Seyyed Hamed Fouladi,Raú l Ch ávez-Santiago,2,3,Pål Ander Floor,2,3,Ilangko Balasingham,2,3, and Tor A.Ramstad

(1.Department of Electronics and Telecommunications，Norwegian University of Science and Technology，Trondheim，Norway;

2.The Intervention Center，Oslo University Hospital，Norway;

3.Institute of Clinical Medicine，University of Oslo，Norway)

Sensing,Signal Processing,and Communication for WBANs

Seyyed Hamed Fouladi1,Raú l Ch ávez-Santiago1,2,3,Pål Ander Floor1,2,3,Ilangko Balasingham1,2,3, and Tor A.Ramstad1

(1.Department of Electronics and Telecommunications，Norwegian University of Science and Technology，Trondheim，Norway;

2.The Intervention Center，Oslo University Hospital，Norway;

3.Institute of Clinical Medicine，University of Oslo，Norway)

A wireless body area network(WBAN)enables real⁃time monitoring of physiological signals and helps with the early detection of life⁃threatening diseases.WBAN nodes can be located on，inside，or in close proximity to the body in order to detect vital signals. Measurements from sensors are processed and transmitted over wireless channels.Issues in sensing，signal processing，and com⁃munication have to be addressed before WBAN can be implemented.In this paper，we survey recent advances in research on sig⁃nal processing for the sensor measurements，and we describe aspects of communication based on IEEE 802.15.6.We also discuss state⁃of⁃the⁃art WBAN channel modeling in all the frequencies specified by IEEE 802.15.6 as well as the need for new channel models for new different frequencies.

wireless body area network;IEEE 802.15.6;signal processing;security;channel modeling

1 Introduction

T raditional healthcare systems can potentially be re⁃placed by wireless body area networks(WBANs). In a WBAN，various sensors are used to sense vital signs.Patients suffering from conditions such as heart disease can be continuously monitored with a WBAN[1]，[2].Data from all the sensors is transmitted over a wireless channel to a base station，and the received measurements are processed to extract the desired information.We give an over⁃view of sensor devices，physical layer(PHY)，data link layer，and radio technology in WBAN.Different kinds of sensors can be used in a WBAN depending on requirements such as data rate and power consumption.In[3]and[4]，the authors give an overview of body area networks and discuss WBAN communi⁃cation types and related topics.In this work，we discuss signal processing，implant communication，and security，which have scarcely been discussed in relation to WBANs.

The IEEE 802.15.6 Task Group was established to standard⁃ize WBAN technologies and communication[5].The main fo⁃ cus of the 802.15.6 standard is low⁃power sensors used on，in，or near the human body[5].The standard supports physical layers，including narrowband(NB)and ultrawideband(UWB) radio interfaces，and human body communication(HBC).

In this paper，we describe wearable and implantable sensors for electrocardiography(ECG)，electroencephalography(EEG)，electromyography(EMG)，blood pressure(BP)，pulse oximetry (SpO2)，accelerometer，andwirelesscapsuleendoscopy (WCE).The measurements from these sensors need to be pro⁃cessed，so we discuss some of the signal⁃processing techniques for WBAN.We also discuss the PHYs of 802.15.6 and channel modeling in all the frequencies specified by the standard.

This paper is organized as follows:In section 2，we describe the sensors used in a WBAN and their requirements.We also discuss some important common data processing techniques for medical applications.In section 3，we discuss the PHYs in 802.15.6.In section 4，we review previous works on on⁃body and in⁃body channel modeling;in particular，we focus on im⁃plant communications.In section 5，we discuss network securi⁃ty.In section 6，we discuss future challenges.In section 7，we make some concluding remarks.

2 Sensors and Their Requirements

In this section，we review sensors typically used in a WBANas well as the requirements of these sensors in terms of data rate and power consumption.We also discuss common signal⁃processing techniques.We consider two main categories of sen⁃sors:wearable and implantable.The first category includes ECG，EMG，EEG，accelerometer，BP，and pulse oximetry sen⁃sors.The second category includes glucose monitoring and WCE sensors.

2.1 Sensors

2.1.1 Electrocardiography

ECG is widely used in biomedical sensing，and many wire⁃less systems for ECG monitoring have been proposed[6].The ECG waveform shows the propagation of electric potentials through the heart muscle as a function of time.This propaga⁃tion is the result of contraction of the heart muscle，and the per⁃formance of the heart can be determined by analyzing the ECG waveform.ECG measurements are based on twelve or six leads of the electrical activity of the heart.However，wireless sensors are generally only used in ambulatory scenarios to take ECG measurements，which are typically based on a subset of these leads.In[6]and[7]，ECG measurements are wirelessly trans⁃mitted at a required data rate 288 kbps and 71 kbps for 12 and 6 leads，respectively.

2.1.2 Electroencephalography

The electrical activity of the brain can be monitored by EEG.Ambulatory EEG(AEEG)is valuable for diagnosing epi⁃lepsy and monitoring patient response to therapy[8].Much of the information gained from AEEG over a 20 to 40 minute peri⁃od cannot be gained from regular EEG over the same period. This has led to improved wireless EEG sensors that reduce the need for more data⁃intensive AEEG recording during daily ac⁃tivities.The required data rate for EEG is only 43.2 kbps[9].

2.1.3 Electromyography

EMG is the recording and analysis of electrical activity of skeletal muscles.The instrument used for this purpose is called an electromyograph，and the record produced is called an electromyogram[10].EMG signals can be used to detect medical abnormalities and analyze biomechanics.EMG is of⁃ten used to examine mechanisms associated with daily physi⁃cal activities that induce pain and to devise related treatment regimes[11].

2.1.4 Blood Pressure

Blood pressure is one of the most important vital signs.In⁃creased blood pressure(hypertension)increases the risk of myocardial infarction，congestive heart failure，stroke，kidney failure，and blindness.Devices that measure blood pressure are mostly based on a sphygmo⁃manometric obstructive arm⁃cuff，which is clumsy，uncomfortable，and allows only intermit⁃tent measurement every several minutes.Continuous cuffless blood pressure monitoring opens up new possibilities for hyper⁃tension diagnosis and treatment，cardiovascular event detec⁃tion，and stress monitoring[12].

2.1.5 Pulse Oximetry

Pulse oximetry is a standard way of measuring arterial blood oxygen saturation(SpO2)in operating rooms，intensive care units，and pediatric care units.SpO2 is one of the most impor⁃tant vital signs，especially for the early detection of hypoxemia. Trauma management involves accurate monitoring of several physiological parameters，including SpO2，so that proper ac⁃tion can be taken to preserve critical functionality.State⁃of⁃the⁃art integrated circuits，wireless communications，and physiolog⁃ical sensing paves the way to miniature，lightweight，low pow⁃er，intelligent pulse oximeters that are appropriate for WBAN applications[7].

2.1.6 Accelerometers

Accelerometers are used to measure acceleration acting on a device and convert this acceleration to an electrical signal.Al⁃gorithms can be used to classify the subject’s movement into one of a few groups[13].Research has also shown that acceler⁃ometers are effective in long⁃term activity monitoring and rec⁃ognition.One application of accelerometers is monitoring the uncontrolled body movements(dyskinesias)of Parkinson’s pa⁃tients[14].This may lead to a more effective use of levodopa，a drug used to treat the symptoms of Parkinson’s disease.

