Desktop version

Home arrow Computer Science

  • Increase font
  • Decrease font

<<   CONTENTS   >>

Privacy and Security

Any system connected to Internet and Cloud must take adequate measures for ensuring security and privacy. The Cloud-enabled BAN is also no exception. Rather, privacy and security is of paramount importance in Cloud-enabled BAN. This is because any compromise of the PHI parameters, MSP parameters, and location of MU and MSP will lead to failure of the system.

Security Notions in Cloud enabled BAN

Cloud-Enabled BAN must satisfy some of the following security requirements:

I. Data Privacy: The PHI data of MU is very sensitive. MU does not want her

personal PHI data that depicts her physical status to be known outside. Neither attacker nor Cloud should be able to get access to this raw data. However, those data are encrypted and then processed by Cloud and MSP because any compromise or alteration of MU PHI data can lead to fatal consequences such as wrong medical treatment or false medical emergency alarm generation, etc. Similarly, MSP also does not want to disclose its infrastructural parameter details to the outside world. However, Cloud can access MSP data in encrypted form for processing.

II. Identity Privacy or Anonymity: Neither MU nor MSP want to disclose their

real identity to a third party such as attacker or semi-trusted Cloud. If real identity of MU gets disclosed then that can be life threatening to those who are very important persons or someone in society whose physical status on diseases can lead to loss of market share, brand values, etc. Similarly, leakage of identity of MSP can be exploited by its rival MSPs.

III. Location Privacy: Location of MU is very important parameter that MU does

not want to share with others. Location information of MU can be exploited by others if disclosed by entities like Cloud.

IV. Authentication: Since multiple entities like MU, MSP, and Cloud exchange

data, their source must be authenticated properly by the system.

V. Unlinkability: Adversarial agents must not be able to link two messages generated from the same BAN.

VI. Access Control: As different types of users might be there who process

the PHI data, there should be differential access rights to process PHI data. For example, the emergency medical technicians should access only the data during a medical emergency whereas the assigned doctors must access past records to find any correlation that can help in proper medical diagnosis.

Attacks and Threats in Cloudenabled BAN

The attacks and threats are generally classified into passive and active. Passive attack does not modify the transmitted data whereas an active attack modifies the intercepted messages. Following are some common attacks on BANs:

I. Eavesdropping attack: Since BAN communication takes place wirelessly, it is

not difficult for an adversary to listen to the data communicated through an insecure channel. The goal of this passive attack is to know the content of the communication by collecting a large amount of exchanged messages. It is very important to stop this, as PHI data is very personal. To prevent eavesdropping, PHI data has to be encrypted and the keys used must be updated regularly.

II. Impersonation attack: This attack enables an attacker to pose itself as an

authorized user of the network. The attacker gains private identity information and encases it to impersonate a legitimate user in the network. As a result, the attacker obtains the same rights as of the legitimate user. Authentication mechanism needs to be adopted by different entities in BAN to protect from this kind of attack.

III. Replay attack: An attacker captures a message sent by user and resends it at

some later time. If appropriate countermeasures are not present, this resent message is accepted by the receiver as a valid message. The attacker can achieve some malicious intent. Since the receiver processes the message, receiver can be misled by the message content. Also sensor energy gets reduced if plenty of non-legitimate replayed messages are processed by the legitimate user. To prevent this attack, the BAN communications should be secured in such a way that it stops old messages from being sent again at a later time. For achieving this, authentication and message freshness are to be ensured in BAN. This is often done by using timestamps or nonces.

IV. Man in the Middle attack: In this attack, the attacker intercepts messages

being sent by sender, modifies the messages and then relays the messages to the receiver. Sender and receiver believes that they are communicating with each other directly. This powerful attack allows the attacker to eavesdrop and manipulate the transmitted message in real time.

V. Denial of Service Attack: Denial of Service (DoS) is an attack that targets to

disrupt the availability of certain resources or services. It is made possible by flooding the target with a huge number of fake messages until the target entity is no longer able to process them all. Besides degradation of performance services, energy of BAN nodes is drained heavily by this type of attack. Defending against this attack is done mainly by trying to identify legitimate messages ignoring the suspicious messages.

Existing Security and Privacy Solutions in Cloud enabled BAN

This section presents different solutions proposed in the literature to ensure privacy and security in Cloud-enabled BAN. [223] proposed a privacy-preserving emergency notification scheme to the nearest doctor through mobile healthcare social network. When medical emergency occurs, the personal digital assistant (PDA) of MU accumulates the emergency data such as PHI record, MU health record, and MU location. It then broadcasts this emergency call along with emergency data through mobile healthcare social network. Thus the message is communicated to the nearest doctor. It has adopted attribute-based encryption (ABE) scheme to make sure about proper access-control of emergency data. It has also discussed the revocation of the access control for maintaining safety from insider attack.

