Wireless Sensor Network Based Home Monitoring ... - IEEE Xplore

IEEE SENSORS JOURNAL, VOL. 12, NO. 6, JUNE 2012

1965

Wireless Sensor Network Based Home Monitoring System for Wellness Determination of Elderly Nagender Kumar Suryadevara, Student Member, IEEE, and Subhas Chandra Mukhopadhyay, Fellow, IEEE

Abstract— Wireless-sensor-network-based home monitoring system for elderly activity behavior involves functional assessment of daily activities. In this paper, we reported a mechanism for estimation of elderly well-being condition based on usage of house-hold appliances connected through various sensing units. We defined two new wellness functions to determine the status of the elderly on performing essential daily activities. The developed system for monitoring and evaluation of essential daily activities was tested at the homes of four different elderly persons living alone and the results are encouraging in determining wellness of the elderly. Index Terms— Activities of daily living, elder care, home monitoring, smart home, wellness, wireless sensor network.

I. I NTRODUCTION

A

NORMAL person performs daily activities at regular interval of time. This implies that the person is mentally and physically fit and leading a regular life. This tells us that the overall well-being of the person is at a certain standard. If there is decline or change in the regular activity, then the wellness of the person is not in the normal state. Elderly people desire to lead an independent lifestyle, but at old age, people become prone to different accidents, so living alone has high risks and is recurrent. A growing amount of research is reported in recent times on development of a system to monitor the activities of an elderly person living alone so that help can be provided before any unforeseen situation happened. In the present work, an intelligent home monitoring system based on ZigBee wireless sensors network has been designed and developed to monitor and evaluate the well-being of the elderly living alone in a home environment. Wellness of elderly can be evaluated for forecasting unsafe situations during monitoring of regular activities. The developed system is intelligent, robust and does not use any camera or vision sensors as it intrudes privacy. Based on a survey among elderly we find that it has a huge acceptability to be used at home due to non use of the camera or vision based sensors. The intelligent software, along with the electronic system, can monitor the usage of different household appliances

Manuscript received November 13, 2011; revised December 18, 2011; accepted December 26, 2011. Date of publication January 3, 2012; date of current version April 25, 2012. The associate editor coordinating the review of this paper and approving it for publication was Dr. V. R. Singh. The authors are with the School of Engineering and Advanced Technology, Massey University, Palmerston North 5301, New Zealand (e-mail: [email protected]; [email protected]). Color versions of one or more of the figures in this paper are available online at http://ieeexplore.ieee.org. Digital Object Identifier 10.1109/JSEN.2011.2182341

and recognize the activities to determine the well-being of the elderly. Also, the system interprets all the essential elderly activities such as preparing breakfast/lunch/dinner, showering, rest-room use, dinning, sleeping and self grooming. Basically, the system function based on the usage data of electrical and non-electrical appliances within a home. The system has been developed to be used in an existing home occupied by an elderly rather than a test bed scenario or for a newly built house. At the hardware level, wireless sensor network with ZigBee components are connected in the form of mesh topology, and a central coordinator of the sensing units collect data from the sensors connected to various appliances. The developed software system continuously reads the data from the coordinator and efficiently stores on the system for further data processing in real time. The data processing involves steps for wellness check based on the knowledge of daily activities performed in conjunction with the usage of house-hold appliances, for predicting change in the daily activity pattern of the system. In this system, a required number of sensors for monitoring the daily activities of the elderly have been used. Increase of a number of sensors increases the cost of the system and may also complicate the installation issues. A variety of systems for monitoring and functional assessment for elderly care have been proposed and developed in recent times. Monitoring activities of the person based on camera based sensors are reported in [1, 2] where the images of the person are taken and analyzed. In real practice applications such as surveillance and security make full use of camera based system but for home monitoring activities it lacks a huge acceptability among the elderly. Other than camera, infrared based Small Motion Detectors (SMDs), passing sensors, operation detectors and IR motion sensors have been incorporated in the house for monitoring the human activity behaviour [3] and the interpretation of human activity is limited to only to a few human activities. There are a number of projects available on wearable health devices [4, 5] personal wellness monitoring and safety [6] integrated with sensors to provide continuous monitoring of person’s health related issues and activity monitoring. Also, systems using RFID communication technology in elderly center were introduced [7, 8]. Though these devices are for specific purposes, they have severe concerns related to security, privacy and legal aspects [9]. Usually people are reluctant to wear a system continuously on their body. So it may not be a viable option for a healthy elderly people. This situation may be acceptable for a patient under rehabilitation.