2.1.7 Wireless Capsule Endoscopy

Endoscopic and radiological investigation of the small intes⁃tine had limited diagnostic operation before the year 2000. This meant that intestinal disease was sometimes diagnosed late，which worsened the patient’s prognosis.WCE is a recent technique for examining the small intestine in a non⁃invasive way.WCE is performed using a camera the size and shape of a pill.The patient swallows the camera so that the doctor can view images of the gastrointestinal tract.However，this real⁃time video imaging requires a high data rate[15]，[16].It has been demonstrated WCE is more effective than other tech⁃niques in detecting small intestine diseases[17].

The patient swallows the WCE camera with water and wears a recorder belt around the waist.Some hours later，medical staff analyzes the video created from the still images transmit⁃ted from the WCE to the recorder belt.The high⁃quality imag⁃es from the WCE camera can be analyzed in real time to en⁃able more precise diagnosis;however，this additional capabili⁃ty increases transmission complexity and power consumption. The WCE camera should consume the lowest amount of power possible，on the order of microwatts，and should only be about 300 mm3.To transmit video in real time，a high⁃rate communi⁃cation link is needed.For VGA images，i.e.，640×480 p and 10 fps，73.8 Mbps is needed[18].Therefore，powerful，low⁃complexity compression algorithms are necessary to decreasedata rate and power consumption.

2.2 Data Rate

Different applications in a WBAN require different data rates.For example，several megabytes per second are required for WCE whereas only several kilobytes per second is required for ECG.Table 1 shows the required data rates for WBAN ap⁃plications[7]，[9]，[19].

▼Table 1.Data rates for different applications

2.3 Power Consumption

Power consumption is one of the most significant constraints in a WBAN.There are limitations on the size of batteries and，as a consequence，the amount of power they can provide.In some WBAN applications，such as implant sensors，the sen⁃sors need to work for several years without battery replace⁃ment.In order to save power，a thermoelectric MEMS genera⁃tor can be used to scavenge energy from the surrounding envi⁃ronment[19].Harvesting energy from commercial radio fre⁃quency transmissions to power WBAN nodes has been shown to be feasible[20].Also，body heat can be used to help power wearable sensor nodes[8].Available power in sensors is used for wireless communication，sensing，and data processing[19]. However，the main cause of power consumption in sensors is wireless communication.

A significant amount of power can be saved by modifying standard protocols according to the specific needs of WBAN application[20].Moreover，power⁃aware sensor nodes can esti⁃mate the amount of transmission power needed to keep con⁃nected to the network[21].With this concept，each node in the network can trade⁃off performance for energy efficiency[22].

2.4 Signal Processing

Because power is limited，low⁃power wireless communica⁃tion and processing algorithms are needed.In a WBAN，signal processing involves processing the measurements of sensors and transmitted data to extract the desired information.Data processing within the sensor nodes must not be complex in or⁃der to reduce power consumption and prolong battery life in the sensors.However，data processing and analysis outside the sensors can be complex and power and time consuming.In this subsection，we introduce some common signal⁃processing algo⁃rithms for the sensors mentioned in subsection 2.1.

Sensing involves detecting a physical presence of data and transforming the signal into a format that can be read by an ob⁃server or instrument.A well⁃designed WBAN can give doctors accurate real⁃time and historical information.Data derived from different applications must be preprocessed so that it is ready for analysis.In[21]，the authors propose a data prepro⁃cessing model for decreasing the power consumed during com⁃munication between nodes and for improving data transmission in wireless sensor networks.The model also provides a way of determining the integrity of data.If the data is incomplete，the error is identified，missing data is added，and corrupted data is repaired.Data processing in nodes helps eliminate redundant information，decrease the data rate，and save power.Some sig⁃nal⁃processing algorithms are used to detect abnormalities and artifacts and track the mobile implantable sensor.Other signal⁃processing algorithms are used for classification.

2.4.1 Compression

Data processing usually requires less power than wireless transmission.Thus，it is important to develop algorithms that reduce the amount of information transmitted.Compression is a well⁃known technique used before transmission to decrease the data rate.In a wireless sensor network，such compression algorithms need to be energy⁃efficient.Much research has been done on data compression for wireless networks.In[23]，data⁃compression approaches are categorized as distributed da⁃ta compression and local data compression.Looking at distrib⁃uted data compression，the authors of[24]aim to find a func⁃tion or model that fits a best set of input measurements ob⁃tained by a specific class of sensor node.Parametric and non⁃parametric modeling is used.The authors of[25]use distribut⁃ed transform coding to decompose data into components or co⁃efficients，which are then coded according to their individual characteristics.There are also several well⁃known compression techniques，such as Karhunen-Loeve transform，cosine trans⁃formation，and wavelet transformation，which are used in im⁃age and video compression applications.Distributed source coding(DSC)is a well⁃known technique，based on the Slepian⁃Wolf theorem，for data compression in wireless sensor net⁃works.The authors of[26]proposed energy⁃efficient distribut⁃ed source⁃coding methods that have a spatial correlation for wireless sensor networks.Recently，a technique based on sam⁃pling theory was proposed by the authors of[27].This tech⁃nique，called compressed sensing，has low complexity at the sensor nodes and saves power.In other recent research，signalsare reconstructed using approaches such as basis pursuit and orthogonal matching pursuit[28]，[29].In[30]，an approach based on compressed sensing was proposed to compress ECG signals.Local data compression is discussed in[23]and its ref⁃erences.

In WCE，the diagnosis can be improved by increasing the data and frame rate.Because power consumption is limited，the data rate and frame rate should also be limited as long as the image quality satisfies hospital staff.To decrease the data rate and save power，the WCE images are compressed and en⁃coded.A differential pulse⁃coded modulation(DPCM)coder re⁃quires little memory and is very easy to implement in a WBAN.The authors of[31]proposed a low⁃complexity algo⁃rithm for encoding WCE images with a DPCM coder and com⁃bined this with decimation，dead⁃zone quantization，and effi⁃cient run⁃length coding of the quantization indices.DPCM can also be used for other medical signals，such as ECG signals.In [32]，the authors studied compression schemes for a two⁃node sensor network.DPCM was used to remove temporal correla⁃tion，and distributed quantization was used to exploit inter⁃sen⁃sor correlation.In[33]，a communication system for wireless sensor networks is proposed.Low⁃complexity，delay direct⁃sum source encoder is used to reduce power consumption.

