SECURE WEARABLE AND IMPLANTABLE BODY SENSOR

NETWORKS IN HAZARDOUS ENVIRONMENTS

Mohamed Hamdi, Noureddine Boudriga

Communications Networks and Security Research Lab., Ariana, Tunisia

Habtamu Abie

Norwegian Computing Center, Oslo, Norway

Mieso Denko

University of Guelph, Ontario, Canada

Keywords: Smart sensor networks, Wearable and implantable sensors, Intelligent session management, Secure

communication.

Abstract: The aim of Wearable and implantable monitoring devices is to collect relevant data from the application-

related environment, and transmit this information to the outside world. Modern microelectronics create

ever increasing opportunities, but it is still true that sensors form the weakest elements in the entire chain of

data collection and processing. The difficulty of deploying smart body sensor networks is exacerbated by

the hostile environments in which they are typically installed. In this paper, we propose a novel architecture

for wearable and implantable body sensor systems that guarantees both real-time responsiveness and

security. We rely on the wavelet packet transform to develop an intelligent session management scheme

where a customizable frame structure allows multiplexing the set of sessions between the elementary

sensors and the analysis center. We introduce a lightweight identity-based encryption protocol suitable for

body smart sensor systems. We also present performance results using simulation experiments.

1 INTRODUCTION

The last decade has witnessed a rapid surge of interest

in new sensing and monitoring devices for healthcare

and the use of wearable/wireless devices for multiple

applications (Puers, 2005). Key developments in this

area include implantable in vivo monitoring,

battlefield monitoring, and human tracking; where

sensors are strategically placed at various locations on

the vest or inside the human body to form a network

(Body Sensor Network) and interact with the human

system to acquire and transmit the data to an

acquisition system. The data acquisition hardware

collects the data from various sensors and transmits

the processed data to the remote monitoring station.

The basic requirement for such systems is that the

data gathered by the body sensors should be available

for transmission in real-time in response to a query

issued by the data acquisition system. When multiple

sensors are involved in the measurement process,

real-time responsiveness becomes hard to achieve

since all sensor nodes share the same communication

channel. Hence, the balance between response delay

and scalability should be carefully addressed. In

addition, security is a matter of concern in these

networks, as the data being monitored are the health

status of the individual. The sensor nodes used to

form these networks are resource-constrained, which

makes security applications a challenging problem.

The data are also vulnerable to external attackers,

who may inject errors in the routing information,

replay old routing information, distort routing

information or send malicious information. The data

are also subject to jamming, tampering, Sybil attack,

and collision (Hamdi and Boudriga, 2008. Attacks of

this nature which have been thoroughly investigated

and neutralized as threats within the context of

traditional wireless sensor networks, still represent a

threat to body sensor networks.

Hamdi M., Boudriga N., Abje H. and Denko M. (2010).

SECURE WEARABLE AND IMPLANTABLE BODY SENSOR NETWORKS IN HAZARDOUS ENVIRONMENTS.

In Proceedings of the International Conference on Data Communication Networking and Optical Communication Systems, pages 85-92

DOI: 10.5220/0002988200850092

 SciTePress

In this paper we develop a secure network

architecture for Wearable and Implantable Smart

Sensor Networks (WISSNs). We first propose an

architecture where intermediate sensor nodes allow

the data collected by the elementary sensors to be

forwarded to the analysis center (i.e., data acquisition

center). To support this architecture, we propose a

session multiplexing scheme that permits multiple

elementary sensors to share the communication

resources of an intermediate. It relies on wavelet

theory since it is necessary to allocate a variable

number of slots to a given elementary sensor in the

multiplexed frames, the number of slots varying with

the volume of date it generates or with its residual

amount of energy. We propose, in addition, a security

protocol addressing specific issues including

authentication/anonymity, accounting, confidentiality,

and investigation. The use of elliptic curve

cryptography minimizes the power consumption of

the cryptographic primitives while the absence of user

information in the authentication protocol preserves

privacy and anonymity.

The reader will notice that four innovative issues

are addressed in this paper:

Layer-2 Multiplexing: Rather than being performed

at the physical layer, the multiplexing of the

information originating from multiple sensor nodes

is dealt with using a specific frame structure based

on the wavelet packet transform. Such multiplexing

provides more fairness.