1530–437X/$31.00 © 2012 IEEE

1966


If many sensors can be installed for the monitor of all appliances used by the elderly in a newly constructed house, it provides necessary data for elderly monitoring [10]. This may not be possible in most practical scenario as the elderly usually lives in their homes, which were built, during their young age. Hence the elderly houses are basically old and existing houses. Systems like remote human monitoring using wireless sensor networks [11, 12] were introduced in recent times. Software systems with different machine learning techniques are incorporated into the wireless systems like [13]. Also, monitoring and modeling of elderly activities of daily living were incorporated [14, 15]. Though technology is effectively implemented but, these systems are limited to a few activities. There is a huge demand for an electronic system with intelligent mechanism, low cost, flexible, easy to install, robust and accurate for monitoring basic Activities of Daily Living (ADLs) of elderly living alone so that help can be provided at the right time. The ultimate goal of personal wellness systems is to provide care for elderly people in the right time no matter where they live, but technology could assist with transitions from one level of care to the next and help prevent premature placement in expensive assistance domains [16]. Activity recognition and Wellness determination are two important functions to be done in a timely manner rather than offline. Hence, real-time processing of data is a must for recognizing activity behaviour and predicting abnormal situations of the elderly. To deal with issues such as monitoring the daily activities, performance tracking of normal behaviour and well-being of the elderly living alone a system which is noninvasive, flexible, low-cost and safe to use is designed and developed. An initial decline or change in regular daily activities can be identified by the home monitoring system and trigger messages to the appropriate care provider about the changes in the functional abilities of the elderly person.

Fig. 1.

Fabricated sensing unit with ZigBee module.

Fig. 2. Electrical appliance monitoring units connected to various house-hold appliances.

Processing circuit

ADC-ZigBee module

Current transformer Main power supply

Electrical appliance

Fig. 3. Block diagram representation of interfacing current sensor with ZigBee module.

II. S YSTEM D ESCRIPTION The system consists of two basic modules as developed in [17, 18]. At the low level module, Wireless sensor network integrated with Zigbee modules of mesh structure exists capturing the sensor data based on the usage of house-hold appliances and stores data in the computer system for further data processing. Collected sensor data are of low level information containing only status of the sensor as active or inactive and identity of the sensor. To sense the activity behaviour of elderly in real time, the next level software module will analyze the collected data by following an intelligent mechanism at various level of data abstraction based on time and sequence behaviour of sensor usage. The low level module consists of a number of sensors interconnected to detect usage of electrical devices, bed usage and chairs along with a panic button. The fabricated sensing unit as shown in Fig.1 communicates at 2.4GHz (Industrial Scientific and Medical band) through radio frequency protocols and

provides sensor information that can be used to monitor the daily activities of an elderly person. A smart sensor coordinator collects data from the sensing units and forward to the computer system for data processing. The major task of our work is to recognize the essential activities of daily living behaviour of the elderly through sensor fusion by using minimal sensors at elderly home. For this, WSN consisting of different types of sensors like electrical, force, contact sensors with zigbee module sensing units are installed at elderly home. The uses of electrical appliances are monitored by the electrical appliance monitoring sensing units as shown in Fig.2. These operate based on the detection of current flow connected to appliances such as microwave, water kettle, toaster, room Heater, television and dishwasher as they are regularly used by the elderly at home. Fig.3 illustrates inter-connection structure of developed current sensor unit

SURYADEVARA AND MUKHOPADHYAY: WSN BASED HOME MONITORING SYSTEM FOR WELLNESS DETERMINATION OF ELDERLY

1967

Temperature sensor Humidity sensor Light intensity sensor Associated indicator On/sleep LED indicator Monitoring LED indicators ZigBee chip (series 2) 3.3 V voltage regulator

Fig. 4. Sensing units connected to bed, toilet and chair house-hold appliances of elderly home. Bed/ couch/ chair/ toilet

Amplifier/ signal conditioning circuit

A D C

ZigBee module

Fig. 5. Block diagram representation of interfacing force sensor to ZigBee module.

Fig. 7. Developed ambient sensing unit for recording temperature, humidity, and light intensity.

to these actions were effectively recorded to recognize the corresponding activities. Designed and developed temperature, humidity and light intensity sensing unit with only one zigbee module as shown in Fig.7 is used to record the ambient readings of room for analyzing the data. Rationale for observing usage of house-hold appliances is based on the fact that these are regularly used by the elderly in various behaviours like preparation of food, during rest, toileting, sleeping and grooming activities. They are useful to determine the wellness of the person in performing these activities. Also, Emergency help and deactivate operations are made-up with zigbee module to facilitate the corresponding operations during the real-time activity monitoring of the elderly.