2.4.2 Localization of Capsule Endoscopy

WCE was first proposed by Given Imaging Ltd[34].The pro⁃cess enables painless diagnosis within the gastrointestinal tract，specifically the small intestine.The capsule moves along the GI tract with the normal peristaltic movement of the gut and transmits the images to outside the body.Original localiza⁃tion approaches for WCE were based on the received signal strength(RSS)[35].However，recent research has shown that localization accuracy can be increased in RSS⁃based approach⁃es by using ultrawide bandwidth[16].In[17]，a technique based on magnetic localization was proposed for WCE.Magnet⁃ic tracking is attractive because the magnetic signal can pass through the human body without attenuation.A magnetic sig⁃nal has low⁃static，low⁃frequency properties，and the capsule does not have to be in the line of sight of the magnetic sensors to be detected.In[36]，the WCE localization problem is consid⁃ered as a tracking problem.Maneuvers of the capsule endo⁃scope，including any sudden stops and starts，can be tracked.

2.4.3 Detection

Detection theory is widely used in biomedical applications to find a special pattern in the signal background.Different kinds of unwanted background signals，created by the human body，deteriorate the ECG signals[34].For example，EMG sig⁃nals from muscle contraction and relaxation decreases the sig⁃nal⁃to⁃noise ratio of ECG signals.Detection algorithms have been developed to eliminate noise and accurately detect the peak points in the ECG signal.In[37]，the authors discuss the complexity and performance of detection schemes.In[38]，the authors proposed a new pre⁃processing technique with Shan⁃non energy envelope estimator to improve detection of R⁃peaks in ECG signals.In[39]，the authors took into account both complexity and accuracy and created a detection algorithm us⁃ing wavelets.

Detection algorithms are also applicable to EMG signals. The duration of EMG onset is short and must be precisely known[39].A threshold is set by the statistical properties relat⁃ed to the amplitude distribution of the EMG signal.In[40]，the authors use the maximum likelihood function in an algorithm for detecting the onset of muscle activity from EMG records.

3 WBAN Communications:IEEE 802.15.6

In this section，we consider the PHY layers specified in IEEE 802.15.6，including NB，UWB，and HBC.UWB and HBC PHYs are not optional whereas the NB PHY is optional [5].Proper selection of PHYs or frequency bands is an impor⁃tant consideration in the development of WBANs[41].Fig.1 shows the available frequency bands for WBANs[2].

3.1 NB PHY Specification

Narrowband PHY is responsible for 1)activating and deacti⁃vating the radio transceiver，2)clear channel assessment with⁃in the current channel，and 3)data transmission and reception. NB PHY provides a way of converting a Physical Layer Service Data Unit(PSDU)into a Physical Layer Protocol Data Unit(PP⁃DU).The PPDU frame of the NB PHY contains a physical⁃lay⁃er convergence procedure(PLCP)preamble，a PLCP header，and a PHY Service Data Unit(PSDU)(Fig.2)[42].

The PLCP preamble helps the receiver with timing synchro⁃nization and carrier offset recovery.The other main part of the PPDU is PLCP header.The main purpose of the PLCP header is to carry essential information about the PHY parameters in order to decode the PSDU at the receiver.The PLCP header can be divided into the following fields:rate，length，burst mode，scrambler seed，reserved bits，header check sequence，and BCH parity bits.The BCH parity is responsible for improv⁃ing the robustness of the PLCP header，which is transmitted af⁃ter the PLCP preamble using the given header data rate in the operating frequency band.The last part of the PPDU is the PS⁃DU.The PSDU is formed by concatenating the MAC header with the MAC frame body and frame check sequence.The PS⁃DU is then scrambled and optionally encoded by a BCH code. The PSDU is transmitted using any of the available data rates in the operating frequency band[43].

In a WBAN，modulation has to be efficient in order to pro⁃long life of the batteries.Gaussian frequency⁃shift keying，pulse⁃position modulation，Gaussian minimum⁃shift keying，differential phase⁃shift keying，offset quadrature phase⁃shift keying，and phase silence⁃shift keying are the predominant modulations in this area[43]，[44].These modulations reduce side lobe，are easy to implement，and are efficient in terms ofbandwidth.

▲Figure 1.IEEE 802.15.6 frequency bands,bandwidths,and ranges of data rates(kbps).

▲Figure 2.Standard PPDU structure for NB PHY.

3.2 Ultrawideband PHY Specification

UWB PHY uses both low⁃and high⁃frequency bands that are divided into 0-10 channels with a bandwidth of 499.2 MHz[5].The low⁃frequency band contains the first three chan⁃nels.The high⁃frequency band contains the remaining eight channels.Channels 1 and 6 are mandatory，and the others are optional.The central frequencies of channels 1 and 6 are 3993.6 MHz and 7978.2 MHz，respectively.However，in prac⁃tice，one of these mandatory channels needs to be supported by UWB devices.The mandatory data rate is 0.4882 Mbps.The transceivers of UWB PHY are not complicated to implement and generate low⁃power signals.The UWB PPDU contains a synchronization header(SHR)，PHY header(PHR)，and PSDU. Figure 4 in[45]shows the UWB PPDU structure.The SHR in⁃cludes a preamble and start frame delimiter(SFD).The PHR provides the data rate of the PSDU，length of the payload，scrambler seed，and decoding procedure in the receiver.The SHR comprises repetitions of Kasami sequences with a length of 63.

3.3 HBC PHY Specification

HBC PHY operates in two frequency bands and has a band⁃width of 4 MHz.The central frequencies of the low and high bands are 16 MHz and 27 MHz，respectively.The WBAN pro⁃ tocol，which specifies packet structure，modulation，preamble/SFD，etc.，is iden⁃tified by HBC.Figure 5 in[45]shows the PPDU structure of electrostatic field communication.This structure contains a preamble，SFD，PHR，and PSDU.The preamble and SFD are generated and sent before the packet header and pay⁃load.To ensure packet synchronization，the preamble sequence is transmitted four times.The preamble sequence is used to find the start of the packet in the receiver，and then the receiver locates the start of the frame by detecting the SFD.HBC has been referred to as intra⁃body communication[46].In[46]，the IBC transceiver design and mathematical models of the human body are presented.

4 Radio Propagation for Body Area Networks

Accurate channel models have to be used to fairly evaluate and compare the performance of different PHYs.Channel mod⁃els have to characterize the mean path loss of WBAN devices as well as the scattering around the mean value caused by dif⁃ferent postures of human bodies or objects in the vicinity，i.e.，

wherePLis the total path loss，Sis the scattering that accounts for different fading phenomena，andPL(d)is the distance⁃de⁃pendent loss.In some WBAN scenarios，PL(d)can be comput⁃ed by the Friis equation given as

WherePL0is the path loss at a reference distanced0andn is the path loss exponent.Seven different propagation scenari⁃os(S1 through S7)in which compliant WBAN devices may op⁃erate were identified in the IEEE 802.15.6 standard[47](see Table 2).

These scenarios are determined according to whether theWBAN nodes[5]are implanted inside the body;on the body surface，i.e.，in contact with the skin or 2 cm from the skin;or external，i.e.，between 2 cm and 5 m from the body.

Much research has been done on characterizing the WBAN channel for body⁃surface communication at different frequen⁃

▼Table 2.Propagation scenarios for WBAN communications[47]

cies for NB and UWB signals.On the other hand，there are sig⁃nificantly fewer channel models for implant communication，in part because it is impossible to perform in⁃body measurements on people，and electromagnetic simulation tools are expensive.