Real-time Responsiveness: The proposed

architecture guarantees that the queries from the

data acquisition center are processed in real-time by

the smart sensor system since a set of intermediate

sensors processes the information collected by the

elementary sensors

Privacy/anonymity Provision: Our security

protocol allows the data sent by the intermediate

sensors to the analysis center to be enciphered using

dynamic public keys. This guarantees the privacy of

the collected data as well as the anonymity of the

wearer.

Low Energy Consumption: The simulations that

have been performed show that our cryptographic

protocol is characterized by a low computational

complexity, making it convenient for use with the

limited resources of the intermediate sensors

The rest of this paper is organized as follows. Section

II presents related work. Section III presents the

WISSN. A novel session multiplexing technique

based on wavelet theory is discussed in Section IV.

Section V discusses security issues. Section VI

provides validation and performance evaluation of the

proposed techniques. And Finally, Section VII

concludes the paper with suggested avenues of future

research.

2 RELATED WORK

In recent years, there has been a proliferation of smart

monitoring based on small sensing devices. A large

portion of these devices have been devised for sports

science and combating obesity. For instance, there are

sophisticated watches available today (Polar),

(Suunto) that provide real-time measurement of

heartrate and allow athletes to store the gathered data

on computers for further analysis using specific

software. Bodymedia (BodyMedia) has developed an

armband that has multiple sensors (galvanic skin

response, skin and near-body temperature, two-axis

accelerometer and heat flux) and collects

physiological data on an on-going basis for days at a

time. Once the data is uploaded to a computer,

relevant and accurate information can be extracted

about, for example, fatigue, duration of physical

activity, consumed calories, etc. However, in all cases

the physiological data is analyzed on a home PC at a

later time, and proprietary data formats prevent users

from consolidating and correlating health monitoring

data from different devices.

In the medical domain, research is being

conducted on the remote monitoring of physiological

reactions (Scannell et al., 1995), (Martin et al., 2000),

(Oliver et al., 2006). However, in existing approaches,

as a rule no automated analysis is performed by the

device, and the raw data is instead sent to a remote

computer for further analysis by humans.

Traditionally, personal medical monitoring systems,

such as Holter monitors, have been used only to

collect data for off-line processing. An exception to

this is the approach proposed in (Oliver et al., 2006)

where a cell phone is used to store, transmit (via

Bluetooth) and analyze the physiological data, and

present it to the user in an intelligible way. In (Leister

et al., 2009) a security and authentication architecture

using MPEG-21 for wireless patient monitoring

systems has been developed based on the threat

assessment of wireless patient monitoring systems. In

(Leister et al., 2008), an architecture that can handle

end-to-end management of multimedia content in

diverse wireless sensor networks have been

proposed.

Martin et al discuss in (Martin et al., 2000) the

usage of wearable computers for health monitoring

where the devices provide real-time feedback to the

patient. In particular, they describe a wearable ECG

DCNET 2010 - International Conference on Data Communication Networking

device, but provide no experimental results. A

wearable health-monitoring device using a Personal

Area Network (PAN) or Body Area Network (BAN)

can be integrated into a user’s clothing (Park and

Jayaraman, 2003), like Foster-Miller’s health

monitoring garment for soldiers. Along these lines,

Paradiso (Paradiso, 2003) describes preliminary work

on the WEALTHY system, a garment with embedded

ECG sensors for continuous monitoring of the heart.

Jovanov et al present in (Jovanov et al., 2005) a

wireless BAN with motion sensors for computer-

assisted physical rehabilitation and ambulatory

monitoring. In (Kemp et al., 2008), Kamp et al

develop a wearable system for manned bomb disposal

missions. Mihovska and Prasad (Mihovska and

Prasad, 2007) have developed an adaptive security

architecture for personal networks with an

asymmetric key agreement scheme on three levels by

using contextual information, such as the location of

the user and the capability of the devices. This

architecture is based on an elliptic curve

cryptosystem. It has, however, one shortcoming. It is

susceptible to impersonation via key compromise.