Fig. 6. Developed contact sensing units connected to grooming cabinet and fridge.

A. Data Acquisition with zigbee module for transmitting digital ON/OFF signal. The output of electrical appliance monitoring sensor unit is either ON or OFF based on the use of connected electrical appliance. Normally, one electric sensing unit is required to sense each electrical appliance. In order to have minimum sensing units to monitor more electrical appliances and reduce cost, the electrical appliance monitoring units are fabricated to support two electrical appliances on a single power inlet, having the intelligence to detect which particular device is on. We have tested by connecting water kettle and toaster appliances through different analog channels of ZigBee module to be monitored by single sensing unit thereby reducing the number of sensing units and cost for monitoring elderly activity behaviour. The system uses force sensors attached to bed, couch, toilet and dining chair as shown in Fig. 4 to monitor their daily usage. Based on the analog values of the force sensor received by the coordinator, the system can recognize the usage of these devices as active or inactive. Whenever the elderly person used these devices, developed system has monitored and recorded the event effectively for further data processing. Fig. 5 illustrates inter-connection structure of developed force sensor unit with zigbee module for transmitting analog value. Developed contact sensing units as shown in Fig.6 are fixed to the fridge and grooming cabinet of the elderly home to detect the open and close of the door operations. Events related

Captured data are dynamically changing and demanding fast, real-time response time for forecasting the irregular behaviour of the elderly. To analyze the data properly, an efficient process of storage mechanism of sensor data onto the computer system is executed. Issues like storage requirements for continuous flow of data streams and processing of data to generate patterns/abnormal events in real time were effectively dealt in the current system. Since there is a continuous in flow of sensor streams we have stored the sensor data in the processing system only when there is a change in the sensor events - Event based storage (i.e) when status (active or inactive) of the sensor is changed then the sensor fusion data is recorded. This is most efficient technique, as it reduces the size of storage to a large extent and more flexible for processing of data in real time. Event monitoring collection of data has enormous benefit over continuous flow collection of data in terms of the amount of data storage and processing of data in real-time applications like home monitoring. B. Activity Annotation Activity labeling for the activities of daily living of the elderly during real-time monitoring of appliances use is directly done with the help of ‘sensor events’. Activities like sleeping, preparing breakfast/lunch/dinner, dining, toileting and self grooming were recognized based on the Sensor-ID status and Time of the Day. Other activities like watching

1968


TABLE I A CTIVITY L ABELLING P ROCESS D URING RUN -T IME OF THE S YSTEM Connected to Appliance

Type of Sensor

Time of Usage

Annotated Activity

Bed

Pressure Sensor

09:00 pm to 06:00 am

Sleeping(SL)

Microwave / Oven/ Water kettle

Electrical Sensor

06:00 am to 10:00 am

Breakfast(BF)

10(Active)

Toilet

Pressure Sensor

Anytime

Toileting(TO)

2011-6-7 02:15:30 10 ON TO begin 2011-6-7 02:16:07 10 OFF TO end

19(Active)

Couch

Pressure Sensor

Anytime

Toileting(TO)

2011-6-8 05:20:45 19 ON RE begin 2011-6-8 05:35:30 19 OFF RE end

26(Active)

Grooming Cabinet

Contact

Anytime

Self Grooming(SG)

2011-6-8 09:20:10 26 ON SG begin 2011-6-8 09:22:40 26 OGG SG end

Sensor-ID/Status 18(Active)

11/12/13 (active)

Run Time Data18 2011-6-9 21:02:10 18 ON SL begin 2011-6-10 05:50:10 18 OFF SL end 2011-6-5 06:16:42 11 ON BF begin 2011-6-5 06:21:35 11 OFF BF end

of food between 6:00 am to 10 am been considered as preparation of breakfast. So sensor event generated in the kitchen between 6:00 am to 10:00 am used labeling as breakfast. We are interested in accuracy of the model to be build based on the activity annotation rather than accuracy of the activity annotation. Activity annotation is validated by cross checking with the ground truth (manually recording the events by the elderly). Fig. 8.

Sensor data acquisition and activity recognition in real-time.

