimodal biometric authentication system, Basma Abd El-Rahiem et al. [39] proposed a multimodal authentication system using ECG and finger vein. They used preprocessing and feature extraction using CNN for authentication. And they also used some machine learning classifiers to authenticate the user. First-time authentication takes more time to authenticate in this proposed model and to obtain high performance it needs a big dataset.

The Table 1 compares the various schemes discussed in the literature review. Comparison is given among the schemes with unique techniques used for biometric keys. The table summarizes the different technique used to generate a secret key from the biometric input. Schemes[40 –43] uses fuzzy extractor for the biometrics secret key. Few schemes like[44, 45] uses fuzzy commitment technique for the biometric processing. The scheme[46] uses the fuzzy embedder technique. All these schemes are using the physical traits as a biometric and which could be fabricated and vulnerable to impersonation attack. The multifactor and multimodal authentication schemes are using techniques like OTP, PUF, ECC and many more, which are computationally costly.

Table 1
Summary of the literature’s

Studies Authentication Factors Techniques Merits Limitations

Two Factor Authentication Schemes

Wang et al. [13] Smart Card, Password Fuzzy Verifier + HoneyWords Provides a systematic benchmark metrics for evaluate a two-factor authentication schemes High Storage bits for communication

Zhang et al. [14] Unlock pattern, Behavioral features using wrist wearable MagAuth Balances the security and usability with magnetic field theory. Authentication accuracy is very low if user swipe is fast.

Yetong Cao et al. [15] Password/pattern, physiological signal Random Forest and Support Vector Initiated a novel nonintrusive Photoplethysmography(PPG) sensors based secure two-factor authentication scheme for mobiles. PPG sensor signal quality is very low for varies signal positioning.

Multi Factor Authentication Schemes

Qi Jiang et al. [17] Password, Smart Card, Biometric(Finger, Iris) Fuzzy Commitment Provided security for remote access of aerial vehicles through a three-factor authentication. High computational overhead than the prevailing schemes.

Shuming Qiu et al. [25] Password, Smart Card, Biometric Extended chaotic-maps, Fuzzy Extractor Proposed a authentication model using the biometric from user mobile device Their scheme vulnerable to Biometric Template attack, Stolen Smart Card attack and Impersonation attack.

Zhang et al. [26] Password, Smart Card, Iris BioHash Proposed a Fast Authentication scheme for Medical IoT application. High Computational Cost

Sahoo et al. [27] Password, Smart Card, Finger Print ECC, Fuzzy Extractor Proposed a scheme for Telecare Medical Information System(TMIS) without computational overhead. Fails to protect the Biometric Template attack and Impersonation attack.

Sudhakar et al. [32] Password, Smart Card, Finger Print Fuzzy Embedder The scheme protect the biometric template and preserve the user anonymity for a client server based application. Not suitable for IoT applications.

Studies	Authentication Factors	Techniques	Merits	Limitations
Two Factor Authentication Schemes
Wang et al. [13]	Smart Card, Password	Fuzzy Verifier + HoneyWords	Provides a systematic benchmark metrics for evaluate a two-factor authentication schemes	High Storage bits for communication
Zhang et al. [14]	Unlock pattern, Behavioral features using wrist wearable	MagAuth	Balances the security and usability with magnetic field theory.	Authentication accuracy is very low if user swipe is fast.
Yetong Cao et al. [15]	Password/pattern, physiological signal	Random Forest and Support Vector	Initiated a novel nonintrusive Photoplethysmography(PPG) sensors based secure two-factor authentication scheme for mobiles.	PPG sensor signal quality is very low for varies signal positioning.
Multi Factor Authentication Schemes
Qi Jiang et al. [17]	Password, Smart Card, Biometric(Finger, Iris)	Fuzzy Commitment	Provided security for remote access of aerial vehicles through a three-factor authentication.	High computational overhead than the prevailing schemes.
Shuming Qiu et al. [25]	Password, Smart Card, Biometric	Extended chaotic-maps, Fuzzy Extractor	Proposed a authentication model using the biometric from user mobile device	Their scheme vulnerable to Biometric Template attack, Stolen Smart Card attack and Impersonation attack.
Zhang et al. [26]	Password, Smart Card, Iris	BioHash	Proposed a Fast Authentication scheme for Medical IoT application.	High Computational Cost
Sahoo et al. [27]	Password, Smart Card, Finger Print	ECC, Fuzzy Extractor	Proposed a scheme for Telecare Medical Information System(TMIS) without computational overhead.	Fails to protect the Biometric Template attack and Impersonation attack.
Sudhakar et al. [32]	Password, Smart Card, Finger Print	Fuzzy Embedder	The scheme protect the biometric template and preserve the user anonymity for a client server based application.	Not suitable for IoT applications.

The discussed schemes are either vulnerable to attacks or failed to preserve the user anonymity, privacy and user biometric template protection. Otherwise, the schemes resulting the overhead in terms of communication cost, computational cost or storage cost by increase the number of security factors. To overcome the listed limitations in the existing schemes, this work proposes a lightweight authentication scheme with better security, biometric template protection and privacy preservation.

The major contributions of the proposed system are:

In the IoMT system, we introduce a lightweight authentication system with frequency-based biometric input(bio-acoustics) instead of structural characteristics-based biometrics(fingerprint, face, iris, Palm and others). As the structural characteristics based biometrics can be tampered.

Most of the authentication system uses either fuzzy extractor or fuzzy commitment schemes to process the biometric features. As, the fuzzy extractor technique is vulnerable to Biometric-Template attack and allows the impersonation attack during the password change phase and the fuzzy commitment technique is computationally high. Hence, In the proposed system, a fuzzy embedder technique is used to process the user Bioacoustic data.

The dynamic nature of the IoT network is detailed in the proposed scheme with the dynamic sensor node join and revocation phases and the proposed scheme security analysis is carried using the AVISPA simulation tool.

An informal security analysis is done to ensure that the proposed bio-acoustic based authentication scheme secure against well-known attacks. Also, the comparative study on Communication Cost, Computation Cost and Security features is done to establish that the proposed scheme is performing better than the other existing schemes.

3 The proposed system model preliminaries

This section discusses the authentication model, Threat model, Notations used and other preliminaries that are required to understand the proposed secure user authentication scheme for medical IoT clearly.

3.1 Authentication model

A remote patient monitoring system with IoT-based smart sensing devices is given in the below Fig. 2. The remote patient monitoring(RPM) system is built to create smart hospital environments and has the IoT sensors and devices to improve data-driven clinical decision making, improve quality of care, and increase the capacity for physicians to treat more patients. The admin of the system will monitor the sensors and devices in regular intervals. Due to the limited availability of resources of IoT sensors and devices in such an environment, it will be a more difficult challenge to provide secure on-demand data transmission. Also, there is a problem with the physical capturing of deployed devices. In order to solve this problem, we propose a secure and efficient user authentication system in which only a legitimate authorized user (such as a doctor or patient) can get real-time data from sensors and smart IoT devices. A legitimate user can set up a secure connection to an IoT sensor using a Gateway Node via the user authentication system under this authentication model.

Fig. 2

Authentication System Model.

3.2 Threat model

As in the DY model, the CK adversary model allows Attacker $A$ to do each and every capability but also to expose the session states and keys as well as the secret credentials. So, a user authentication system designed for the IoMT environment must guarantee that regardless of whether some kind of secret credentials like the session ephemeral secrets and session key are exposed to $A$ , it should be able to have the least possible impact on the secret of the users’ other credentials involved in communication. In the IoMT environment, the Gateway Nodes will be put under a locking framework to safeguard the devices from physical capture by $A$ , and therefore, they will be considered as trusted participants in the IoMT network.