A global notice about these approaches shows that

traditional communication protocols are used to

transmit the collected data from the human body to an

external system (e.g., cellphone, laptop).

Unfortunately, this does not guarantee a real-time

transmission of this information since an important

variable delay can occur, especially when some

sensors transmit large units of data such as images.

Moreover, due to the use of radio communication, the

confidentiality of the transmitted data is not

intrinsically guaranteed, which may lead to privacy

violation. In several applications, including

healthcare, even the identity of the wearer should be

hidden.

3 PROPOSED WISSN

ARCHITECTURE

In this paper, we address two crucial issues regarding

wearable sensor systems:

•Improving Real-time Responsiveness: This is

achieved by building special communication frame

structures based on the non-uniform multiplexing of

the data generated by different types of sensors

•Combining Sensor Authentication and user

anonymity through the use of lightweight

cryptographic protocols: In order to adapt to the

severe resource limitations characterizing WISSNs,

we use an elliptic curve implementation of the

proposed security functions

In spite of its apparent simplicity, WISSNs exhibit

several complex features and therefore require

sophisticated engineering approaches in order to be

set up. In the following, we list the most relevant

factors that may shape the communication models

used in smart sensor networks.

1. Multi-functional framework: A sensor node

may be able to carry out multiple functions that can

be set on/off depending on the situation. Obviously,

the communication requirements may differ greatly

from one functionality to another according to the

data sent across the WISSN. For instance, when the

network is deployed in a mining structure, a first

category of sensor may be used to monitor the amount

of several toxic gases in the atmosphere. A second

type of sensor can serve to estimate the opacity of the

encountered obstacles. IRM sensors can be used in

such a context in order to predict, and possibly

prevent, disasters. Since the volume of data generated

by the latter category is by far greater than that

generated by the former, much more bandwidth must

be reserved to transmit image data.

2. Independent monitoring capability: Due to the

non-uniform nature of the monitored events

(irrespective of the application), some sensors may

exhaust their energy more rapidly than others. This

may result in the presence of uncovered regions

where the nodes in charge of gathering data related to

the environment are out of power. Since such a

situation significantly affects the efficiency of the

WISSN, solutions should be proposed to avoid it. One

alternative is to tune the quality of the data gathered

by a sensor node according to its residual energy

resources. This would extend considerably the

lifetime of this node at the cost of losing some refined

data, which is definitely better than totally losing the

functionalities provided by the node. As a result the

communication resources required to transmit the

data may vary from one sensor to another.

3. Exportable configuration: Configurations can

be exported from one sensor to another in order to

turn on/off several functionalities. Even though this

feature allows energy to be saved (by triggering

power-consuming time only when necessary), it

creates a significant security hole since node

imposture can be easily carried out. Hence,

authentication mechanisms should be set up to

prevent non-authorized nodes from manipulating the

WISSN. Two important issues must be taken into

consideration: First, the security algorithms must be

based on non-complex algorithms and use small

cryptographic credentials (to adapt to limited CPU

time and memory resources) and; Second For a wide

range of applications, the anonymity of the person

holding the wearable or implantable smart sensor

system should be preserved. Since this conflicts with

authentication, specific security infrastructures will

SECURE WEARABLE AND IMPLANTABLE BODY SENSOR NETWORKS IN HAZARDOUS ENVIRONMENTS

have to be developed

From the foregoing discussion, it transpires that

sessions on WISSNs should be managed bearing in

mind that the specific features of such networks. In

fact, a session should typically be initiated by a

central external (i.e., not wearable) node, called the

analysis center, in order to collect data from the

WISSN. Henceforth, the underlying bandwidth

management scheme should guarantee fairness for all

body sensors. Unlike traditional networks, fairness in

WISSNs should take into account the differences in

the nature of the generated data and the available

power level.

Figure 1: Wavelet packet decomposition.

Figure 1 illustrates this reasoning. In order to improve

the scalability of the communication structure, we

propose to divide the body area network into a set of

clusters. We make the assumption that within each of

these clusters, there is a central node which is in

charge of forwarding the data gathered by the sensors

present within the cluster and the analysis center.

Since the contents of the frames sent out by the

central node of a cluster to the analysis center

originate from multiple body sensors, an intelligent

multiplexing scheme is needed.