In the following，we summarize different WBAN path⁃loss models.We survey the literature emphasizing implant commu⁃nication because this topic requires more attention.

4.1 Channel Models for Implant Communications

Scenarios S1，S2，and S3 correspond to implanted BAN nodes.Two IEEE 802.15.6 channel models，CM1 and CM2，can be used to characterize the propagation scenarios for im⁃plant nodes.

4.1.1 Narrowband Signals

CM1 and CM2 have been developed for 402-405 MHz，which is allocated to medical implant communication services in many parts of the world.This band offers good propagation through human tissue and enables the use of reasonable⁃size antennas[48].However，its limited bandwidth constrains the communication devices to low transmission rates.The models are described by(1)and(2)，with the scattering term being a normally distributed random variable.The parameters of these models for deeply implanted and body⁃surface WBAN devices can be found in[47].An approximation of S3 can be obtained by combining S2 with S6 or S7.These models were the result of highly innovative research based on a 3D immersive visual⁃ization and simulation platform that included frequency⁃depen⁃dent dielectric properties of more than 300 parts of the male anatomy[49].

Other propagation models for implants using 418 MHz and 916.5 MHz were presented in[50].In the higher frequency band，the loss was greater than expected.In[51]，the propaga⁃ tion loss of antennas implanted in the body was computed us⁃ing numerical simulations for industrial⁃scientific⁃medical (ISM)frequency bands.These bands were 433 MHz，915 MHz，2450 MHz，and 5800 MHz.However，the simulations were done with simplistic single⁃layer and triple⁃layer tissue struc⁃tures，and no mathematical formulas for path loss were provid⁃ed.In[52]，numerical and experimental path loss were investi⁃gated using ingested wireless implants in 402 MHz，868 MHz，and 2.4 GHz.Likewise，measurements were taken in phantoms (chemical solutions specially formulated to mimic the dielec⁃tric properties of human tissues)，and path loss was numerical⁃ly simulated for insulated dipole antennas in the ISM band at 2.457 GHz in order to derive formulas for a propagation path of up to 80 mm[53].However，as in the previous case，the homo⁃geneous propagation scenario was very simplistic.

More accurate channel models for implant communication in different ISM frequency bands have not yet been reported in the literature.There are opportunities to innovate in realistic anatomical voxel models and in⁃vivo measurements on animals to characterize path loss[54].

4.1.2 Ultrawideband Signals

There have been limited attempts to model the implant prop⁃agation channel for UWB signals.Only two models for WBAN devices implanted in the human chest have been reported.The first model[55]was developed through numerical simulations using a voxel anatomical model that included nearly 50 types of tissue with a spatial resolution of 2 mm.This model predicts a root mean square(RMS)delay spread of around 0.2 ns.The second model[56]predicts an RMS delay spread of less than 1 ns，which agrees with the results in[55]，and the path loss is given as

whereais a fitting constant，nis an empirical exponent，and Sis a normally distributed RV.This formula does not have the same form as the Friis equation，but it is a better fit to the data obtained from numerical simulations using an anatomical mod⁃el that is not homogenous and includes 24 different kinds of tis⁃sues with voxel resolution of 2 mm.A similar simulation ap⁃proach as that in[56]was taken to obtain a UWB path⁃loss model for the abdominal region[57].In⁃vivo experiments[58]，[59]have demonstrated the feasibility of high⁃data⁃rate trans⁃mission over implant UWB links，and further research in this area is encouraged，especially in the characterization of fre⁃quency⁃dependent loss that cannot be neglected in UWB chan⁃nels[60].

4.2 Channel Models for Body-Surface Communication

In 802.15.6，body surface to body surface(BS2BS)links over 5-50 MHz are established by using the human body as the communication medium.The HBC channel comprises thefrequency response and noise.BS2BS in 400 MHz，600 MHz，900 MHz，2.4 GHz，and 3.1-10.6 GHz are described by(2)，and the corresponding parameters for hospital room and an⁃echoic chamber measurements are found in[47].Alternative and more detailed channel models are described in[50].

5 Security

Because many signals transported in a BAN are considered biometric data，it is important to protect this data from being observed and analyzed by unwanted parties.Encryption of sen⁃sitive data is one solution.The encryption algorithm of choice depends on how much complexity the system allows.Data from the fusion center and network coordinator can be easily en⁃crypted using asymmetric cryptography a fusion center and net⁃work coordinator allows more complex algorithms，such as RSA，to be implemented.For sensor nodes，especially im⁃plants，the situation is different because of complexity.Sym⁃metric cryptography solutions，especially stream ciphers，can be uncomplicated.

5.1 Cipher Security

The security of a cipher depends on several factors but is generally quantified by equivocation and work characteristic.

5.1.1 Equivocation

Equivocation is the uncertainty surrounding what was trans⁃mitted(source data)and the encryption key after an unknown party，referred to as a cryptanalyst(CA)，has observed the en⁃crypted data stream.Equivocation can be expressed in terms of conditional entropies:Let X denote a vector ofNχsamples from the data source and K denote a vector ofNkkey sam⁃ples.Equivocation is then given byH(X|Y)andH(K|Y)，where Y is a collection ofNχencrypted samples available to the CA.A cipher is considered secure if no information about the transmitted vector or key has been revealed when the enci⁃phered vector has been observed，i.e.，whenH(K|Y)=H(K) andH(X|Y)=H(X).When these conditions are there is perfect secrecy or strongly ideal secrecy[61].

Perfect secrecy is achieved for arbitrary distribution on X as long as the key is uniformly distributed(completely random) overNk≥Nχsamples.That is，the key cannot repeat，and its length is the same as that of the data sequence transmitted.For sensors monitoring medical conditions over hours and even days，perfect secrecy is impractical.

Strongly ideal secrecy can be achieved for a key that repeats several times over the data sequence as long as that data se⁃quence has no redundancy and is uniformly distributed.

In practice，especially with a complexity constraint，it may be difficult to remove all redundancy in the data sequence.If Nk

5.1.2 Work Characteristic

AlthoughNχ≥NUP，the CA may still have problems isolat⁃ing the correct solution to the cipher because this solution may be difficult to compute.The computational work that the CA must do to break the cipher is called the work characteristic. To ensure a high work characteristic，it is necessary to create mathematical problems that are known to be computationally difficult to solve，e.g.，factorization of prime numbers.Apply⁃ing chaotic maps prior to the encryption algorithm confuses the CA and makes it difficult to collect the statistics needed to break the cipher.Some chaotic maps，such as cat maps[62] are easy to implement.

5.2 Low-Complexity Symmetric Cryptography

A good low⁃complexity enciphering system should contain redundancy removal，map from given data distribution to a uni⁃form distribution，chaotic map，and simple encryption method. 5.2.1 Simple Encryption Method

We assume that each source sample is limited so that χi∈[-A，A]，and each key sample is uniformly distributed over [-A，A].

Medical signals contain redundancy as correlation between consecutive samples.By removing this correlation，both com⁃pression and higher security can be achieved.There are many methods for removing correlation，some of which are men⁃tioned in subsection 2.4.1.