3.3 Fuzzy based biometric feature extraction techniques

In this section, introduces various techniques available for generating secret key based on biometric input data. Most commonly used techniques are the fuzzy extractor[47], fuzzy commitment[48] & fuzzy embedder[46].

Fuzzy Embedder: The fuzzy embedding model provides two operations, namely embed and rep. The biometric and random secret number are passed into the embed function, and the resulting string is made available to the public. To authenticate a user, the rep function accepts the public string and a biometric input that is similar to the biometric input used during registration, and returns a random secret number.

Embedded function (Embed): This function used to user registration and re-registration phase, which accepts the biometric identifier BA_i and a random secret numberm as input and returns a publicly accessible string CW_i.

Embed (BA_i, m) = CW_i

Reproduce function(Rep): This function used to login & authentication phase, which takes new user biometric ${BA}_{i}^{*}$ and public parameter CW_i as input and reproduce the secret key ‘m’ as output.

$Rep ({BA}_{i}^{*}, {CW}_{i}) = m$

3.4 Notations table

The notations used in this work are listed in the Table 2.

Table 2
Notations Used

Notation Description

G Gateway Node

SK _G G’s Secret Master key

P _i User

U _id Identity of the User i

(Physician/Patient)

IS _j IoMT Sensor Device

$A$ Attacker/Adversary

Pwd Password

SM _i Smart Card

BA _i User Bio-acoustics input

⊕ Bit-wise exclusive-OR

∥ String Concatenation

h (.) One-way Hash Function

Embed (.) Fuzzy Embedder Function

Rep (.) Fuzzy Reproduce Function

m, n, a, b, NC_i, NC_j Random numbers

Notation	Description
G	Gateway Node
SK _G	G’s Secret Master key
P _i	User
U _id	Identity of the User i
(Physician/Patient)
IS _j	IoMT Sensor Device
$A$	Attacker/Adversary
Pwd	Password
SM _i	Smart Card
BA _i	User Bio-acoustics input
⊕	Bit-wise exclusive-OR
∥	String Concatenation
h (.)	One-way Hash Function
Embed (.)	Fuzzy Embedder Function
Rep (.)	Fuzzy Reproduce Function
m, n, a, b, NC_i, NC_j	Random numbers

4 The proposed fuzzy embedder based secure user authentication scheme

This section introduce a lightweight fuzzy embedder based user authentication system with bio-acoustics signal for the IoMT applications. The proposed work is explained with following phases namely Pre-deployment, User Registration, Login & Authentication and Bio-acoustics based biometric Password change, followed by the Bio-acoustics based biometric Extraction process. The dynamic nature of the medical IoT environment the sensor nodes can enter and exit from the system after the deployment. Therefore, Two additional phases namely Dynamic join and Revocation of Sensor Nodes has been given in the proposed scheme for the flexible architecture[49, 50] of IoT environment.

4.1 Bio-acoustics biometric extraction

An acoustic sensor and a transducer are placed on the finger as shown in Fig. 3 The transducer sends a signal through the finger which is then captured by the acoustic sensor. The captured signal now contains several information about the finger such as the structure of bone, cartilage, etc., The biometrics feature is extracted as frequency domain information as similar to [11, 51] and fed as input to our proposed scheme.

Fig. 3

A system model for extracting bio acoustics signals.

4.2 Pre-deployment phase

In the Pre-deployment Phase, the IoMT sensors(IS_j) are register with the Gateway Node(G) before they are deployed in the network. The G computes a secret master key (SK_G) based on the number of devices are being deployed using the chinese remainder theorem principle. In addition the G generates secret keys for for each sensor. The G selects a n-dimensional matrix, where n denotes the number of sensors, assigns a unique identities for all the sensors namely IS_j and generates sec_j = j^throw × x_j along with pairwise prime numbers p₁, . . . p_n, where j= 1, 2,..., n. Then, G computes $GropuMul = \sum_{j = 1}^{n} p_{j}$ and SensorMul_j = GropuMul/p_j. Random secret keys Xkey_j for each IS_j are being generated by the G which are satisfies the condition SensorMul_j × Xkey_j ≡ 1 mod p_j. The group secret is computed as ${SK}_{G} = \sum_{j = 1}^{n} {SensorMul}_{j} \times p_{j}$ . Finally, {IS_j, sec_j, p_j} are stored into the sensor’s memory by the G.

4.3 User registration phase

In this phase, the Users can register themselves with the Gateway Node(G) in a closed environment via a secure channel. The detailed procedure and the computations involved with the user registration are given in Table 3.

Table 3
User Registration Protocol

User(Doctor/Patient) Gateway Node(G)

Select U_id pwd

Gets bio - acoustic input BA_i and

Generates random integers m and n

Calculates CW_i = embed (BA_i, m)

AID_i = h (U_id||m||pwd||n)

Registration request :

xrightarrow [Securechannel] {AID_i, U_id, CW_i}

G picks random numbers a, b and then Calculates

P_i = h (U_id||a||b)

Q_i = h (U_id||h (b) ||AID_i||CW_i)

R_i = h (P_i||h (a||b)) ⊕ AID_i

S_i = P_i ⊕ h (a||b)

Issues Smartcard :

             SM_i = {Q_i, R_i, S_i, h (.) , h (b)}

              ← - - - - - - - - - - - - - -

             Secure channel

Stores the following in Smartcard,

SM_i = {n, CW_i, Q_i, R_i, S_i, h (.) , h (b)}

User(Doctor/Patient)	Gateway Node(G)
Select U_id pwd
Gets bio - acoustic input BA_i and
Generates random integers m and n
Calculates CW_i = embed (BA_i, m)
AID_i = h (U_id\|\|m\|\|pwd\|\|n)
Registration request :
xrightarrow [Securechannel] {AID_i, U_id, CW_i}
	G picks random numbers a, b and then Calculates
	P_i = h (U_id\|\|a\|\|b)
	Q_i = h (U_id\|\|h (b) \|\|AID_i\|\|CW_i)
	R_i = h (P_i\|\|h (a\|\|b)) ⊕ AID_i
	S_i = P_i ⊕ h (a\|\|b)
	Issues Smartcard :
	SM_i = {Q_i, R_i, S_i, h (.) , h (b)}
	← - - - - - - - - - - - - - -
	Secure channel
Stores the following in Smartcard,
SM_i = {n, CW_i, Q_i, R_i, S_i, h (.) , h (b)}

The user(Doctor/Patient) selects a User ID and password. The bio-acoustic data is read with the help of sensors and it will be processed using the fuzzy embedded’s Embedd (.) function. Two random numbers m and n are generated. Then, User(P_i) calculates the following CW_i = embed (BA_i, m) , AID_i = h (U_id ∥ m ∥ pwd ∥ n). Then, the user P_i sends the registration request which contains AID_i, U_id, CW_i to the Gateway Node(G) via a secure channel. Upon receiving the registration request, G picks random numbers a, b and then Calculates P_i = h (U_id||a||b), Q_i = h (U_id||h (b) ||AID_i||CW), R_i = h (P_i||h (a||b)) ⊕ AID_i, and S_i = P_i ⊕ h (a||b). Then the G stores Q_i, R_i, S_i, h (.) , h (b) in a smartcard and issues the smartcard to the User(Pi) via a secure channel. When the smartcard is received, the User(Pi) stores the following {n, CW_i, Q_i, R_i, S_i, h (.) , h (b)} in smartcard.

4.4 Login and authentication phase

Once the User(Pi) is registered with the Gateway Node(G), they can login to the system with the smartcard, password and Bioacastics biometric. The protocol authenticate the user and then authorize them to access the data from the sensors as discribed in Table 4.