4 INTELLIGENT SESSION

MANAGEMENT

This section develops a novel session multiplexing

technique based on wavelet theory. We first discuss

the mathematical aspects related to the wavelet packet

transform. Then, we develop a multiplexing scheme

where data emanating from multiple elementary

sensors can be carried in a unique frame flow. For this

purpose, we introduce a frame structure based on the

parent-child relationship defined in wavelet theory.

The Wavelet Transform (WT) is a time-scale

transform that can be used to perform signal analysis.

It offers effective time-frequency representation of

signals. Wavelet theory and application have matured

in recent decades and have proven to have

tremendous application in fields such as data

compression, multi-scale analysis, transient signal

processing, and more. In practice, the wavelet

transform is implemented using a couple of filters; a

low-pass filter is used to generate approximation

coefficients and a high-pass filter is used to generate

detail coefficients. A decimation phase is also used so

that the size of each of the approximation and detail

signals is half the size of the input signal.

Mallat (Mallat, 1989) showed that a multi-

resolution decomposition of a signal f(t) can be

achieved by iterating the wavelet decomposition on

the approximation signal (which will be initialized to

f(t)). More recently (Feil and Uhl, 1998), a more

sophisticated multi-resolution analysis, based on the

Wavelet Packet Transform (WPT), has been

proposed to apply the wavelet transform to both the

approximation and the detail coefficients at every

decomposition stage. Figure 2 illustrates this

transform where H

denotes the low-pass filter and

denotes the high-pass filter.

Figure 2: Wavelet packet decomposition.

For n levels of decomposition the WPT produces 2n

different sets of coefficients (or nodes) as opposed to

(n + 1) different sets for the DWT. However, due to

the downsampling process the overall number of

coefficients is still the same and there is no

redundancy.

The basic idea of our work is that larger time

slots should be allocated to the sensor nodes that

provide more refined data. For this purpose, we

define a parent–child relationship between wavelet

coefficients, and let the coarser resolution transport

the most refined data. We have investigated

dependencies between wavelet coefficients on this

traffic. As shown in (Feil and Uhl, 1998), the

dependencies between parent-child are very

important. Therefore, the wavelet packet transform

minimizes the cross-correlation between two

decomposed signals. Therefore, a frame issued by an

intermediate sensor node can carry data from

DCNET 2010 - International Conference on Data Communication Networking

multiple flows generated by different sensor nodes.

This idea is detailed in the following.

Importance should be attached to the values on

diagonals. The main diagonal is not so important in

our case. More important are other diagonals, which

directly show the dependencies between predecessors

and successors. For example, second diagonal reveals

direct parent-child dependency.

According to Figure 2, if the size of a signal f(t) is

denoted by s, then the size of the signals obtained

after n wavelet stages is s/2i (the rounding operator is

omitted because we suppose that s is a power of 2).

Therefore, if n is the number of elementary sensors

and F is the frame size; then, a fair decomposition of

the frame gives that <F/n> bits are allocated to every

elementary sensor, where <.> denotes the rounding

operator. Hence, the frame can be structured so that

the analysis center reconstructs the signals

corresponding to lower decomposition depths before

those corresponding to deep depths. The role of the

intermediate sensor is simply to increase the depth of

the wavelet packet transform according to the

priority of the corresponding sensor node.

5 SECURITY PROTOCOL

The first step in ensuring secure data aggregation at

intermediate nodes is to enable the intermediate nodes

to have appropriate encryption/decryption keys to

communicate and decipher the incoming data, apply

the aggregation function and relay them forward.

When data is relayed, it is assumed that it is broadcast

(using omni-directional antenna). Furthermore, if the

same data has to be transmitted to several nodes and if

the nodes are operating using distinct pair-wise keys,

then, care must be taken to transmit the data multiple

times, each time encrypted differently with a different

key. This could potentially be a drain on energy and a

hindrance to in-network processing. As we have seen

earlier, having a common key for the group of nodes

is a possible solution to this problem, but has an

inherent weakness in that the whole network could be

compromised if an attacker successfully attacks any

one node.