C. Wellness Determination of Elderly television, preparing tea are done with the help of respective sequence patterns of sensor ids which are active. Fig.8 elucidates the flow of sensor data and preprocessing for activity annotation. Table. I depict the activity annotation involving recognizing elderly activity behaviour over time and storage of data in system. Sensor fusion data was not segmented into separate sequences for each activity rather it was processed as a continuous stream. Activity annotation process will help the monitoring system to recognize the various behaviours of the elderly at different instant of time. This process is done based on the collection of sensor identity from the sensor fusion of various sensing units connected to different house-hold appliances. Appropriate time slot size is to consider for labelling the activity based on the sensor id and time of the day. It provides sufficient information for doing data analysis. Even if the sensors are active for multiple times during a particular time slot, activity labeling is done according to the definition specified in the system. We experimented with models that used time slot sizes of one hour, three hours, four hours and six hours duration. Activity recognition in terms of three hours and four hour time slot sizes are giving more modelling accuracy for labelling the activity processing. Table. I show that sensors id 11, 12, 13 are used for kitchen appliances. If multiple times of sensor id 11, 12 or 13 are active during four hour time slot the event is annotated with defined activity as breakfast, lunch and dinner respectively. Obviously an event like preparing breakfast, lunch or dinner doesn’t happen at the same time every day, but it is usually happened within a specified time interval. Hence preparation

Health care providers assisting the elderly can have a more comprehensive, longitudinal evaluation of the monitored elderly activities than the snap shot assessment obtained during an annual physical examination. If the elderly person needs assistance with some of their Activities of Daily Living (ADLs) - An index or scale which measures a patient’s degree of independence in bathing, dressing, using the toilet, eating and transferring (moving from a bed to a chair, for example) [19] as these are usedto determine the need for long-term care or Instrumental Activities of Daily Living (IADLs), professional caregivers accessing the elderly activity reports will have an objective assessment of their actual needs and appropriate care services based on the daily functional assessments of the person. There are numerous wellness concepts suggested by experts from various domains, each of which is defined from their specialist perspective and contain several dimensions of wellness [20, 21, 22]. Several authors are of the same opinion that wellness is not just the state of mind or free from illness and disease; it is not a single state [20, 21]. Wellness does have multiple dimensions or levels. However, an integrated definition does not exist. Hence, there are various instruments and methods for wellness assessment. Wellness is a very wide and multifaceted perception. It is difficult to define the term wellness completely because the term wellness is developed overtime and changed by different influential factors such as culture, experience, belief, religion, context etc [20, 23, 24]. Wellness meaning in our context is how “healthy” the elderly living alone is able to perform his essential daily


S E N S O R

S T R E A M

Close port

Data acquisition & Activity annotation

Wellness function β1, β2

COM 1

M.Oven

Toaster

Toilet

Electircal monitoring sensor Dchair

Force sensor Water flow monitoring sensor Panic button

Couch

Fig. 9.

1969

Bed

Functional description of wellness computation functions. W.kettle

activities in terms of the usage of the house-hold appliances. We introduced two wellness functions to determine the wellness of the elderly person under the monitoring environment. The first function (β1 ) is determined from the non-usage or inactive duration of the appliances. The second function (β2 ) is determined from the over-usage of a few specific appliances. The two functions β1 and β2 determine the wellness of elderly are based on the usage of house-hold appliances. Fig. 9 shows the functional description in determining the Wellness functions. The wellness functions were calculated during the runtime of the system as background process taking the activity durations from the respective files of the computer system. These indices were simultaneously recorded in the database for future data processing and prediction of the unusual behaviour of the elderly. β1 and β2 are helpful in deriving the reliability of performing ADLs as regular or irregular over a long period of execution of the system. 1) Wellness Function 1, β1 : The wellness function 1, designated as ‘β1 ’ is defined by the following equation β1 = 1 −

t T

(1)

where β1 = Wellness function of the elderly based on the measurement of inactive duration of appliances t = Time of Inactive duration of all appliances (i.e.) duration time no appliances are used. T = Maximum inactive duration during which no appliances are used, leading to an unusual situation If β1 is equal to 1.0 indicates the elderly is in healthy wellbeing situation. If β1 is less than 1.0 the situation indicates some unusual situation. If β1 goes below 0.5 then care is required. 2) Wellness Function 2, β2 : The wellness function 2, designated as ‘β2 ’ is defined by the following equation Ta β2 = 1 + 1 − (2) Tn Where β2 = Wellness function of the elderly based on excess usage measurement of appliance. Ta = Actual usage duration of any appliance. Tn = Maximum usage duration use of appliances under normal situation. Under normal condition, Ta < Tn ; No Abnormality Only if Ta > Tn then β2 is calculated using the eq. (2). The value of β2 close to 1 to 0.8 or so may be considered as normal situation. If β2 goes less than 0.8, then it indicates

TV

Exit

Fig. 10.