Table 4
Login and Authentication Protocol

User P_i with SM_i = Gateway Node (G) IoT Sensing Sevice IS_j

{n, CW_i, Q_i, R_i, S_i, h (.) , h (b)}

Insert SM_i into a card reader .

$Input U_{id} pwd / & {BA}_{i}^{'} is read$

Computes

AID_i = h (U_id||m||pwd||n)

$Q_{i}^{} = h (U_{id} | | h (b) | | {AID}_{i} | | {CW}_{i})$

$Checks Q_{i}^{} \overset{?}{=} Q_{i}$

If so, create two nonce NC_i NC_j

Generate timestamp TS₁

Calculate,

CW_ij = S_i ⊕ h (IS_j||h (b) ||NC_i)

T_i = R_i ⊕ AID_i

A1_i = AID_i ⊕ N_j ⊕ T_i

IDC_i = A1_i ⊕ h (T_i||IS_j||NC_i)

MSG₁ = h (CW_ij||IDC_i||A1_i||NC_i)

xrightarrow [(P_ixrightarrow G)] M₁ = {CW_ij, IDC_i, MSG₁, NC_i, TS₁}

$Check if | {TS}_{1}^{'} - {TS}_{1} | < Δ T ?$

If so, compute

S_i = CW_ij ⊕ h (IS_j||h (b) ||NC_i)

P_i = S_i ⊕ h (a||b)

T_i = h (P_i||h (a||b))

A1_i = IDC_i ⊕ h (T_i||IS_j||NC_i)

Check if

${MSG}_{1} \overset{?}{=} h ({CW}_{ij} | | {IDC}_{i} | | A 1_{i} | | {NC}_{i})$

If true, then generate timestamp TS₂

Then Calculate, NC_k = h (IS_j|| (k_j ⊕ SK_G))

MSG₂ = h (A1_i||NC_k||IS_j)

MSG₃ = A1_i ⊕ NC_i ⊕ NC_k

xrightarrow [(Gxrightarrow IS_j)] M₂ = {NC_i, MSG₂, MSG₃}

$Check if | {TS}_{2}^{'} - {TS}_{2} | < Δ T ?$

If so, compute

NC_k = h (XKey_j||IS_j)

A1_i = MSG₃ ⊕ NC_i ⊕ NC_k

$Checks if, h (A 1_{i} | | {NC}_{k} | | {IS}_{j}) \overset{?}{=} {MSG}_{2}$

If true, then generate timestamp TS₃ and a nonce NC_l

Then *calculate,

MSG₄ = h (A1_i||NC_i||IS_j)

MSG₅ = NC_l||MSG₄

             M₃ = {MSG₃, MSG₅}

             ← - - - - - - - - - - - - - -

              (P_i ← IS_j)

$Check if | {TS}_{3}^{'} - {TS}_{3} | < Δ T ?$

If so compute,

NC_k = MSG₃ ⊕ A1_i ⊕ NC_i

MSG₄ = h (A1_i||NC_i||ISj)

NC_l = MSG₄||MSG₅

Recompute

If true, Mutual authentication is success

session key ${SK}_{ij}^{} = h (A 1_{i} | | {NC}_{i} | | {NC}_{k} | | {NC}_{l} | | {IS}_{j}) = {SK}_{ij}$

The user(Doctor/Patient) inserts the smart card(SM_i) into a card reader. The user inputs the User ID and password. The bio-acoustic data is read with the help of sensors. Then, the user calculates, $m = rep ({BA}_{i}, {CW}_{i}), {AID}_{i} = h (U_{id} | | m | | pwd | | n), Q_{i}^{*} = h (U_{id} | | h (b) | | {APW}_{i} | | {CW}_{i})$ . The system checks whether $Q_{i}^{*}$ is equal to Q_i. If so, create two nonce NC_i, NC_j and generate a timestamp TS₁. Then, computes, CW_ij, T_i, A1_i, IDC_i and MSG₁ as given mentioned in the Table 4. The computed values are sent to the Gateway Node(G) via a public channel.

The Gatewaynode(G) receives the parameters and checks for the freshness of the message by calculating |TS1’-TS1| < ΔT, where TS1’ is the time at which the parameters were received by the Gateway and ΔT is the maximum time delay allowed between each message. If |TS1’-TS1| < ΔT is satisfied, then it computes the following, S_i = CW_ij ⊕ h (IS_j||h (b) ||NC_i), P_i = S_i ⊕ h (a||b), T_i = h (P_i||h (a||b), A1_i = AID_i ⊕ N_j ⊕ T_i and Checks if MSG₁ is matching with {h (CW_ij||ICD_i||A1_i||NC_i)}. If so, generates a timestamp TS₂ and computes the values for NC_k = h (IS_j|| (k_j ⊕ SK_G)) , MSG₂ & MSG₃. The gateway node sends the parameters NC_i, A1_i, MSG₂, MSG₃ to the sensor node IS_j.

Upon receiving the parameters from the gateway node, the sensing device checks the freshness of the message by computing |TS2’-TS2| < ΔT. If so, then compute NC_k = A1_i ⊕ NC_i ⊕ MSG₃ ⊕ h (XKey_j||IS_j). Checks if the received MSG₂ is equal to the computed one. If so, generate a timestamp TS₃. Computes, MSG₄ = h (T_i||A1_i||NC_i||IS_j). The IS_j sends the parameters MSG₄, IS_j to the User(P_i).

Upon receiving the parameters from the gateway node, the sensing device checks the freshness of the message by computing |TS3’-TS3| < ΔT. If so, then compute the value of MSG₄ with the parameters as follows h (T_i||A1_i||NC_i||IS_j). The user then verified, if the computed MSG₄ is matched with the received MSG₄. If the condition is satisfied, then the secure communication session between the P_i and the Sensing device IS_j is established with the secret key ${SK}_{ij}^{*}$ . Using this secret key ${SK}_{ij} = {SK}_{ij}^{*}$ , the User and the sensing device can now communicate securely.

4.5 Bio-acoustics based biometric password change phase

The user can update their credentials as the steps detailed in this phase. The user has to insert their smart card SM_i and enter their old credentials $pwd, {BA}_{i}^{'}$ . The credentials are validated with the secret parameters stored in the smart card as follows, The fuzzy embedder rep function is used to regenerate the secret key from the bioacoustics based biometric input.

$m = r e p (B {A^{'}}_{i}, C W_{i})$

AID_i = h (U_id||m||pwd||n)

$Q_{i}^{*} = h (U_{id} | | h (b) | | {AID}_{i} | | {CW}_{i})$

If the $Q_{i}^{*}$ is matched with the stored secret parameter Q_i in the smart card, then the user is considered as a valid and permitted to update their credentials.

User selects new secret parameters m_new, n_new and imprints the biometric input BA_{i
_new}. The fuzzy embedder function Embedd (.) generates a fresh secret key CW_{i
_new} = embed (BA_{i
_new}, m_new) and finally the smart card parameters AID_i, Q_i and R_i are updated with the recomputed $A I {D^{'}}_{i}, {Q^{'}}_{i}, {R^{'}}_{i}$ values as below.

$A I {D^{'}}_{i} = h (U_{i d} | | m_{n e w} | | p w d | | n_{n e w})$

${Q^{'}}_{i} = h (U_{i d} | | h (b) | | A I {D^{'}}_{i} | | C W_{i_{n e w}})$

${R^{'}}_{i} = R_{i} \oplus A I {D^{'}}_{i} \oplus A I D_{i}$

This way new keys are being generated for varying the user secret parameter for the same user every time during the biometric password updation phase. Thus, the key freshness/renewability is achieved in the proposed scheme using the fuzzy embedder technique, which prevents the biometric template attack.