The basic idea behind our protocol is to make a

sensor independently generate a public key using an

arbitrary string. For example, a sensor collecting data

of type T at time t will first create a string σ =

(sensor_id|t|T). Using this string, the sensor can

derive a public key, π

to encrypt the data and send

them to the storage site. There is no corresponding

secret key created. In fact, the sensor cannot create

the secret key needed to decrypt the message.

When the sensor wishes to release this

information to the analysis center, the sensor can

derive the corresponding secret key, κ

, by using the

same string σ. This secret key only allows the analysis

center to decrypt messages encrypted by a sensor

using the same string. This simplifies key

management, since the sensor can generate the secret

key on-demand without keeping track of which keys

were used to encrypt which data. The only

requirement is that the string used to describe the

event is the same.

Setup: We select an elliptic curve E over GF(p),

where p is a big prime number. We also denote P as

the base point of E and q as the order of P, where q is

also a big prime. A set of n secret keys κ

,…, κ

∈

GF(q) is chosen to generate the master secret key,

denoted by K = (κ

,…, κ

). The n public keys are then

generated to make up the master public key, denoted

by Π = (π

,…, π

), where π

= κ

.P, 1 ≤ i < n. Finally,

a collision resistant one-way hash function is chosen,

The parameters (Π, P, p, q, h(.)) are released as

the system public parameters.

Keygen: To derive a secret key κ

corresponding to a

public key generated by a string σ, the sensor

executes keygen(σ) = κ















.



where h

(σ) is the i-th bit of h(σ).

Encrypt: To encrypt a message m using a public key

derived from string σ, the sensor does encrypt(m,σ) to

determine the ciphertext c.

Algorithm encrypt

Determine string σ using agreed-upon syntax

Generate public key π

where

= Pn

=1 h

(σ) · yi

Execute EccEncrypt(m, π

) to obtain c

Decrypt: The analysis center executes decrypt(c, κ

)

to obtain the original message m which was encrypted

using a secret key derived from σ.

Algorithm decrypt

Requests permission from sensor to obtain data

described by σ

Sensor runs Keygen(σ) to derive κ

Analysis center executes EccDecrypt(c, κ

) to obtain m

Based on these functions, we develop the

following protocols for secure data collection,

transfer, and aggregation.

Secure Data Collection: Having collected an event

d, the sensor executes the following algorithm to

encrypt it.

SECURE WEARABLE AND IMPLANTABLE BODY SENSOR NETWORKS IN HAZARDOUS ENVIRONMENTS

Algorithm secure_data_collection

Derive the string σ, and generate a random number n

Calculate m1 = (flag|n) where flag is a known bitstring

Calculate m2 = (d|n)

Calculate c1 =Encrypt(σ,m1)

Calculate c2 =Encrypt(σ,m2)

Secure Data Transfer: Periodically, each sensor in

the WISSN will transfer its data to the analysis center.

This is done by first aggregating all the data into an

intermediate sensor node, which then forwards the

aggregated data to the storage site. Assuming that

there are k tuples generated by the WISSN, the

intermediate sensor will forward the set {(c

, c

), …

, (c

, c

)}.

Secure Data Querying: An analysis center wishing

to obtain data collected under some σ will first

contact the CA for permission. After the CA agrees,

the CA will run Keygen(σ) to derive the

corresponding secret key κ

needed to decrypt data.

Then, the following algorithm is executed to decrypt

the data:

Algorithm Secure data querying

for every (c

, c

) i

∈

k for sensor do

Storage site sends ci1 to analysis center

Analysis center runs Decrypt(c

, σ)

if the initial bits of the result match flag then

Analysis center requests corresponding

from storage site

Analysis center executes Decrypt(c

, σ)

and checks whether the n matches the

value from c

Analysis center accepts d if both are correct

end if

end for

Since all the data are encrypted, the storage site

cannot return a specific encrypted tuple to the analysis

center. Instead, the storage site simply lets the

analysis center try to decrypt each tuple (c

, c

)

belonging to the sensor. The reason for returning c

the analysis center first instead of returning c

directly

is to improve efficiency. Since the length of c

much shorter than c

, letting the analysis center first

attempt to decrypt c

before sending the much longer

reduces transmission time.