Front end of the data acquisition unit.

the excess usage of the appliance corresponding to an unusual situation. In ideal case, β1 and β2 equals to one indicate the elderly activities are recurring with equal durations every time. However, human behavior is not consistent; hence the optimum alarm level for β1 and β2 are determined so that false warning messages are minimized. Based on the experiments conducted at different elderly houses (as further discussed in section III) there are instances of the maximum inactive and active duration of the appliances. Deriving β1 and β2 accordingly from the experiments at the elderly houses, warning messages are generated when β1 goes below 0.5 and β2 goes less than 0.8. Maximum inactive duration and Maximum usage duration use of appliances can be obtained during the trial run period of the system. The trial run period may be varied depending on the elderly activities of daily living conditions. Once the system learns the daily activity behaviour (i.e.) once the daily activities are cyclic then the trial run execution phase will be shifted to test phase and wellness indices are determined. (e.g) Section III, table II depicts the obtained maximum duration of the different appliances at the end of one week trial run. III. E XPERIMENTAL R ESULTS AND D ISCUSSION The experimental setup is as follows: WSN consisting of six electrical sensors, four force sensors, two contact switch sensors, one combined temperature/humidity monitoring sensor and one alarm/reset button are installed in the home to monitor elderly behaviour and assist the elderly living alone if there is any irregular behaviour at a particular time. Along with the wireless sensor network a laptop installed with the developed intelligent software connected with zigbee module acting as coordinator is associated with WSN to collect and monitor the elderly behaviour. Program for data acquisition, activity recognition and wellness determination are written using Microsoft Visual Studio. The fabricated sensing modules along with Zigbee components are configured as mesh topology to have effective communication with zigbee coordinator for recording sensor values in the system for further machine learning process. This section concludes with a presentation of the acquired results. Fig. 10 depicts the front end of the developed software system indicating which sensor is active or inactive.

1970


TABLE II S UBJECT 1 M AXIMUM A CTIVE D URATION OF THE A PPLIANCES

TABLE III S UBJECT 1 M AXIMUM A CTIVE D URATION OF THE A PPLIANCES

D URING O NE W EEK T RIAL RUN

D URING O NE W EEK T ESTING P HASE

Date/Appliance

Maximum Active Duration(hh:mm:ss) Bed Toilet Chair TV Couch 05/06/2011(Sun) 9:35:40 0:12:20 0:17:45 1:10:50 0:57:45 06/06/2011(Mon) 7:50:10 0:10:35 0:15:35 0:45:20 1:45:50 07/06/2011(Tue) 9:20:10 0:14:45 0:25:28 2:15:10 2:30:10 08/06/2011(Wed) 8:45:50 0:13:55 0:10:20 1:45:50 0:55:20 09/06/2011(Thu) 8:35:25 0:12:20 0:19:45 1:55:30 2:20:10 10/06/2011(Fri) 8:50:25 0:15:45 0:20:35 1:30:20 1:30:45 11/06/2011(Sat) 9:25:15 0:10:55 0:28:30 1:40:10 2:10:35 Maximum 9:35:40 0:15:45 0:28:30 2:15:10 2:20:10

Subject: 3

9:25:20, 1.01795

0:11:10, 1.291005

0:18:55, 1.336257

1:27:45, 1.3736

13/06/2011(Mon)

7:20:45, 1.234366

0:12:15, 1.222222

0:16:25, 1.423977

3:15:50, 0.60285

14/06/2011(Tue)

8:50:37, 1.078257

0:10:45, 1.31746

0:20:18, 1.287719

0:20:18, 1.287719

15/06/2011(Wed)

9:15:15, 1.035466

0:12:55, 1.179894

0:34:30, 0.789474

1:15:20, 1.46253

16/06/2011(Thu)

9:35:35, 1.000145

0:15:20, 1.026455

0:15:20, 1.026455

2:50:40, 0.78204

17/06/2011(Fri)

8:30:55, 1.112478

0:13:45, 1.126984

0:13:45, 1.126984

1:45:50, 1.24456

18/06/2011(Sat)

10:25:15 0.91386

0:12:15, 1.222222

0:18:40, 1.384675

1:55:35, 1.17538

Subject: 2 MO TR WK AD HT TV DC BD CO TO NA

Subject: 4

Fig. 11. Depict the percentage use of different appliances1 at various subject houses.