4.6 Dynamic sensor node join phase

Connect In the dynamic nature of IoT network, the sensor devices are often join the network post deployment. Sooner the new sensor initiates the registration request to the G, the secret parameters $I {S^{'}}_{j}, s e {c^{'}}_{j}, {p^{'}}_{j}, X k e {y^{'}}_{j}$ are calculated as explained in the section 4.2. The G recomputes its secret key ${SK}^{'}_{G}$ with the new sensors added in the network. In the login & authentication phase, the user can able to access the newly added sensor ${IS}^{'}_{j}$ as the G’s secret keys is being renewed and shared using the Chinese remainder theorem along with secret sharing technique.

4.7 Dynamic sensor node revocation phase

As a Reverse of section 4.6, few sensor nodes may leave the network dynamically after the deployment. The G should protect the network from these sensor nodes capturing attacks after they moved out of network. The G recomputes its secret key ${SK}^{″}_{G}$ sooner a node is revoked. The computation happens as per the steps detailed in the section 4.2 with the active nodes of the system. As the G secret is updated with ${SK}^{″}_{G}$ , even if a user wants to access the redacted sensor, the session is not established as the MSG₂ check is failed in the authentication phase(refer the Table 4) with the old secret key SK_G of the gateway node.

5 Simulation results and security analysis discussion

In this section, The simulation of the proposed authentication scheme against the potential attacks of IoMT has been presented using the AVISPA simulator. The security of the authentication protocol is verified using Formal and Informal security analysis. In formal security verification, the designed algorithm is written using HLPSL (High Level Protocol Specification Language) for running the algorithm in the AVISPA tool. Informal security analysis is carried out by analyzing the algorithm theoretically against various security attacks.

5.1 Formal Security Verification using AVISPA Simulator

The AVISPA simulator is commonly used to simulate to verifying the security of the authentication protocols. The tool integrates four backends for processing the code. They are namely i) On-the-fly Model-Checker (OFMC), ii) Constraint Logic based Attack Searcher (CL-AtSe), iii) SAT-based Model-Checker (SATMC) and 4) Tree Automata based on Automatic Approximations for the Analysis of Security Protocols (TA4SP). These backends utilize different analysis techniques. The proposed protocol is first defined in HLPSL(High Level Protocol Specification Language). The code is then converted into Intermediate Format(IF) code by the HLPSL2IF translator. The IF code is run by one of the four backends and the result is then produced which displays whether the proposed protocol is secure or not.

5.1.1 The HLPSL implementation of the proposed scheme

The proposed system is simulated using Security Protocol ANimator (SPAN) for the AVISPA tool to imitate the RPM system of IoMT. The proposed scheme’s registration and authentication phases are implemented in HLPSL code. The scheme involves five fundamental roles. The User, Gateway, Sensor, Environment, Session, and Goal sections are written in HLPSL code. The HLPSL code for the user role in registration and authentication is shown in Fig. 4. Like this, Code for other roles in the system are implemented in HLPSL.

Fig. 4

HLPSL code for User role in Registration.

5.1.2 AVISPA simulation results and discussion

Figures 5,6,7 and 8 , Figures 5,6,7 and 8 shows the simulation results of the proposed scheme using CL-AtSe and OFMC backends. The simulation result shows that the protocol has been proved to be safe against the potential attacks in both backends of the AVISPA simulator.

Fig. 5

Result of User Registration using OFMC Backend.

Fig. 6

Result of User Registration using CL-ATSE Backend.

Fig. 7

Result of Login and Authentication using OFMC Backend.

Fig. 8

Result of Login and Authentication using CL-ATSE Backend

5.2 Informal security analysis

This section explores the proposed model’s functionality and security aspects using theoretically (non-mathematical) security investigation against various security attacks.

5.2.1 Privileged insider and stolen smartcard attack

We assume that a registered user P_i’s smart card has been misplaced or stolen in these attacks. During the user registration, P_i sends a registration request to the G by sending {AID_i, U_id, CW_i} where CW_i = embed (BA_i, m) , AID_i = h (U_id ∥ m ∥ pwd ∥ n) via the secure channel. In this example, we assume that an adversary $A$ knows the values of {AID_i, U_id, CW_i}. We have now assumed that $A$ is in possession of the stolen smart card SM_i of P_i after the registration process has been completed, and that using the power analysis attacks, he can able to extract the secret information’s ({n, CW_i, Q_i, R_i, S_i, h (.) , h (b)}) of SM_i. As AID_i uses a collision-resistant hash function h(.), it is computationally impossible to gain access to or guess pwd from AID_i without knowing the random numbers m and n. In other words, A will be unable to perform the identity guessing attack. Also, as P_i is generates a biometric password using the fuzzy embedder’s embed function, which takes a random secret and the U_i biometric imprint. It is computationally infeasible to get the U_i’s password even with the tampered biometric template, as it required the value of m. Thus the system is secure against privileged insider and stolen smartcards attacks.

5.2.2 Untraceability and user anonymity

In accordance with the threat model, we assume that $A$ detects and captures transmitted messages M₁, M₂ and M₃ through the public channel during the login & authentication phases. $A$ cannot compute the valid P_i’s identitiy U_id from these transmitted messages without knowing the secret parameters T_i, A1_i, NC_i and NC_k. Thus, the anonymity of the user is preserved. In addition, due to the fact that all messages are dynamic M_i (i = 1,2,3), which involves current timestamps and random nonce’s, it is computationally impossible to try and track the activities of the same user across multiple sessions. We are therefore able to preserve the untraceability property through the proposed scheme.

5.2.3 Replay attack

Let us consider that the adversary $A$ is able to trap the transmitted messages M₁, M₂ and M₃ and replays the messages in a while. In this protocol, the validity of the transmitted messages are verified by the timestamps TS₁, TS₂ and TS₃ attached with the messages in a small time interval ΔT, which is very small in the protocol implementations. So, even if $A$ manages a replay the messages after some time, they will not be validated as the timestamp will vary. Hence, the proposed system is secure against replay attacks.

5.2.4 Sensing device capturing attack

Let us imagine, based on the threat model presented, that $A$ is able to physically seize some IS_j and obtain the credentials stored on them. Therefore, from a captured device IS_j’s memory, $A$ can now have the parameters IS_j and XKey_j, where Xkey_j = h (IS_j||SK_G). The XKey_j is unique for all the IS_j available on the network. Because of the unique and separate credentials contained in the IS_j, $A$ can neither able to obtain G’s secret key nor be able to compute the session key between P_i and other non-compromised IoMT sensing devices using the recorded session key SK_ij in the captured IS_j’s memory. Even if $A$ captures a sensing device, the computation of session keys between P_i and other non-compromised sensing devices is unaffected. As a result, the suggested system is Resist to sensing device capture attacks.

5.2.5 Man-in-the-middle attack

Let us consider that the adversary $A$ is able to trap the transmitted messages M₁, M₂ and M₃. The $A$ can able to alter the transmitted messages to convince the G, IS_j and User P_i. If $A$ needs to modify the message M₁, $A$ must know the hidden parameters A1_i, T_i, m, NC_k and NC₁ to calculate MSG₁, MSG₂ and MSG₃. Computing the values of these secret parameters are infeasible, because they are computed using h(.) and fuzzy embedder techniques. Hence, if $A$ modified the messages M₁, M₂ and M₃, which are not processed in G, IS_j and U_i respectively. Thus, the proposed scheme is secure against man-in-the-middle attacks.