The analysis center can check if the data obtained

from the storage site belongs to his sensor by

checking whether the same random number n is used

in both c

and c

. Since this random n is known only

to the sensor encrypting the data, only that sensor can

embed the same n in both c

and c

6 ASSESSMENT

AND EVALUATION

In this section, we validate the proposed session

management and security protocols. We first analyze

the features of the developed functionalities with

respect to the requirements given in Section III.

Then, we proceed to a performance evaluation based

on simulation of the wavelet-based session

management scheme. Finally, we discuss the security

properties guaranteed by our cryptographic protocols.

A. Proving Features

We discuss the features of the developed

functionalities with respect to real-time

responsiveness, fairness, privacy and anonymity.

Real-time responsiveness: The data acquisition

center is able, via structured queries, to have the data

collected by the sensor nodes nearly in real-time. In

fact, the period between two queries has to be

sufficient for the transmission of n.l.p.s bits, where n

is the number of elementary sensors, l is the event

rate, p is the even gathering periodicity, and s is the

average signal size. For n=10, l=2, p=1mn, and s=2

we find that the transmission rate between the

elementary sensor and the analysis center should be

approximately 6kb to fulfill the real-time

requirement. The period of time needed to upload all

the collected events to the analysis center would be

0.6 seconds, in that case, using a 10kb/s link.

Fairness: the allocation scheme used when building

the upward frame guarantees an equal slot of time for

all nodes constituting the BAN. Nodes that have

larger quantities of information to send are provided

with greater depth, using wavelet transform, to send

more data in the same period of time. This approach

reduces the latency measured for the arriving data at

the analysis center.

Privacy and Anonymity: privacy provided by a BAN

in a hazardous environment covers personal

information related to the wearer and information

related to the collecting sensors (e.g., used algorithms

and nature of the data collected). After multiplexing

the collected data, the transmitted frame is unable to

show any of the private information since the wavelet

transform will mix these data at variable depths. In

addition, a public encryption is added to this process.

B. Security Evaluation

The main overhead of our protocols is the amount of

time needed to generate a single π

from a string σ

using n number of public keys π

,…, π

. Note that the

DCNET 2010 - International Conference on Data Communication Networking

value of n is not related to the number of different π

s that can be generated. The WISSN can continue to

generate as many π

on-the-fly as needed, regardless

of the value of n. Once π

is generated, the remaining

encryption is the same as that for a regular ECC

encryption.

Figure 3 shows the amount of time needed to

generate a single π

with varying values of n. All n

public keys are initially stored in the flash memory.

Figure 4 shows the amount of flash storage need to

store n different public keys. We see from Figure 3

that, for n = 360, we need only 0.9 seconds to

generate π

Figure 3: Time needed to derive one public key versus the

number of elementary keys.

Figure 4: Flash memory storage needed to store one public

key versus the number of elementary keys.

Figure 5 shows the time needed to perform the

encryption once the public key π

has been derived.

For a given piece of data, encrypting with just one π

requires about 1.5 seconds. Again this is the

encryption time for the symmetric key (r), which will

then be used to encrypt the raw data. The symmetric

key can be used for a period, say 10 minutes. The

cost of the 1.5s can be compensated over the 10

minute period. The amount of time needed for

multiple π

s to encrypt the same data is proportional

to the number of π

s. While in Figure 5 the amount

of time needed for 10 different π

is close to 15

seconds while it is worth mentioning that ,in practice,

we are unlikely to use many different public keys to

encrypt the same event.

Figure 5: Time needed for encryption versus number of

keys.

7 CONCLUSIONS

AND PERSPECTIVES

In this paper, we defined an architecture for secure

wearable and implantable smart sensor networks

where an analysis center periodically launches

queries to gather data related to the monitored

environment. To adapt to the hazardous nature of the

contexts where such systems are typically deployed,

we proposed a multiplexing scheme and a

cryptographic protocol based on wavelet packet

decomposition and elliptic curve cryptography

respectively. We have shown that these approaches

provide real-time responsiveness (through intelligent

session management) as well as anonymity (since the

human identity is not involved in the cryptographic

protocol).An extension of the session multiplexing

technique to a physical layer is under development

for use in situations where the optical sensors are

linked to the analysis center via laser beams.