A map from an arbitrary source distribution to a uniform dis⁃tribution is relatively easy to create.If the source densitypχ(χ) is monovalent and symmetric around 0，then

is uniformly distributed over[] -A，A.For a vector，each sam⁃ple can be transformed through this function.If correlation has been removed，the output will have a jointly uniform distribu⁃tion at the output.

Many simple symmetric schemes can be implemented.One such scheme is the Shannon cipher[63]for encrypting analog information sources.

This cipher is very uncomplicated and has been shown to provide both perfect secrecy[63]and strongly ideal secrecy [64]under the conditions mentioned in subsection 5.1.1.

In a more practical situation，where there is some redundan⁃cy/structure left in the signal，the UP indicates the security of the algorithm and must be determined for relevant data.We consider a first⁃order Gaussian autoregressive process AR(1) with correlationρχbetween consecutive samples.If each sam⁃

ple is mapped through(4)，the lower bound of the UP is

Fig.3 shows the lower bound of the UP for AR(1)as a func⁃tion of correlation when 1)the signal(blue)is encrypted direct⁃ly，and 2)the transformed vector(green)is encrypted.

By reducing correlation down to about 0.2，which is possible in practice，the key has to repeat at least 100 times before the CA can break the cipher.Direct encryption is insufficient:the key can only repeat a couple of times before UP is reached in⁃dependent ofρχ.It is important to map the data sequence through(4)before encryption.

▲Figure 3.Key repetition rate as a function of correlation.

5.2.2 Security and Fidelity

If Shannon’s cipher(5)is applied to noisy channels，both source and key must be quantized prior to encryption;other⁃wise，large decoding errors will corrupt the reconstructed sig⁃nal.The larger the noise is，the larger the quantizer step⁃size should be.The fidelity in the reconstruction is therefore re⁃duced whenever the noise increases.

In[65]，the authors argue that there is a tradeoff between se⁃curity and fidelity when analog signals are encrypted.Fidelity drops because of increased quantizer step size，and this means that more secure encryption is possible because fewer samples need to be transmitted when the signal is heavily compressed. Another reason why there is a tradeoff between security and fi⁃delity whenNk

5.2.3 Key Distribution

A problem with symmetric key cryptography is secure distri⁃ bution of the key(s)being used.However，for medical BAN configured at the hospital，several keys can be pre⁃stored and changed by a simple rule decided by the person responsible for installing the medical sensors.If the UP is high，relatively few keys need to be installed.If a high enough UP is not possi⁃ble，chaotic maps can be implemented to increase the work characteristic.

6 Future Challenges

Recent advancements in microelectronics make it possible to have a WBAN with numerous sensor nodes，and high⁃speed connectivity is possible with multiple antennas.As WBANs be⁃come bigger，they will encounter problems in terms of multi⁃hop routing，end⁃to⁃end delay，and service provisioning with priority⁃transmission.All these features will have to be care⁃fully optimized in a system with constraints such as small foot⁃print，limited power，and limited computational possibilities.

WBANs can be arbitrarily deployed，and this leads to inter⁃ference.For example，passengers with WBANs in a confined area，such as a bus，will experience mutual interference.It has to be determined whether cognitive radio networking technolo⁃gy can help mitigate interference in WBANs[62].Another in⁃teresting research problem is the possibility of using WBAN⁃to⁃WBAN interaction to improve connectivity and communication service.However，for this to be achieved，it has to be deter⁃mined whether different WBANs can communicate and ex⁃change private(possibly sensitive)medical information for the mutual benefit of the WBANs.Privacy and access control for WBAN⁃to⁃WBAN communications both have to be extensively researched.Multiple sensors operating at different rates and having different priorities in terms of mobility raise difficult problems.Such a system needs to treat PHY，MAC，and net⁃work layers in a cross⁃layer manner.

7 Conclusion

In this paper，we have described the different aspects of a WBAN，including sensors and their requirements in terms of data rate and power consumption.We have discussed signal⁃processing techniques，such as compression，detection and lo⁃calization，in mobile implant sensors.The PHY layers of 802.15.6，and body⁃surface and in⁃body channel modeling were described for the frequencies specified in 802.15.6.Possi⁃ble security techniques for WBANs were also discussed.

[1]S.Ullah，H.Higgins，B.Braem，B.Latre，C.Blondia，I.Moerman,et al.，“A com⁃prehensive survey of wireless body area networks，”J.of Medical Syst.,vol.36，no.3，pp.1065-1094，Jun.2012.doi:10.1007/s10916⁃010⁃9571⁃3.

[2]B.Latré，B.Braem，I.Moerman，C.Blondia，and P.Demeester，“A survey on wireless body area networks，”Wireless Networks,vol.17，no.1，pp.1-18，Jan. 2011.doi:10.1007/s11276⁃010⁃0252⁃4.

[3]M.Chen，S.Gonzalez，A.Vasilakos，H.Cao，and V.C.Leung，“Body area net⁃works:A survey，”Mobile Networks and Applicat.,vol.16，no.2，pp.171-193，Apr.2011.doi:10.1007/s11036⁃010⁃0260⁃8.

[4]R.Cavallari，F.Martelli，R.Rosini，C.Buratti，and R.Verdone，“A survey on wireless body area networks:technologies and design challenges，”IEEE Com⁃mun.Surveys&Tutorials,to be published.

[5]I.S.Association，“IEEE standard for local and metropolitan area networks⁃part 15.6:wireless body area networks，”IEEE Standard for Inform.Technology,vol. 802，pp.1-271，2012.

[6]M.M.Baig，H.Gholamhosseini，and M.J.Connolly，“A comprehensive survey of wearable and wireless ECG monitoring systems for older adults，”Medical&bio⁃logical Eng.&computing,vol.51，no.5，pp.485-495，May 2013.doi:10.1007/ s11517⁃012⁃1021⁃6.

[7]S.Arnon，D.Bhastekar，D.Kedar，and A.Tauber，“A comparative study of wire⁃less communication network configurations for medical applications，”IEEE Wireless Commun.,vol.10，no.1，pp.56-61，Feb.2003.doi:10.1109/ MWC.2003.1182112.

[8]B.Gyselinckx，C.Van Hoof，J.Ryckaert，R.F.Yazicioglu，P.Fiorini，and V.Le⁃onov，“Human++:autonomous wireless sensors for body area networks，”in Proc. of the IEEE 2005 on Custom Integrated Circuits Conf.，2005 San Jose，USA，pp. 13-19.doi:10.1109/CICC.2005.1568597.

[9]J.F.Rizzo，J.L.Wyatt Jr，and L.Theogarajan，“Minimally invasive retinal pros⁃thesis，”US 6976998 B2，Dec.20，2005.

[10]J.Viby⁃Mogensen，“Neuromuscular monitoring，”Current Opinion in Anesthesi⁃ology,vol.14，no.6，pp.655-659，Dec.2001.

[11]P.Bonato，P.J.Mork，D.M.Sherrill，and R.H.Westgaard，“Data mining of motor patterns recorded with wearable technology，”IEEE Eng.Med.Biol. Mag.,vol.22，no.3，pp.110-119，2003.doi:10.1109/MEMB.2003.1213634.