5.2.6 Bio-metric template attack

In a bio-metric template attack, the adversary $A$ tries to deceive the system by providing a false bio-metric template of a registered user. Since, the proposed scheme uses smartcard, password and bio-acoustic for authentication, it is difficult for A to reproduce or capture the smartcard as well as it is not feasible to guess the password. Let us assume that $A$ manages to capture the smartcard and guesses the password perfectly. Now, he needs the bio-acoustic of the U_i to login to the system. Bio-acoustic signals are unique for everyone and they are not easily reproducible. Hence, the proposed system is secure against bio-metric template attacks.

5.2.7 Impersonation attacks

Consider the scenario in which $A$ tries to impersonate a legitimate user, gateway or a sensor nodes. The $A$ needs to generate a valid login request to the system with all the necessary secret information of a U_i, G & IS_j, The sections 5.2.1, 5.2.3 and 5.2.6 are clearly explained that the proposed scheme is resist to all kind of impersonation attacks.

5.2.8 Mutual authentication

In this authentication system, the mutual authentication among P_i, G and IS_j is done by the following three scenarios: 1) G validates P_i by checking ${MSG}_{1} \overset{?}{=} h ({CW}_{ij} | | {IDC}_{i} | | A 1_{i} | | {NC}_{i})$ 2) The IS_j authenticates G directly by checking $h (A 1_{i} | | {NC}_{k} | | {IS}_{j}) \overset{?}{=} {MSG}_{2}$ received. and finally P_i validates IS_j by checking MSG₅ directly by compare the received value with the recomputed one , thus the mutual authentication is achieved using the session key SK_ij = h (A1_i||NC_i||NC_k||NC_l||IS_j).

5.2.9 Suitable for IoT applications

The typical IoT network, connects a of physical IoT devices such as sensors, actuators, gateway nodes and servers as shown in the Fig. 1. The sensor nodes are communicated to the servers via the Gateway nodes. The computational capacities and storage of the devices are differs. The sensor devices are very low in computational power and Servers has more. As these devices are resource constrained in nature, the cryptographic schemes designed for these devices must use only the lightweight operations like one-way hash functions and ex-OR. As the proposed scheme uses only the lightweight operations and follows the IoT network architecture. Thus, the proposed scheme is suitable for IoT applications like medical and Industrial IoT.

6 Comparative analysis

In this section, a complete analysis of the proposed method is given in terms of security features, communication, and computation cost in comparison with the existing authentication schemes. In this comparative study, only the fuzzy extractor, fuzzy commitment, and fuzzy embedder based authentication schemes are considered. The schemes chosen for the comparison are Vinoth et al. (2021)[42], Sudhakar et al. (2020)[46], Wu et al. (2021)[45], Rehman et al. (2021)[44], Abdi et al. (2021)[40], Zhang et al. (2022)[26] & Shao et al. (2022)[41].

6.1 Comparison of security and functionality features

In Table 6, the other existing schemes are compared with the proposed scheme in terms of Various security feature and functionality features. As given in the Table 6, the schemes of Sudhakar et al. (2020)[46] and Wu et al. (2021)[45] are not suitable for the IoT environment, as these schemes are failed to include the G in their proposal. The schemes (Vinoth et al. (2020)[42],Abdi et al. (2021)[40] & Shao et al. (2022)[41]) using fuzzy extractors are prone to biometric template attack and Impersonation attacks. Even though the Rehman et al. (2021)[44] and Zhang et al. (2022)[26] are secure against the various well known attacks, they are using the structural information of physiological characteristics as the biometric input. These structural information of physiological characteristics are not reliable and has the problem of being vulnerable to spoofing in future. Our proposed scheme is using frequency-domain based bioacoustic trait as teh biometric input, which is safe against all the attacks including biometric template attack.

Table 6
Security and Functionality Features Comparison

Schemes → Vinoth et al.[42] Sudhakar et al.[46] Wu et al.[45] Rehman et al.[44] Abdi et al.[40] Zhang et al.[26] Shao et al.[41] Proposed

Security Features ↓ (2021) (2020) (2021) (2021) (2021) (2022) (2022) Scheme

Resist to Privileged Insider Yes Yes Yes Yes Yes Yes Yes Yes

Resist to Replay Attack No Yes Yes Yes Yes Yes Yes Yes

Resist to Sensor Capturing Attack No N/A N/A Yes Yes Yes Yes Yes

Resist to Man-in-the-Middle Attack Yes Yes Yes Yes Yes Yes Yes Yes

Resist to Biometric Template Attack No Yes Yes Yes No Yes No Yes

Resist to Smart Card Stolen Attack Yes Yes Yes Yes Yes Yes Yes Yes

Resist to Impersonation Attack No Yes Yes Yes No Yes No Yes

Mutual Authentication Yes Yes Yes Yes Yes Yes Yes Yes

Untraceability and User Anonymity Yes Yes Yes Yes Yes Yes Yes Yes

Perfect Forward Secrecy Yes Yes Yes Yes Yes Yes Yes Yes

Biometric Technique Fuzzy Extractor Fuzzy Embedder Fuzzy Commitment Fuzzy Commitment Fuzzy Extractor Bio-Hash Fuzzy Extractor Fuzzy Embedder

Suitable for IoT Applications Yes No No Yes Yes Yes Yes Yes

Schemes →	Vinoth et al.[42]	Sudhakar et al.[46]	Wu et al.[45]	Rehman et al.[44]	Abdi et al.[40]	Zhang et al.[26]	Shao et al.[41]	Proposed
Resist to Privileged Insider	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes
Resist to Replay Attack	No	Yes	Yes	Yes	Yes	Yes	Yes	Yes
Resist to Sensor Capturing Attack	No	N/A	N/A	Yes	Yes	Yes	Yes	Yes
Resist to Man-in-the-Middle Attack	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes
Resist to Biometric Template Attack	No	Yes	Yes	Yes	No	Yes	No	Yes
Resist to Smart Card Stolen Attack	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes
Resist to Impersonation Attack	No	Yes	Yes	Yes	No	Yes	No	Yes
Mutual Authentication	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes
Untraceability and User Anonymity	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes
Perfect Forward Secrecy	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes
Biometric Technique	Fuzzy Extractor	Fuzzy Embedder	Fuzzy Commitment	Fuzzy Commitment	Fuzzy Extractor	Bio-Hash	Fuzzy Extractor	Fuzzy Embedder
Suitable for IoT Applications	Yes	No	No	Yes	Yes	Yes	Yes	Yes

6.2 Comparison of computation costs

The various notation and their experiment operational computing time[44, 52] as listed in Table 5, which are used as it’s to calculating the computation cost of the proposed scheme. The Table 7 provides the computation cost estimation for the proposed and all the comparable schemes. The proposed scheme’s computational time is calculated as ≈2.2428 ms, which is lesser than the Vinoth et al. (2021)[42], Wu et al. (2021)[45], Rehman et al. (2021)[44], Abdi et al. (2021)[40], Zhang et al. (2022)[26] & Shao et al. (2022)[41] schemes by 22.50%, 396.39%, 1.29%, 0.73%, 222.89% & 196.52% respectively. Even though the Sudhakar et al. (2020)[46] scheme’s computational time is 0.92% lesser then the proposed schemes, that scheme is designed for a single client server model, which is not suitable for IoT applications. As per the computation cost comparison wise the proposed scheme is the ideal candidate for the various IoT applications.