The work proposed in [345] considered separation of access rights for the health records between emergency medical technician (EMT) and primary physician. Primary physician is fully trusted, whereas EMT is considered as honest-but- curious. Remote server is considered as semi-trusted. The EMT, remote server, and credential authority may collude to compromise patient's privacy. This work allows EMT to get access to only those data that are required for addressing the medical emergency, whereas primary physician is allowed to get unrestricted access to all the past health records of MU. The different security privacy concerns that have been addressed in this work are authentication, anonymity, location privacy and unlinkability of multiple medical data of same MU but sent at different time. It has used cryptographic primitives such as bilinear pairing, commitment scheme, proof of knowledge, anonymous credentials, and pseudorandom number generator (PRNG) for achieving the privacy and access control.

[91] has developed an event-aided packet forwarding protocol that enables patients to communicate with each other in mobile healthcare social network (MHSN), built upon BAN and mobile communication platform, whenever illness related events or activities occur. They have used predicate encryption to provide patient identity privacy, patient illness privacy and message confidentiality.

Another work considered privacy-preserving and secret sharing of data and collaboration scheme in MHSN of smart cities by [172]. They have used identity- based broadcast encryption and ABE scheme for secure and privacy-preserving social and health data sharing, respectively. Further, to provide a secure data collaboration from independent cloud servers, they have used proxy re-encryption (PRE) scheme. As a result, the authorized health data analysers can access reencrypted data.

Tackling of infectious disease in smart city has been explored in [396], which demands fusion of health cloud data and social network data analysis. This is a promising area, which has not been explored much up until this date. [236] has considered a different scenario in MHSN, where a patient can share its data to other members who also have the same symptoms. They have proposed a secure same symptom based handshaking scheme using bilinear pairing.

After sensing of PHI data by sensors in the BAN, the data read by the sensors may be noisy due to reasons such as lossy human body medium and uncertainty of MU motion. To remove these noise, data has to be filtered. [216] has proposed the unscented Kalman filter based emergency fusion model to increase the data accuracy when emergency occurs.

So far all the works we have mentioned focussed on content privacy. Content privacy is all about protecting the MU information privacy. Whereas [73] has considered both content privacy and contextual privacy. Context privacy is about the communication context and it has four parts: sender/receiver anonymity, un-linkability, un-observability and pseudonymity. It has proposed contextual privacy based on the concept of onion routing, fake message injection and multicast, whereas it has used identity based cryptography to provide content privacy.

Another work by [96] focussed on attending of MU in emergency by nearest available medical personnel. During medical emergency, when no one is around to help, their system enables the nearest medical personnel to be alerted so that the MU is attended at the earliest. The Cloud, aware of the emergency event, grants necessary access to medical data of the patient to the medical personnel heading towards the MU in emergency. This system has provision for cancel phase in case someone nearby arrives to help the MU.

A lightweight encryption algorithm based on SHA-3 has been proposed by [385] for secure communication between BAN and server. Apart from keeping the data in servers safe, it employs Sharemind, a framework easy for use by non-crypto users. Sharemind splits the patient data into three numbers such that its sum equals the original data. As an improvement to this work, [363] supports any number of participants and is secure till half of the participants are not compromised. This proposal uses another lightweight multi-party computation protocol based on FairplayMP framework. This scheme has an advantage that a sensor needs to keep just one key to communicate with n number of different servers.

Since the BAN nodes operate in lossy medium, it is very natural that some data readings will be faulty. [190] has attempted to identify the data faults and reconstruct them in pervasive healthcare. It considers that there exists correlation in multiple attributes among different sensor nodes in the MU. Any activity is reflected in the change of readings of at least two sensors. Here every node maintains a trust rating for other nodes using Cosine similarity.

Attribute based encryption has been used by many works for providing privacy and security in BAN. ABE has the advantage that it is capable of providing differential access rights for different stake holders in BAN-centric system. There are two flavours of ABE: Key-Policy ABE (KP-ABE) and Ciphertext-Policy ABE (CP- ABE). [352] studied the suitability of these two approaches for application in BAN and found that KP-ABE is more preferable.