[12]F.Adochiei，C.Rotariu，R.Ciobotariu，and H.Costin，“A wireless low⁃power pulse oximetry system for patient telemonitoring，”in 7th Int.Symp.on Ad⁃vanced Topics in Elect.Eng.(ATEE)，Bucharest，Romania，May 2011，pp.1-4.

[13]T.R.Burchfield and S.Venkatesan，“Accelerometer⁃based human abnormal movement detection in wireless sensor networks，”in HealthNet'07，San Juan，Puerto Rico，pp.67-69.doi:10.1145/1248054.1248073.

[14]N.L.Keijsers，M.W.Horstink，and S.C.Gielen，“Automatic assessment of le⁃vodopa-induced dyskinesias in daily life by neural networks，”Movement disor⁃ders,vol.18，no.1，pp.70-80，2003.

[15]Y.Bar⁃Shalom，X.R.Li，and T.Kirubarajan，Estimation with Applications to Tracking and Navigation:Theory Algorithms and Software，New York:John Wiley&Sons，2004.

[16]B.Moussakhani，J.T.Flåm，S.Støa，I.Balasingham，and T.Ramstad，“On lo⁃calisation accuracy inside the human abdomen region，”IET Wireless Sensor Syst.,vol.2，no.1，pp.9-15，2012.

[17]C.Hu，M.Li，S.Song，R.Zhang，and M.⁃H.Meng，“A cubic 3⁃axis magnetic sensor array for wirelessly tracking magnet position and orientation，”IEEE Sensors J.，vol.10，no.5，pp.903-913，May 2010.doi:10.1109/ JSEN.2009.2035711.

[18]R.Chávez⁃Santiago，A.Khaleghi，I.Balasingham，and T.A.Ramstad，“Archi⁃tecture of an ultra wideband wireless body area network for medical applica⁃tions，”in 2nd Int.Symp.Applied Sciences in Biomedical and Commun.Technol⁃ogies，Bratislava，Slovakia，2009，pp.1-6.doi:10.1109/ISA⁃BEL.2009.5373624.

[19]I.F.Akyildiz，W.Su，Y.Sankarasubramaniam，and E.Cayirci，“A survey on sensor networks，”IEEE Commun.Mag.,vol.40，no.8，pp.102-114，Aug. 2002.doi:10.1109/MCOM.2002.1024422.

[20]V.Leonov，P.Fiorini，S.Sedky，T.Torfs，and C.Van Hoof，“Thermoelectric MEMS generators as a power supply for a body area network，”in 13th Int. Conf.Solid⁃State Sensors,Actuators and Microsystems,2005.Digest of Tech.Pa⁃pers，Seoul，South Korea，vol.1，pp.291-294.doi:10.1109/SEN⁃SOR.2005.1496414.

[21]M.Wang，J.Cao，M.Liu，B.Chen，Y.Xu，and J.Li，“Design and implementa⁃tion of distributed algorithms for WSN⁃based structural health monitoring，”Int.J.Sensor Networks,vol.5，no.1，pp.11-21，Feb.2009.doi:10.1504/IJS⁃NET.2009.023312.

[22]S.Xiao，A.Dhamdhere，V.Sivaraman，and A.Burdett，“Transmission power control in body area sensor networks for healthcare monitoring，”IEEE J.Sel. Areas Commun.,vol.27，no.1，pp.37-48，Jan.2009.doi:10.1109/ JSAC.2009.090105.

[23]T.Srisooksai，K.Keamarungsi，P.Lamsrichan，and K.Araki，“Practical data compression in wireless sensor networks:A survey，”J.Network and Computer Applicat.,vol.35，no.1，pp.37-59，Jan.2012.doi:10.1016/j.jn⁃ca.2011.03.001.

[24]A.Oka and L.Lampe，“Energy efficient distributed filtering with wireless sen⁃sor networks，”IEEE Trans.Signal Process.,vol.56，no.5，pp.2062-2075， May 2008.doi:10.1109/TSP.2007.911496.

[25]J.A.Gubner，“Distributed estimation and quantization，”IEEE Trans.Inf.The⁃ory，vol.39，no.4，Jul.1993.doi:10.1109/18.243470.

[26]J.Chou，D.Petrovic，and K.Ramachandran，“A distributed and adaptive sig⁃nal processing approach to reducing energy consumption in sensor networks，”in INFOCOM 2003，San Francisco，USA，vol.2，pp.1054-1062.doi:10.1109/ INFCOM.2003.1208942.

[27]D.L.Donoho，“Compressed sensing，”IEEE Trans.Inf.Theory,vol.52，no.4，pp.1289-1306，Apr.2006.doi:10.1109/TIT.2006.871582.

[28]H.Rauhut，K.Schnass，and P.Vandergheynst，“Compressed sensing and re⁃dundant dictionaries，”IEEE Trans.Inf.Theory,vol.54，no.5，pp.2210-2219，May 2008.doi:10.1109/TIT.2008.920190.

[29]J.A.Tropp and A.C.Gilbert，“Signal recovery from random measurements via orthogonal matching pursuit，”IEEE Trans.Inf.Theory,vol.53，no.12，pp. 4655-4666，2007.doi:10.1109/TIT.2007.909108.

[30]L.F.Polania，R.E.Carrillo，M.Blanco⁃Velasco，and K.E.Barner，“Com⁃pressed sensing based method for ECG compression，”in IEEE Int.Conf. Acoustics,Speech and Signal Processing(ICASSP)，Prague，Czech Republic，2011，pp.761-764.doi:10.1109/ICASSP.2011.5946515.

[31]A.N.Kim，T.A.Ramstad，and I.Balasingham，“Very low complexity low rate image coding for the wireless endoscope，”in Proc.4th Int.Symp.Applied Sci⁃ences in Biomedical and Commun.Technologies，2011，p.90.doi:10.1145/ 2093698.2093788.

[32]P.A.Floor，I.Balasingham，T.A.Ramstad，E.Meurville，and M.Peisino，“Compression schemes for in⁃body and on⁃body UWB sensor networks，”in 3rd Int.Symp.Applied Sciences in Biomedical and Commun.Technologies(ISA⁃BEL)，Rome，Italy，2010，pp.1-5.doi:10.1109/ISABEL.2010.5702843.

[33]H.T.Nguyen，T.A.Ramstad，and I.Balasingham，“Wireless sensor communi⁃cation system based on direct⁃sum source coder，”IET Wireless Sensor Syst., vol.1，pp.96-104，2011.doi:10.1049/iet⁃wss.2010.0094.

[34]L.Smital，M.Vítek，J.Kozumplík，and I.Provaznik，“Adaptive Wavelet Wie⁃ner Filtering of ECG Signals，”IEEE Trans.Biomed.Eng.,vol.60，no.2，pp. 437-445，Feb.2013.doi:10.1109/TBME.2012.2228482.

[35]K.A.Germansky and D.A.Leffler，“Best practice&research clinical gastro⁃enterology，”Best Practice&Research Clinical Gastroenterology,vol.25，pp. 387-395，2011.