Table 5
Computational Cost Operations

Notations Operation Cost(in ms)

T _h One-way cryptographic hash ≈ 0.0026

T _e Block Cipher Encryption &Decryption ≈0.0046

T _fe Fuzzy Extractor ≈2.183

T _fc Fuzzy Commitment ≈2.204

T _fem Fuzzy Embedder ≈2.183

T _mm Modular multiplication ≈0.2568

T _m ECC point multiplication ≈2.226

T _bh BioHash ≈2.3891

Notations	Operation	Cost(in ms)
T _h	One-way cryptographic hash	≈ 0.0026
T _e	Block Cipher Encryption &Decryption	≈0.0046
T _fe	Fuzzy Extractor	≈2.183
T _fc	Fuzzy Commitment	≈2.204
T _fem	Fuzzy Embedder	≈2.183
T _mm	Modular multiplication	≈0.2568
T _m	ECC point multiplication	≈2.226
T _bh	BioHash	≈2.3891

Table 7

Comparison of Computation Cost

Schemes →	Vinoth et al.[42]	Sudhakar et al.[46]	Wu et al.[45]	Rehman et al.[44]	Abdi et al.[40]	Zhang et al.[26]	Shao et al.[41]	Proposed
Entity ↓/	(2021)/	(2020)	(2021)	(2021)	(2021)	(2022)	(2022)/	Scheme
User/	1T_fe + 1T_e	8T_h + 1T_fem	2T_fc + 1T_m	1T_fc + 1T_e	2T_e + 1T_fe	1T_bh + 2T_e	13T_h + 1T_fe	10T_h + 1T_fem
	+9T_h		+1T_e + 6T_h	+7T_h	+9T_h	+5T_h
Gateway	2T_mm + 3T_e	7T_h	2T_m + 3T_e	2T_e + 8T_h/	1T_e + 10T_h	1T_bh + 6T_e	18T_h + 1T_fe	9T_h
	+6T_h		+5T_h			+4T_h
Sensing	2T_e + 4T_h	-	-	1T_e + 4T_h	5T_h	1T_bh + 2T_e	8T_h + 1T_fe	4T_h
device/						+2T_h
Total	1T_fe + 2T_mm	15T_h + 1T_fem	2T_fc + 3T_m	1T_fc + 4T_e	1T_fe + 3T_e	3T_bh + 10T_e	39T_h + 3T_fe	23T_h + 1T_fem
Operations	+6T_e + 9T_h		+4T_e + 11T_h	+19T_h	+24T_h	+11T_h/
Total Cost	≈2.7476	≈2.222	≈11.133/	≈2.2718	≈2.2592	≈7.2419	≈6.6504	≈2.2428
(in ms)/

6.3 Comparison of communication costs

The suggested scheme’s efficiency is examined in terms of the communication expenses associated with sending messages during the login and authentication stage, as the registration is one time activity, its expense is not considered for the comparison. The bit sizes of the random number, identity, timestamp, Symmetric Encryption/Decryption points, and hash output (if we apply h(.)) are 160, 160, 32, 320, 320, 160 bits, respectively, with the presumption. Furthermore, the bit values utilized for modular exponentiation and inversion operations are 1024 bits because the 160-bit ECC gives the same level of security as the 1024-bit RSA security. The Table 8 summarizes the comparative result on communication costs during the login and authentication phases. The proposed scheme consumes the number of bits in three messages are, in message 1 = (160 + 160 + 160 + 160 + 32) = 672 bits, in message 2 = (160 + 160 + 160) = 480 bits and in message 3 = (160 + 160) = 320 bits, respectively. The total communication cost of the proposed scheme is (672 + 480 + 320) = 1472 bits.

Table 8
Comparison of Communication Cost

Schemes → Vinoth et al.[42] Sudhakar et al.[46] Wu et al.[45] Rehman et al.[44] Abdi et al.[40] Zhang et al.[26] Shao et al.[41] Proposed

Messages &Cost ↓ (2021) (2020) (2021) (2021) (2021) (2022) (2022) Scheme

No of Messages 7 4 3 4 4 3 7 3

Total Cost in bits 3040 1800 1088 1804 2816 1220 5360 1472

Schemes →	Vinoth et al.[42]	Sudhakar et al.[46]	Wu et al.[45]	Rehman et al.[44]	Abdi et al.[40]	Zhang et al.[26]	Shao et al.[41]	Proposed
No of Messages	7	4	3	4	4	3	7	3
Total Cost in bits	3040	1800	1088	1804	2816	1220	5360	1472

As shown in Table 8, the proposed scheme consumes lesser bits for the communication than the Vinoth et al. (2021)[42], Rehman et al. (2021)[44], Abdi et al. (2021)[40] & Shao et al. (2022)[41] schemes.

The schemes Sudhakar et al. (2020)[46] and Wu et al. (2021)[45] are failed to implement the G. So, achieved the lesser communication cost, which are not suitable for IoT applications. The Zhang et al. (2022)[26] scheme achieves 17.11% lesser in communication cost than the proposed scheme but consumes 222.89% higher in computation cost wise. As a summary, the proposed scheme is the ideal scheme with respect to Communication cost comparison.

7 Conclusions and future works

In this work, a lightweight bioacoustic based authentication scheme has been proposed for securing the IoMT applications with the fuzzy embedder technique. The proposed Fuzzy-embedder based user authentication scheme security has been formally verified using the AVISPA simulator and the results confirms the proposed scheme safeness. Additionally, the security and functionality features of the scheme has been compared with the prevailing schemes and affirmed that the proposed scheme withstands well-known security attacks and features. Finally, the proposed bioacoustic scheme performance has been compared with prevailing schemes in terms of computation and communication costs and found that the proposed scheme has outperformed.

In future, the proposed scheme can be adopted to the multimodal authentication systems for e-healthcare. It is also possible to expand it further and extend its use to other areas of the Internet of Things, such as the industrial Internet of Things, the Internet of Vehicles, the Internet of Drones, Smart Farming, and many other sectors for the secure authentication. In the future research direction, the frequency domain based biometrics can be fused with physical biometric traits to authenticate a user.

References

Goel

, Sharma

D.K.

and Gupta

K.D.

, LEOBAT: Lightweight encryptionand OTP based authentication technique for securing IoT networks, Expert Systems 39(5) (2022), e12788. doi: 10.1111/exsy.12788.

Patient Portal - eClinicalWorks. https://www.eclinicalworks.com/products-services/patient-engagement/patient-portal/.

Manage My Health NZ - Take Charge of Your Medical Records. https://www.managemyhealth.co.nz/.

Boynton Health. https://boynton.umn.edu/.

National Health Portal of India, Gateway to Authentic Health Information. https://www.nhp.gov.in/.

Jacob

T.P.

, Pravin

and Kumar

R.R.

, A secure IoT based healthcare framework using modified RSA algorithm using an artificial hummingbird based CNN, Transactions on Emerging Telecommunications Technologies (2022), e4622. doi: 10.1002/ett.4622.

Sudhakar

, Ramalingam

and Jagatheswari

, An improvedproxy-vehicle based authentication scheme for vehicular adhocnetworks, International Journal of Information Technology 14 (2022), 2441–2449. doi:10.1007/s41870-022-00938-5.

Jagatheswari

, Ramalingam

and Chandra Priya

, Improved greyrelational analysis-based TOPSIS method for cooperation enforcingscheme to guarantee quality of service in MANETs, International Journal of Information Technology 14(2) (2022), 887–897. doi:10.1007/s41870-022-00865-5.

Saad

, Shalaby

and Mohamed

A.A.

, A lightweight access management scheme for the internet of things based smart applications, Transactions on Emerging Telecommunications Technologies (2022), e4600. doi:10.1002/ett.4600.

10.

, Janakiraman

, Ramalingam

and An

M.D.

, improved ensemble classification-based secure two stage bagging pruning technique for guaranteeing privacy preservation of DNA sequences in electronic health records, Journal of Intelligent & Fuzzy Systems Preprint 44(1) (2023), 149–166. doi:10.3233/JIFS-221615.

11.

Sim

J.Y.

, Noh

H.W.

, Goo

, Kim

, Chae

S.-H.

and AhnIdentity

C.-G.