The problem of data privacy in Cloud-assisted healthcare systems has been reviewed by [320]. An outstanding work on Cloud-assisted privacy preserving mobile health monitoring is by [225]. It has used outsourcing decryption technique and multi-dimensional range query to shift decryption complexity from

MU to Cloud. They have further used key private proxy re-encryption to reduce computation complexity of service providers. A systematic literature review on distributed denial of service attack in Cloud-assisted BAN has been carried out by [212]. [360] has proposed a model for Cloud-assisted mobile-access of health data where both private Cloud and public Cloud have been used. Data processing and analysis tasks are done at private Cloud and processed results are stored in public Cloud.

Privacy-preserving priority based data aggregation for Cloud-assisted BAN has been discussed in [395]. The health data is categorized into different types and accordingly assigned different priorities. Then separate forwarding strategies are chosen as per the data priority. They have used bilinear pairing and Paillier cryptosystem for achieving privacy-preserving data aggregation.

Authentication in BAN

Authentication is one of the desirable properties in secure systems. As communication is mostly wireless in Cloud-enabled BAN, the source authentication is very important. [217] has proposed a secure authentication scheme using Chebyshev chaotic maps for Cloud-assisted BAN. In this work, the MSP to treat MU is pre-decided by MU herself.

[378] proposed a revocable certificateless encryption scheme and revocable certificateless signature scheme to provide certificateless remote anonymous authentication protocol. The salient feature is that the revocation scheme is scalable. [151] proposed a distributed attribute-based authentication system, where MU/MSP use their verifiable attributes for authenticating each other. It also provides privacy protection and verifiability of MU/MSP attributes.

[229] has proposed an anonymous authentication protocol that used an anonymous account index instead of actual identity of MU to access BAN service, thereby preventing the potential privacy leakage. However, [377] and [161 ] proved that this scheme is insecure against public key replacement attack, and impersonation attack. [377] has proposed a lightweight certificateless scalable and remote anonymous authentication protocol by using the certificateless encryption and two-party authenticated key agreement protocol. An excellent work on anonymous authentication in BAN is by [161]. They have proved that their anonymous authentication scheme is mutual authentication-secure based on the hardness of Diffie Heilman problem.

Key Management in BAN

Since the data in BANs are transmitted through wireless medium, the communication has to be secured, to prevent eavesdropping and distortion PHI data. For this, a cryptographic scheme needs to be employed, and that requires secret keys. That is why it is necessary to have a secure key agreement and distribution of them amongst the sensor nodes in the BAN. As sensors in BANs have very limited resources, it is important for the key agreement scheme to be lightweight. We can classify the existing key agreement schemes into four types [196], [33]:

I. Traditional key agreement schemes

II. Physiological value-based key agreement schemes

III. Hybrid key agreement schemes

IV. Secret key generation schemes

The traditional key agreement schemes are generally based on public-key cryptography or some kind of key pre-distribution techniques. This requires more key storage space, but reduces the processing effort. The second type of key agreement schemes are physiological value-based key agreement schemes. These use humans' vital physiological characteristics such as ECG signal and then transform these values into a secret key by adopting some appropriate artificial intelligence technique. Since the sender and the receiver are in the same human body, they generate the same key. This removes the need for key pre-distribution and explicit key distribution. The third kind of scheme is the hybrid scheme, which uses both of the previous kinds of key agreement approaches. For the key generation and agreement, they often use key pre-distribution and physiological values from the human body. For an introduction to key pre-distribution, one can refer to [302]. Lastly, the secret key generation approach is mainly based on the characteristics of communication channel and physiological data from the user. These signals are generally used for independent key generation of a key, while physiological signals are used for key exchange. [119] has proposed a key management scheme for group communication in BAN by using physical unclonable function (PUF). Group communication in BAN is an area which has not been explored much.


In this chapter, we have discussed the Cloud-enabled Body Area Network technology. This is a promising area of research with so many critical issues involved. We have attempted to introduce the sensor nodes, bio-sensors, and the body area network gradually. The architecture of BAN and Cloud-enabled BAN has been presented. Although there are many applications of BAN in many path-breaking domains, we have focussed our discussions on mainly application of BAN in healthcare, particularly in continuous healthcare monitoring. Since BAN enabled medical user's physiological data and location information is extremely private information, any BAN-based system will not succeed if proper security and privacy measures are not taken. This privacy and security issue is more relevant in the presence of Cloud. Cloud-enabled BAN demands that Cloud will be processing the medical user health data and location information in encrypted form so that it can not extract any contextual meaning of BAN data. We have presented the security and privacy issues in Cloud-enabled BAN and discussed some of the research done in this direction. Also, we have very briefly discussed the key agreement in BAN. With the advent of Internet of Things, this horizon of Cloud-enabled Body Area Network is going to enormously expand in the near future.

<<   CONTENTS   >>

Related topics