[36]B.Moussakhani，R.Chavez⁃Santiago，and I.Balasingham，“Multi model track⁃ing for localization in wireless capsule endoscopes，”in Proc.4th Int.l Symp. Applied Sciences in Biomedical and Commun.Technologies，Spain，2011，p. 159.doi:10.1145/2093698.2093857.

[37]Y.⁃J.Min，H.⁃K.Kim，Y.⁃R.Kang，G.⁃S.Kim，J.Park，and S.⁃W.Kim，“Design of Wavelet⁃Based ECG Detector for Implantable Cardiac Pacemakers，”IEEE Trans.Biomed.Circuits Syst.,vol.7，no.4，pp.426-436，Aug.2013.doi: 10.1109/TBCAS.2012.2229463.

[38]M.S.Manikandan and K.P.Soman，“A novel method for detecting R⁃peaks in electrocardiogram(ECG)signal，”Biomedical Signal Processing and Control, vol.7，no.2，pp.118-128，Mar.2012.doi:10.1016/j.bspc.2011.03.004.

[39]S.Kadambe，R.Murray，and G.F.Boudreaux⁃Bartels，“Wavelet transform⁃based QRS complex detector，”IEEE Trans.Biomed.Eng.,vol.46，no.7，pp. 838-848，Jul.1999.doi:10.1109/10.771194.

[40]A.P.Stylianou，C.W.Luchies，and M.F.Insana，“EMG onset detection using the maximum likelihood method，”in Proc.2003 Summer Bioengineering Conf.，Key Biscayne，USA，pp.1075-1076.

[41]M.Hernandez and R.Miura，“Coexistence of IEEE Std 802.15.6 TM⁃2012 UWB⁃PHY with other UWB systems，”in IEEE Int.Conf.Ultra⁃Wideband (ICUWB)，Syracuse，USA，2012，pp.46-50.doi:10.1109/ICU⁃WB.2012.6340496.

[42]S.Ullah，M.Mohaisen，and M.A.Alnuem，“A review of ieee 802.15.6 MAC，PHY，and security specifications，”Int.J.Distributed Sensor Networks,vol. 2013，2013.doi:dx.doi.org/10.1155/2013/950704.

[43]B.Choi，B.Kim，S.Lee，K.Wang，Y.Kim，and D.Chung，“Narrowband physi⁃cal layer design for WBAN system，”in First Int.Conf.Pervasive Computing Signal Processing and Applicat.(PCSPA)，Harbin，China，2010，pp.154-157. doi:10.1109/PCSPA.2010.46.

[44]H.T.Nguyen，T.A.Ramstad，and I.Balasingham，“Coded pulse position mod⁃ulation communication system over the human abdominal channel for medical wireless body area networks，”in IEEE 23rd Int.Symp.Personal Indoor and Mobile Radio Commun.(PIMRC)，Sydney，Australia，2012，pp.1992-1996. doi:10.1109/PIMRC.2012.6362680.

[45]K.S.Kwak，S.Ullah，and N.Ullah，“An overview of IEEE 802.15.6 stan⁃dard，”in 3rd Int.Symp.Applied Sciences in Biomedical and Communication Technologies(ISABEL)，Rome，Italy，2010，pp.1-6.doi:10.1109/ISA⁃BEL.2010.5702867.

[46]M.Seyedi，B.Kibret，D.T.Lai，and M.Faulkner，“A survey on intrabody com⁃munications for body area network applications，”IEEE Trans.Biomed.Eng., vol.60，no.3，pp.2067-2079，Aug.2013.doi:10.1109/TBME.2013.2254714.

[47]K.Y.Yazdandoost and K.Sayrafian⁃Pour，“Channel model for body area net⁃work(BAN)，”IEEE P802,vol.15，2009.

[48]P.Soontornpipit，C.M.Furse，and Y.C.Chung，“Design of implantable mi⁃crostrip antenna for communication with medical implants，”IEEE Trans.Mi⁃crow.Theory Techn.,vol.52，no.8，pp.1944-1951，Aug.2004.

[49]K.Sayrafian⁃Pour，W.⁃B.Yang，J.Hagedorn，J.Terrill，K.Y.Yazdandoost，and K.Hamaguchi，“Channel models for medical implant communication，”Int.J. Wireless Inform.Networks,vol.17，no.3-4，pp.105-112，Dec.2010.doi: 10.1007/s10776⁃010⁃0124⁃y.

[50]W.G.Scanlon，B.Burns，and N.E.Evans，“Radiowave propagation from a tis⁃sue⁃implanted source at 418 MHz and 916.5 MHz，”IEEE Trans.Biomed. Eng.,vol.47，no.4，pp.527-534，Apr.2000.doi:10.1109/10.828152.

[51]J.Gemio，J.Parron，and J.Soler，“Human body effects on implantable anten⁃nas for ISM bands applications:models comparison and propagation losses study，”Progress in Electromagnetics Research,vol.110，pp.437-452，2010. doi:10.2528/PIER10102604.

[52]A.Alomainy and Y.Hao，“Modeling and characterization of biotelemetric ra⁃dio channel from ingested implants considering organ contents，”IEEE Trans. Antennas Propag.,vol.57，no.4，pp.999-1005，Apr.2009.doi:10.1109/ TAP.2009.2014531.

[53]D.Kurup，W.Joseph，G.Vermeeren，and L.Martens，“Path loss model for in⁃body communication in homogeneous human muscle tissue，”Electronics let⁃ters,vol.45，no.9，pp.453-454，2009.

[54]R.Chavez⁃Santiago，K.Sayrafian⁃Pour，A.Khaleghi，K.Takizawa，J.Wang，I. Balasingham，and H⁃B Li，“Propagation models for IEEE 802.15.6 standard⁃ization of implant communication in body area networks，”IEEE Commun. Mag.,vol.51，no.8，Aug.2013.doi:10.1109/MCOM.2013.6576343.

[55]J.Wang and Q.Wang，“Channel modeling and BER performance of an im⁃plant UWB body area link，”in 2nd Int.Symp.Applied Sciences in Biomedical and Commun.Technologies，Bratislava，Slovakia，2009，pp.1-4.doi:10.1109/ ISABEL.2009.5373707.

[56]A.Khaleghi，R.Chávez⁃Santiago，and I.Balasingham，“Ultra⁃wideband statis⁃tical propagation channel model for implant sensors in the human chest，”IET Microwaves,Antennas&Propagation,vol.5，no.15，pp.1805-1812，Dec. 2011.doi:10.1049/iet⁃map.2010.0537.

[57]S.Støa，R.Chavez⁃Santiago，and I.Balasingham，“An ultra wideband commu⁃nication channel model for the human abdominal region，”in IEEE GLOBE⁃COM Workshops(GC Wkshps)，Miami，USA，2010，pp.246-250.doi:10.1109/ GLOCOMW.2010.5700319.

[58]R.Chavez⁃Santiago，I.Balasingham，J.Bergsland，W.Zahid，K.Takizawa，R. Miura，and H⁃B Li，“Experimental implant communication of high data rate video using an ultra wideband radio link，”in 35th Annu.Int.Conf.IEEE on Eng.in Medicine and Biology Society(EMBC)，Osaka，Japan，2013，pp.5175-5178.doi:10.1109/EMBC.2013.6610714.