, Recognition Based on Bioacoustics of Human Body, IEEETransactions on Cybernetics 51(5) (2021), 2761–2772. doi:10.1109/TCYB.2019.2941281.

12.

Lee

, Choi

and Lee

D.H.

, Usable user authentication on a smartwatch using vibration, in: Proceedings of the 2021 ACM SIGSAC Conference on Computer and Communications Security, (2021), pp. 304–319.

13.

Wang

and Wang

, Two Birds with One Stone: Two-FactorAuthentication with Security Beyond Conventional Bound, IEEETransactions on Dependable and Secure Computing 15(4) (2018), 708–722. doi:10.1109/TDSC.2016.2605087.

14.

Zhang

, Han

, Li

, Zhang

, Li

and Zhang

, MagAuth: Secure and Usable Two-Factor Authentication with Magnetic Wrist Wearables, IEEE Transactions on Mobile Computing (2021), 1–1. doi:10.1109/TMC.2021.3072598.

15.

Cao

, Li

, Zhang

, Yang

and Wang

, Towards Nonintrusive and Secure MobileTwo-Factor Authentication on Wearables, IEEE Transactions on Mobile Computing (2021), 1–1. doi:10.1109/TMC.2021.3133275.

16.

Wang

, Wang

, Cheng

and He

, Quantum2FA: Efficient Quantum-Resistant Two-Factor Authentication Scheme for Mobile Devices, IEEE Transactions on Dependable and Secure Computing (2021), 1–1. doi:10.1109/TDSC.2021.3129512.

17.

Jiang

, Zhang

, Ni

, Ma

and Choo

K.-K.R.

, Unified Biometric Privacy Preserving Three-Factor Authentication and Key Agreement for Cloud-Assisted Autonomous Vehicles, IEEE Transactions on Vehicular Technology 69(9) (2020), 9390–9401. doi:10.1109/TVT.2020.2971254.

18.

Hsu

C.-L.

and Le

T.-V.

, A time bound dynamic group key distribution scheme with anonymous three-factor identification for IoT-based multi-server environments, in: 2020 15th Asia Joint Conference on Information Security (AsiaJCIS), IEEE, (2020), pp. 59–65. doi:10.1109/AsiaJCIS50894.2020.00021.

19.

El-Mesady

, Hamed

and Abualnaja

K.M.

, A novel application onmutually orthogonal graph squares and graph orthogonal arrays, AIMS Mathematics 7(5) (2022), 7349–7373. doi:10.3934/math.2022410.

20.

El-Shanawany

and El-Mesady

, Mutually orthogonal graph squares for disjoint unions of stars, Ars Combinatoria 149(2020), 83–91.

21.

Higazy

, El-Mesady

and Mohamed

M.S.

, On graph orthogonal arrays by mutually orthogonal graph squares, Symmetry 12(11) (2020), 1895. doi:10.3390/sym12111895.

22.

El-Mesady

, Bazighifan

, et al., Construction of Mutually Orthogonal Graph Squares Using Novel Product Techniques, Journal of Mathematics 2022 (2022). doi:10.1155/2022/9722983.

23.

El-Mesady

and Shaaban

S.M.

, Generalization of Mac Neish’s Kronecker product theorem of mutually orthogonal Latin squares, AKCE International Journal of Graphs and Combinatorics 18(2) (2021), 117–122. doi:10.1080/09728600.2021.1966349.

24.

El-Mesady

, Bazighifan

and Shabana

, On Graph-Transversal Designs and Graph-Authentication Codes Based on Mutually Orthogonal Graph Squares, Journal of Mathematics 2022 (2022). doi:10.1155/2022/8992934.

25.

Qiu

, Wang

, Xu

and Kumari

, Practical and Provably SecureThree-Factor Authentication Protocol Based on Extended Chaotic-Maps for Mobile Light weight Devices, IEEE Transactions on Dependable and Secure Computing 19(2) (2022), 1338–1351. doi:10.1109/TDSC.2020.3022797.

26.

Zhang

, Zhu

, Ren

, Zhang

and Choo

K.-K.R.

, Privacy-Preserving Fast Authentication andKey Agreement for E-health Systems in IoT, based on Three-Factor Authentication, IEEE Transactions on Services Computing (2022), 1–1. doi:10.1109/TSC.2022.3149940.

27.

Sahoo

S.S.

, Mohanty

and Majhi

, A secure three factor basedauthentication scheme for health care systems using IoT enableddevices, Journal of Ambient Intelligence and HumanizedComputing 12(1) (2021), 1419–1434. doi:10.1007/s12652-020-02213-6.

28.

Zhang

, Xiao

, Sun

and Ma

, Efficient Multi-FactorAuthenticated Key Exchange Scheme for Mobile Communications, IEEE Transactions on Dependable and Secure Computing 16(4) (2019), 625–634. doi:10.1109/TDSC.2017.2700305.

29.

Kora'c

and Simi'c

, Fishbone model and universal authenticationframework for evaluation of multifactor authentication in mobileenvironment, Computers & Security 85 (2019), 313–332. doi:10.1016/j.cose.2019.05.011.

30.

Zhu

, Qu

, Xu

, Zhang

, Shao

, Chen

, Prabhakar

and Yang

, RiskCog: Unobtrusive Real-Time User Authentication on Mobile Devices in the Wild, IEEE Transactions on Mobile Computing 19(2) (2020), 466–483. doi:10.1109/TMC.2019.2892440.

31.

Abbasinezhad-Mood

, Ostad-Sharif

, Mazinani

S.M.

and Nikooghadam

, Provably Secure Escrow-Less Chebyshev Chaotic Map-Based Key Agreement Protocol for Vehicle to Grid Connections With Privacy Protection, IEEE Transactions on Industrial Informatics 16(12) (2020), 7287–7294. doi:10.1109/TII.2020.2974258.

32.

Dharminder

, Kumar

and Gupta

, A construction of a conformal Chebyshev chaotic map based authentication protocol for health care telemedicine services, Complex & Intelligent Systems 7(5) (2021), 2531–2542. doi:10.1007/s40747-021-00441-7.

33.

, Cheng

, Wang

and Liang

, Practical ThresholdMulti-Factor Authentication, IEEE Transactions on InformationForensics and Security 16 (2021), 3573–3588. doi:10.1109/TIFS.2021.3081263.

34.

Kumar

, Saha

, Conti

, Kumar

, Buchanan

W.J.

and Kim

T.H.

, A comprehensive survey of authentication methods in Internet-of-Things and its conjunctions, Journal of Network and Computer Applications (2022), 103414. doi:10.1016/j.jnca.2022.103414.

35.

Das

A.K.

, Afroz

, Chandra

and Sinha

, MACP: Multifactor Authentication using Physical Unclonable Function and Fuzzy Extractor based Chebyshev Polynomial for Industrial Internet of Things devices, Transactions on Emerging Telecommunications Technologies (2022), e4581. doi:10.1002/ett.4581.

36.

Vhaduri

and Poellabauer

, Multi-Modal Biometric-Based Implicit Authentication of Wearable Device Users, IEEE Transactions on Information Forensics and Security 14(12) (2019), 3116–3125. doi:10.1109/TIFS.2019.2911170.

37.

Labayen

, Vea

, Florez

, Aginako

and Sierra

, Online Student Authentication and Proctoring System Based on Multimodal Biometrics Technology, IEEE Access 9 (2021), 72398–72411. doi:10.1109/ACCESS.2021.3079375.

38.

Evangelin

L.N.

and Fred

A.L.

, Securing recognized multimodal biometric images using cryptographic model, Multimedia Toolsand Applications 80(12) (2021), 18735–18752. doi:10.1007/s11042-021-10541-8.

39.