Maximum Active Duration(hh:mm:ss)

12/06/2011(Sun)

Wellness function “β1” S4 Subject locations

Subject: 1

Date/Appliance

S3 S2 S1

Real-time activity status of the elderly can be easily seen on the front-end of the system. This interface enables the care provider to know immediately the present activity status of the elderly (i.e) whenver a house–hold appliance connected by a sensing unit is in use then the interface will highlight the icon indicating the location of the elderly. System can also simulataneously store the sensor activity information and analyze the wellness indices. Another advantage of the interface is that remote monitoring of the elderly can be easily done. Real-time sensor activity status at the corresponding hour of the day is recorded simultaneously in the respective files of the computer for data processing. Continuous sensor activity status is recorded in respective files of the computer for effective data processing. Fig.13 shows the various activity sequences performed by the elderly during one week trial run. The pie-chart as shown in Fig.11 indicates the uses of different appliances at four different subject homes. It can be inferred from the figure that the bed is very important for the life of the elderly person. During one week trial run maximum active duration of the appliances is given in Table II. Fig.13 gives the pictorial representation of activity occurrence-based on data obtained from a running system. During the testing phase β2 is calculated using the eq. (2). Table III shows the corresponding β2 values for four different appliances. The value of β2 close to 1 to 0.8 or so may be 1 MO:MicrowaveOven, TR:Toaster, WK:WaterKettle, AD:Audiodevice, HT: Room Heater, TV:Telivison, DC:Dinning Chair, BD:Bed, CO:Couch, TO:Toilet.NA:No Appliance.

0.00 0.10 0.20 0.30 0.40 0.50 0.60 0.70 0.80 0.90 1.00 β1 Fig. 12. β1 indices at four different elderly home for one week of trial and one week testing phase. β1 close to 1 indicates more healthy of the person in performing the regular daily activities.

considered as normal situation. If the value β2 is less than 0.8 indicate, excess usage of the appliance corresponds to an abnormal condition. “β1 and β2 ” functions can inform how well the elderly is performing daily activity behaviour is executed. Fig 12 and Fig 14 show the wellness indices at different elderly homes. In Fig.12 it is seen that the β1 for the subject 2 on a particular day has gone below 0.5.In practice it has been observed that the elderly went outside the house for quite a long duration without deactivating the monitoring system. In Fig.14 it is seen β2 for subject 2 has gone to very low value for the use of chair. It has been observed that on that particular day, the elderly had a visitor and took lunch sitting on the chair for a long duration. For the subject 3, it has been observed that the elderly slept quite a long time as he was not feeling well. These observations tell clearly about the wellness determination of the system. The alarm can be set depending on values of β1 and β2 . These should be diverse for different elderly people. While the alarm is set, the system can generate a sound to inform the elderly that a message is going to be sent to the care provider.


1971

Activity occurrences 2011-06-04 2011-06-05 2011-06-06 BF RE LN DN SK TO DI SL CU SG 21

2011-06-07 2011-06-08 2011-06-09 2011-06-10 2011-06-11 12 am

3 am

β2

β2

β2

β2

Fig. 13.

1.8 1.6 1.4 1.2 1 0.8 0.6 0.4 0.2 0

1.8 1.6 1.4 1.2 1 0.8 0.6 0.4 0.2 0

1.8 1.6 1.4 1.2 1 0.8 0.6 0.4 0.2 0

1.8 1.6 1.4 1.2 1 0.8 0.6 0.4 0.2 0

6 am

9 am

12 pm

2

6 pm

9 pm

12 am

Pictorial representation of activity+ occurrence-based on data obtained from a running system.

β2 at subject 1

1

3 pm

3

4 5 Day β2 at subject 2

6

duration usage time of the house-hold appliances will be more accurate and the wellness state of the elderly will be precisely determined. The calculations of β1 and β2 are done simultaneously when the sensor activity status is plotted on the respective files of the system. Wellness functions were helpful in deducing no appliance and excess used by the elderly at their houses. These are also helpful in predicting the early abnormal situation of the elderly in performing their ADLs.

7

IV. C ONCLUSION

1

2

3


6

7

1

2

3

6

7


Bed Toilet Chair Couch

1

2

3

4 Day

5

6

7

Fig. 14. β2 values at four different elderly homes indicating their activity with the corresponding house-hold appliances.