[59]D.Anzai，K.Katsu，R.Chavez⁃Santiago，Q.Wang，D.Plettemeier，J.Wang，and I.Balasingham，“Experimental evaluation of implant UWB⁃IR transmis⁃sion with living animal for body area networks，”IEEE Trans.Microw.Theory Techn.,vol.62，no.1，pp.183-192，Jan.2014.doi:10.1109/TM⁃TT.2013.2291542.

[60]A.Khaleghi，R.Chávez⁃Santiago，and I.Balasingham，“An improved ultra wideband channel model including the frequency⁃dependent attenuation for in⁃body communications，”in Annu.Int.Conf.IEEE on Eng.in Medicine and Biol⁃ogy Society(EMBC)，San Diego，USA，2012，pp.1631-1634.doi:10.1109/EM⁃BC.2012.6346258.

[61]C.E.Shannon，“Communication Theory of Secrecy Systems，”Bell Syst.Tech. J.,vol.28，no.4，pp.656-715，1949.

[62]R.Chávez⁃Santiago，K.E.Nolan，O.Holland，L.De Nardis，J.M.Ferro，N. Barroca，L.M.Borges，F.J.Velez，V.Goncalves，and I.Balasingham，“Cogni⁃tive radio for medical body area networks using ultra wideband，”IEEE Wire⁃less Commun.,vol.19，no.4，pp.74-81，Aug.2012.doi:10.1109/ MWC.2012.6272426.

[63]C.E.Shannon，“Analogue of the Vernam system for continuous time series，”in Claude Elwood Shannon:Collected Papers,Memorandum MM,Los Alami⁃tos，USA:IEEE Computer Society Press，1943，pp.43-110.

[64]P.A.Floor，I.Balasingham，and T.A.Ramstad，“Analysis of the Shannon⁃Ger⁃sho(SG)cipher，”internal document.

[65]A.Gersho，“Perfect secrecy encryption of analog signals，”IEEE J.Sel.Areas Commun.,vol.2，no.3，pp.460-466，May1984.doi:10.1109/ JSAC.1984.1146071.

Biographiesphies

Seyyed Hamed Fouladi(hamed.fouladi@iet.ntnu.no)received his B.S.degree in electrical engineering at the Department of Electrical Engineering，Shahed Universi⁃ty，Tehran，Iran in 2009 and M.Sc degree in Communication Systems at the Depart⁃ment of Electrical Engineering，Amirkabir University of Technology，Tehran，Iran，in 2012.He is currently a PhD student at Norwegian University of Science and Technology(NTNU)，Trondheim，Norway.His research interests include statistical signal and image processing，multi⁃resolution signal analysis，blind signal process⁃ing，statistical modeling，detection and estimation.

Raúl Chávez⁃Santiago(raul.chavez⁃santiago@rr⁃research.no)graduated as an Elec⁃tronics and Telecommunications Engineer at the National Polytechnic Institute，Mexico，in 1997.In 2001 he obtained a M.Sc.degree in Electrical Engineering at CINVESTAV，Mexico.He received the Ph.D.degree in Electrical and Computer En⁃gineering from Ben⁃Gurion University of the Negev，Israel，in 2007.Thereafter he held a postdoctoral position at the University Paris⁃Sud XI，France，where he inves⁃tigated radio resource management for OFDMA systems.He later held a second postdoctoral position at Bar⁃Ilan University，Israel.There，he researched the infor⁃mation theory aspects of ad hoc and cognitive radio networking.He joined the Inter⁃vention Centre，Oslo University Hospital，Norway，in 2009，where he currently in⁃vestigates short⁃range radio communication technologies for body area network (BAN)solutions.His research is focused on implant communications and ultra wide⁃band(UWB)technology.He is a Management Committee member in various Euro⁃pean COST Actions.

Pål Anders Floor(andflo@rr⁃research.no)received his B.Sc.degree from Gjøvik University College(HIG)，Norway in 2001，his M.Sc degree in 2003 and his PhD，degree in 2008，both from the Department of Electronics and Telecommunications，Norwegian University of Science and Technology(NTNU)，Trondheim，Norway.All three degrees are in electrical engineering.He was working as a Post.Doc.at the In⁃tervention Centre at Oslo University Hospital and at the Institute of Clinical Medi⁃cine at University of Oslo from 2008 to 2013.He is currently a Post.Doc at NTNU. His research interests include joint source⁃channel coding，information theory and signal processing applied on point⁃to⁃point links，in small and large networks，as well as in Neuroscience，and lightweight cryptography solutions for medical BAN.

Ilangko Balasingham(ilangkob@medisin.uio.no)received the M.Sc.and Ph.D.de⁃grees from the Department of Electronics and Telecommunications，Norwegian Uni⁃versity of Science and Technology(NTNU)，Trondheim，Norway in 1993 and 1998，respectively，both in signal processing.He performed his Master’s degree thesis at the Department of Electrical and Computer Engineering，University of California Santa Barbara，USA.From 1998 to 2002，he worked as a Research Scientist at Fast Search&Transfer ASA，Oslo，Norway，which is now part of Microsoft Inc.Since 2002 he has been with the Intervention Center，Oslo University Hospital，Oslo，Nor⁃way，where he heads the Wireless Sensor Network Research Group.He was appoint⁃ed as a Professor in Signal Processing in Medical Applications at NTNU in 2006. His research interests include wireless body sensor networks，microwave sensing and imaging，short range localization and tracking，and nano⁃neural communication networks.He has authored or co⁃authored 168 papers and has been active in orga⁃nizing special sessions，workshops，and conferences.

Tor A.Ramstad(ramstad@iet.ntnu.no)is Professor Emeritus at the Norwegian Uni⁃versity of Science and Technology(NTNU formerly NTH)，Norway.He got his MSc PhD degrees from NTH in 1968 and 1972，where he was assistant and associate pro⁃fessor until he became a full professor of Communication Theory in 1983，and re⁃tired in 2012.He has been a visiting professor at UCSB，Georgia Tech，and Eure⁃com，France.He was Associate Editor of IEEE Acoustics，Speech and Signal Pro⁃cessing，and chair of the IEEE Signal Processing Workshop in 1996.He is a mem⁃ber of the Norwegian Academy of Technological Sciences，and was awarded the Honorary Price from the Norwegian Signal Processing Society.Professor Ramstad has had a leading role in establishing and developing the field of digital signal pro⁃cessing in Norway.His main interests include efficient digital filtering methods and implementations，signal compression of speech，images and video，signal interpola⁃tion for sample rate conversion，and joint source⁃channel coding.

t

2014⁃04⁃04

10.3969/j.issn.1673-5188.2014.03.001

http://www.cnki.net/kcms/detail/34.1294.TN.20140822.1138.001.html,published online 22 August,2014

The research work was performed,in part,of the research project Medical sensing,localization and communications using ultra wideband

technology(MELODY)contract no.285885,and Adaptive Security for

Smart Internet of Things in eHealth(ASSET)contract no.213131,which

both are funded by the Research Council of Norway.