El-Rahiem

B.A.

, El-Samie

F.E.A.

and Amin

, Multimodal biometric authentication based on deep fusion of electrocardiogram (ECG) and finger vein, Multimedia Systems 28(4) (2022), 1325–1337. doi:10.1007/s00530-021-00810-9.

40.

Abdi Nasib Far

, Bayat

, Kumar Das

, Fotouhi

, Pournaghi

S.M.

and Doostari

M.-A.

, LAPTAS: lightweight anonymousprivacy-preserving three-factor authentication scheme for WSN-basedIIoT, Wireless Networks 27(2) (2021), 1389–1412. doi:10.1007/s11276-020-02523-9.

41.

Shao

, Guo

and Guo

, A PUF-based anonymous authentication protocol for wireless medical sensor networks, Wireless Networks (2022), 1–18. doi:10.1007/s11276-022-03070-1.

42.

Vinoth

, Deborah

L.J.

, Vijayakumar

and Kumar

, Secure Multifactor Authenticated Key Agreement Scheme for Industrial IoT, IEEE Internet of Things Journal 8(5) (2021), 3801–3811. doi:10.1109/JIOT.2020.3024703.

43.

Ramasamy

, Srirangan

and Ramalingam

, Fuzzy and position particle swarm optimized routing in VANET, International Journal of Electrical and Computer Engineering Systems 12(4) (2021), 199–206. doi:10.32985/ijeces.12.4.3.

44.

Rehman

H.U.

, Ghani

, Chaudhry

S.A.

, Alsharif

M.H.

and Nabipour

, A secure and improved multi server authentication protocol usingfuzzy commitment, Multimedia Tools and Applications 80(11) (2021), 16907–16931. doi:10.1007/s11042-020-09078-z.

45.

, Meng

, Zhao

, Wei

and Wang

, Privacy-Preserving Cancelable Biometric Authentication Based on RDM and ECC, IEEE Access 9 (2021), 90989–91000. doi:10.1109/ACCESS.2021.3092018.

46.

Sudhakar

and Natarajan

, A new three-factor authentication andkey agreement protocol for multi-server environment, Wireless Networks 26(7) (2020), 4909–4920. doi:10.1007/s11276-018-01922-3.

47.

Fuller

, Reyzin

and Smith

, When Are Fuzzy Extractors Possible? IEEE Transactions on Information Theory 66(8) (2020), 5282–5298. doi:10.1109/TIT.2020.2984751.

48.

Shamshad

, Ayub

M.F.

, Mahmood

, Rana

, Shafiq

and Rodrigues

J.J.P.C.

, An Identity-Based Authentication Protocol for the Telecare Medical Information System (TMIS) Using a Physically Unclonable Function, IEEE Systems Journal 16(3) (2022), 4831–4838. doi:10.1109/JSYST.2021.3118014.

49.

Rashid

, Imran

, Jafri

A.R.

and Al-Somani

T.F.

, Flexible architectures for cryptographic algorithms—A systematicliterature review, Journal of Circuits, Systems and Computers 28(03) (2019), 1930003. doi:10.1142/S0218126619300034.

50.

Imran

, Rashid

, Raza Jafri

and Najam-ul-Islam

, A Cryp-Proc: Flexible Asymmetric Crypto Processor for Point Multiplication, IEEE Access 6 (2018), 22778–22793. doi:10.1109/ACCESS.2018.2828319.

51.

Noh

H.W.

, Ahn

C.-G.

, Chae

S.-H.

, Ku

and Sim

J.Y.

, Multichannel Acoustic Spectroscopy of the Human Body for Inviolable Biometric Authentication, Biosensors 12(9) (2022), 700. doi:10.3390/bios12090700.

52.

Kilinc

H.H.

and Yanik

, A Survey of SIP Authentication and Key Agreement Schemes, IEEE Communications Surveys & Tutorials 16(2) (2014), 1005–1023. doi:10.1109/SURV.2013.091513.00050.

53.

Praveen

and Pabitha

, Improved Gentry–Halevi’s fully homomorphic encryption-based lightweight privacy preserving scheme for securing medical Internet of Things, Transactions on Emerging Telecommunications Technologies (2023), e4732. doi:10.1002/ett.4732.

User P_i with SM_i =	Gateway Node (G)	IoT Sensing Sevice IS_j
{n, CW_i, Q_i, R_i, S_i, h (.) , h (b)}
Insert SM_i into a card reader .
$Input U_{id} pwd / & {BA}_{i}^{'} is read$
Computes

AID_i = h (U_id\|\|m\|\|pwd\|\|n)
$Q_{i}^{*} = h (U_{id} \| \| h (b) \| \| {AID}_{i} \| \| {CW}_{i})$
$Checks Q_{i}^{*} \overset{?}{=} Q_{i}$
If so, create two nonce NC_i NC_j
Generate timestamp TS₁
Calculate,
CW_ij = S_i ⊕ h (IS_j\|\|h (b) \|\|NC_i)
T_i = R_i ⊕ AID_i
A1_i = AID_i ⊕ N_j ⊕ T_i
IDC_i = A1_i ⊕ h (T_i\|\|IS_j\|\|NC_i)
MSG₁ = h (CW_ij\|\|IDC_i\|\|A1_i\|\|NC_i)

xrightarrow [(P_ixrightarrow G)] M₁ = {CW_ij, IDC_i, MSG₁, NC_i, TS₁}
	$Check if \| {TS}_{1}^{'} - {TS}_{1} \| < Δ T ?$
	If so, compute
	S_i = CW_ij ⊕ h (IS_j\|\|h (b) \|\|NC_i)
	P_i = S_i ⊕ h (a\|\|b)
	T_i = h (P_i\|\|h (a\|\|b))
	A1_i = IDC_i ⊕ h (T_i\|\|IS_j\|\|NC_i)
	Check if
	${MSG}_{1} \overset{?}{=} h ({CW}_{ij} \| \| {IDC}_{i} \| \| A 1_{i} \| \| {NC}_{i})$
	If true, then generate timestamp TS₂
	Then Calculate, NC_k = h (IS_j\|\| (k_j ⊕ SK_G))
	MSG₂ = h (A1_i\|\|NC_k\|\|IS_j)
	MSG₃ = A1_i ⊕ NC_i ⊕ NC_k

	xrightarrow [(Gxrightarrow IS_j)] M₂ = {NC_i, MSG₂, MSG₃}
		$Check if \| {TS}_{2}^{'} - {TS}_{2} \| < Δ T ?$
		If so, compute
		NC_k = h (XKey_j\|\|IS_j)
		A1_i = MSG₃ ⊕ NC_i ⊕ NC_k
		$Checks if, h (A 1_{i} \| \| {NC}_{k} \| \| {IS}_{j}) \overset{?}{=} {MSG}_{2}$
		If true, then generate timestamp TS₃ and a nonce NC_l
		Then calculate,
		MSG₄ = h (A1_i\|\|NC_i\|\|IS_j)
		MSG₅ = NC_l\|\|MSG₄

		M₃ = {MSG₃, MSG₅}
		← - - - - - - - - - - - - - -
		(P_i ← IS_j)
$Check if \| {TS}_{3}^{'} - {TS}_{3} \| < Δ T ?$
If so compute,
NC_k = MSG₃ ⊕ A1_i ⊕ NC_i
MSG₄ = h (A1_i\|\|NC_i\|\|ISj)
NC_l = MSG₄\|\|MSG₅
Recompute

If true, Mutual authentication is success
session key ${SK}_{ij}^{*} = h (A 1_{i} \| \| {NC}_{i} \| \| {NC}_{k} \| \| {NC}_{l} \| \| {IS}_{j}) = {SK}_{ij}$