Wellness is a wide and multifaceted phrase. In this research Wellness is about well-being of elderly in performing their daily activities effectively at their home. This will facilitate the care providers in assessing the performance of the elderly activities doing independently. The developed home monitoring system using WSN is low cost, robust, flexible and efficiently monitor and assess the elderly activities at home in real-time. Real-time activity behaviour recognition of elderly and determination of wellness function of the elderly using the activity of appliances was encouraging as the system was stable in executing the tasks for few weeks. If the system is executed for required number of months the optimal maximum utilization of the appliances used by the elderly will be derived. Also, the efficiency of wellness functions to predict the abnormal behaviour of the elderly in using the daily household appliances will also increase. In the near future, the system will be augmented with the physiological parameter monitoring sub-system. This will supplement to get information about health parameters like body temperature, heart rate etc., so that elderly health perception and daily activity behaviour recognition together can be assessed to determine the wellness of the elderly. R EFERENCES

To deactivate the alarm, considering as regular in future, corresponding event can be reset to regular if the elderly press the reset button. It can be inferred that as the system run for a longer period of time continuously then the maximum and minimum

[1] A. H. Nasution and S. Emmanuel, “Intelligent video surveillance for monitoring elderly in home environments,” in Proc. IEEE 9th Workshop Multimedia Signal Process., Oct. 2007, pp. 203–206. [2] Z. Zhongna, D. Wenqing, J. Eggert, J. T. Giger, J. Keller, M. Rantz, and H. Zhihai, “A real-time system for in-home activity monitoring of elders,” in Proc. Annu. Int. Conf. IEEE Eng. Med. Biol. Soc., Sep. 2009, pp. 6115–6118.

1972

[3] S. J. Hyuk, L. Boreom, and S. P. Kwang, “Detection of abnormal living patterns for elderly living alone using support vector data description,” IEEE Trans. Inf. Technol. Biomed., vol. 15, no. 3, pp. 438–448, May 2011. [4] A. Wood, J. Stankovic, G. Virone, L. Selavo, H. Zhimin, C. Qiuhua, D. Thao, W. Yafeng, F. Lei, and R. Stoleru, “Context-aware wireless sensor networks for assisted living and residential monitoring,” IEEE Netw., vol. 22, no. 4, pp. 26–33, Jul.–Aug. 2008. [5] J. K. Wu, L. Dong, and W. Xiao, “Real-time physical activity classification and tracking using wearble sensors,” in Proc. 6th Int. Conf. Inf., Commun. Signal Process., Dec. 2007, pp. 1–6. [6] Z. Bing, “Health care applications based on ZigBee standard,” in Proc. Int. Conf. Comput. Design Appl., vol. 1. Jun. 2010, pp. V1-605– V1-608. [7] K. P. Hung, G. Tao, X. Wenwei, P. P. Palmes, Z. Jian, W. L. Ng, W. T. Chee, and H. C. Nguyen, “Context-aware middleware for pervasive elderly homecare,” IEEE J. Sel. Areas Commun., vol. 27, no. 4, pp. 510–524, May 2009. [8] H. Yu-Jin, K. Ig-Jae, C. A. Sang, and K. Hyoung-Gon, “Activity recognition using wearable sensors for elder care,” in Proc. 2nd Int. Conf. Future Generat. Commun. Netw., vol. 2. Dec. 2008, pp. 302–305. [9] A. A. Moshaddique and K. Kyung-Sup, “Social issues in wireless sensor networks with healthcare perspective,” Int. Arab J. Inf. Technol., vol. 8, no. 1, pp. 34–39, Jan. 2011. [10] K. Hara, T. Omori, and R. Ueno, “Detection of unusual human behavior in intelligent house,” in Proc. 12th IEEE Workshop Neural Netw. Signal Process., Nov. 2002, pp. 697–706. [11] S.-W. Lee, Y.-J. Kim, G.-S. Lee, B.-O. Cho, and N.-H. Lee, “A remote behavioral monitoring system for elders living alone,” in Proc. Int. Conf. Control, Autom. Syst., Oct. 2007, pp. 2725–2730. [12] D. Lymberopoulos, A. Bamis, T. Eixeira, and A. Savvides, “BehaviorScope: Real-time remote human monitoring using sensor networks,” in Proc. Int. Conf. Inf. Process. Sensor Netw., Apr. 2008, pp. 533–534. [13] E. Nazerfard, P. Rashidi, and D. J. Cook, “Discovering temporal features and relations of activity patterns,” in Proc. IEEE Int. Conf. Data Min. Workshops, Dec. 2010, pp. 1069–1075. [14] P. George, X. George, and P. George, “Monitoring and modeling simple everyday activities of the elderly at home,” in Proc. 7th IEEE Consumer Commun. Netw. Conf., vol. 7. Jan. 2010, pp. 1–5. [15] H. Medjahed, D. Istrate, J. Boudy, and B. Dorizzi “Human activities of daily living recognition using fuzzy logic for elderly home monitoring,” in Proc. IEEE Int. Conf. Fuzzy Syst., Aug. 2009, pp. 2001–2006. [16] E. Dishman, “Inventing wellness systems for aging in place,” IEEE Comput., vol. 37, no. 5, pp. 34–41, May 2004. [17] A. Gaddam, S. C. Mukhopadhyay, and G. S. Gupta, “Elder care based on cognitive sensor network,” IEEE Sensors J., vol. 11, no. 3, pp. 574–581, Mar. 2011. [18] N. K. Suryadevara and S. C. Mukhopadhyay, “Wireless sensors network based safe home to care elderly people: A realistic approach,” in Proc. IEEE Recent Adv. Intell. Comput. Syst., Sep. 2011, pp. 1–5. [19] Covering Health Issues, 6th ed., Alliance Health Reform, Washington D.C., 2005. [20] C. Soomlek and L. Benedicenti, “Operational wellness model: A wellness model designed for an agent-based wellness visualization system,” in Proc. 2nd Int. Conf. eHealth, Telemed. Social Med., 2010, pp. 45–50. [21] A. Kailas, “A generic conceptual model linking wellness, health lifestyles, and user assistance,” in Proc. 13th IEEE Int. Conf. eHealth Netw. Appl. Serv., Jun. 2011, pp. 266–269. [22] A. Kailas, C.-C. Chong, and F. Watanabe, “A simple iterative algorithm for wellness applications,” in Proc. IEEE Wireless Commun. Netw. Conf., Apr. 2010, pp. 1–6.