Moreover, a simpler security protocol not involving

intervention by the certification authority is being

developed.

Our future work will also include the design and

deployment of wearable and implantable smart

sensor nodes with light-weight self-abilities to detect

in real time unknown activity patterns, to swiftly

respond to them, and to learn activity patterns over

time and adapt to the dynamism of the hazardous

environment and to changing degree of security and

privacy breaches. Such abilities may enable the

reduction of communication overhead between

SECURE WEARABLE AND IMPLANTABLE BODY SENSOR NETWORKS IN HAZARDOUS ENVIRONMENTS

nodes. Applications scenarios such as healthcare and

smart homes will be investigated.

REFERENCES

Robert Puers, “Implantable sensor systems”, DISens

symposium- book 2005.

M. Hamdi, N. Boudriga, “Security In Wireless Sensor

Networks,” Handbook of Research on Wireless

Security, Information Science Reference, ISBN-10:

159904899X, March 2008.

Polar. Polar watches. http://www.polarusa.com.

Suunto. T6, foot pod, n6hr. http://www.suunto.com.

BodyMedia. Healthwear armband, bodybugg. http://

www.bodymedia.com.

K. M. Scannell, D. A. Perednia, and H. M. Kissman.

Telemedicine: Past, present, future. Technical report,

U.S. Department of Health and Human Services.

National Library of Medicine. Reference Section.,

1995.

Martin. T., E. Jovanov, and Raskovic. D. Issues in

wearable computing for medical monitoring

applications: A case study of a wearable ecg

monitoring device. In Proc. Intl. Symp. Wearable

Computers (ISWC’00), pages 43–49, 2000.

Oliver, N. Flores-Mangas, F. HealthGear: a real-time

wearable system for monitoring and analyzing

physiological signals. International Worskhop on

Wearable and Implantable Body Sensor Networks,

2006.

S. Park and S. Jayaraman. Enhancing the quality of life

thourgh wearable technology. IEEE Engineering in

Medicine and Biology Magazine, 22:3:41–48,

2003.Dfd

R. Paradiso. Wearable health care system for vital signs

monitoring. In Proc. IEEE Int. Conf. Information

Technology Applications in Biomedicine, pages 283–

286, 2003.

E. Jovanov, A. Milenkovic, C. Otto, and P. C. de Groen. A

wireless body area network of intelligent motion

sensors for computer assisted physical rehabilitation.

Journal of Neuroengineering and Rehabilitation, 22:6,

2005.

Kemp, J., Gaura, E. I., Brusey, J. 'Instrumenting Bomb

Disposal Suits with Wireless Sensor Networks.' In

Proceedings of the 5th International Conference on

Informatics in Control, Automation and Robotics

(ICINCO 2008). May 11-15 2008 at Funchal, Madeira

– Portugal, 2008.

S. G. Mallat "A Theory for Multiresolution Signal

Decomposition : the Wavelet Representation",. IEEE

PAMI, Vol.11(7), pp.674−693, 1989.

M. Feil and A. Uhl. Wavelet Packet Decomposition and

Best Basis Selection on Massively Parallel SIMD

Arrays. Proceedings of the International Conference

“Wavelets and Multiscale Methods” (IWC’98),

Tangier, 1998.

A. Mihovska and N. R. Prasad, Adaptive Security

Architecture based on EC-MQV Algorithm in Personal

Network (PN), Fourth Annual International Conference

on Mobile and Ubiquitous Systems: Networking &

Services (MobiQuitous 2007), Volume, Issue, pp. 1-5,

6-10 August, 2007

W. Leister, H. Abie, A.-K. Groven, T. Fretland and I.

Balasingham, “Threat Assessment of Wireless Patient

Monitoring Systems,” ICTTA'08 (Intl. Conf. on

Information and Communication Technologies: From

Theory to Practice), Damascus, Syria, 7-11 April 08,

2008.

W. Leister, T. Fretland, and I. Balasingham, Security and

Authentication Architecture Using MPEG-21 for

Wireless Patient Monitoring Systems, International

Journal on Advances in Security, 2009 Vol. 2, No. 1

(in press)

DCNET 2010 - International Conference on Data Communication Networking