[23] M. Alwan, “Passive in-home health and wellness monitoring: Overview, value and examples,” in Proc. IEEE Annu. Int. Conf. Eng. Med. Biol. Soc., Sep. 2009, pp. 4307–4310. [24] J. Healey and B. Logan, “Wearable wellness monitoring using ECG and accelerometer data,” in Proc. 9th IEEE Int. Symp. Wearable Comput., Oct. 2005, pp. 220–221.

Nagender Kumar Suryadevara (S’11) received the Bachelors degree from Sri Krishnadevaraya University, Anantapur, India, and the Masters degree from Madurai Kamaraj University, Madurai, India, in 1996 and 1998, respectively. He is currently pursuing the Ph.D. degree with the School of Engineering and Advanced Technology, Massey University, Palmerston North, New Zealand. His career demonstrates consistent success as an Administrator and Educator with the graduate and post-graduate education levels in India, Ethiopia, and Oman. He is currently involved in the development of software systems for the home monitoring project using wireless sensors networks. His current research interests include domains of wireless sensor networks and machine learning.

Subhas Chandra Mukhopadhyay (M’97–SM’02– F’11) received the Degree from the Department of Electrical Engineering, Jadavpur University, Kolkata, India, the Masters degree in electrical engineering from the Indian Institute of Science, Bangalore, India, the Ph.D. degree in engineering from Jadavpur University, and the Doctor of Engineering degree from Kanazawa University, Kanazawa-Shi, Japan. He is currently a Professor of sensing technology with the School of Engineering and Advanced Technology, Massey University, Palmerston North, New Zealand. He has over 21 years of teaching and research experience. He has authored/co-authored over 250 papers in different international journals, conferences, and books. He has edited ten conference proceedings, ten special issues of international journals as a Lead Guest Editor and eleven books, nine of which were with Springer-Verlag. His current research interests include sensors and sensing technology, electromagnetics, controls, electrical machines, and numerical field calculation. Dr. Mukhopadhyay has been awarded numerous awards in his career. He is a Fellow of the Institution of Engineering and Technology, U.K. He is an Associate Editor of the IEEE S ENSORS J OURNAL and the IEEE T RANSACTIONS ON I NSTRUMENTATION AND M EASUREMENTS . He is with the Editorial Board of e-Journal on Non-Destructive Testing, Sensors and Transducers, the Transactions on Systems, Signals, and Devices, and the Journal on the Patents on Electrical Engineering. He is the Co-Editor-inChief of the International Journal on Smart Sensing and Intelligent Systems. He is with the Technical Program Committee of the IEEE Sensors Conference, the IEEE Instrumentation and Measurement Technology Conference, and numerous other conferences. He was the Technical Program Chair of ICARA 2004, ICARA 2006, and ICARA 2009. He was the General Chair/Co-Chair of ICST 2005, ICST 2007, IEEE ROSE 2007, IEEE EPSA 2008, ICST 2008, IEEE Sensors 2008, ICST 2010, IEEE Sensors 2010, and ICST 2011. He has organized the IEEE Sensors Conference at Christchurch, New Zealand, as the General Chair in October 2009. He is the Chair of the IEEE Instrumentation and Measurement Society New Zealand Chapter. He is a Distinguished Lecturer of the IEEE Sensors Council.