INTRODUCTION:

IoT is a collection of sensor enabled physical devices connected to the internet, that able to exchange the information between them without the human involvement [1-3]. AAL is a recent communication technology to help the aged people, keep an active and independent to do their daily activities. In recent years, AAL is involving the application areas like health monitoring, smart home, cloud computing, assistive robotics and wireless communication [4].

The advancement of technology growth in medical field, which allows the people to live longer and provides the good health condition than previous generations.

The world population division prepared a report, stated that “Approximately 20% of population will be age nearly 60 by 2050” [5]. The older people face the challenges like chronic diseases, physical activity, hearing and vision problem [6]. Due to increase the population of older people, the society and health care system face the challenges are increasing the diseases and health care costs, shortage of caregivers, dependency and larger impact on society.

In recent years, many researchers are involved to develop a variety of assistive technology called “Ambient Intelligence”. It is a technology, which aims to build a safe environment around the assisted people [7-8]. AAL contains assisted living technologies and ambient intelligence. It can be used to monitor the health condition and also improve the wellness. AAL tools are medication reminder for older people, mobile emergency response system, fall detection system, video surveillance system, monitoring human activities and etc.

In this section, we address the problem of AAL, for monitoring the people health. Parisa Rashidi and Alex Mihailidis et’al investigated about Ambient Assisted Living Tools for older Adults. The aging population creates the problem as increase the cost of health care, caregiver burden and led their life individually. In recent years, a rapid growth of AAL technologies due to drastic increase of elderly people population. It is summarized the importance of AAL tools for aged people based AI paradigm. It is discussed various AAL applications such as Health and activity monitoring tools and wandering prevention tools [9]. Ghayvat Hemant and Subhas Chandra Mukhopadhyay investigated about wearable sensors for human activity monitoring. The advancement of sensing technologies, Wireless Sensor Networks (WSN), embedded system makes it as a smart, intelligent system and monitors the human activities continuously. It addressed the challenges of the wearable Wireless Sensor for human activity, behavior monitoring and also addressed the energy harvesting issues [10].

Nagender Kumar suryadevara and Subhas Chandra Mukhopadhyay developed WSN based home monitoring system for aged people. This system is mainly concentrated to estimate well-being condition of elderly people. The home appliances, embedded with various sensor units. Whenever the human being uses the home appliances, sensor enabled appliances generates the value and exchange the information through communication module. This system defined two wellness functions for finding the status of elderly people. It is tested at home about four different elderly people living in independently and it provided the best results to useful for elderly people as well as our society [11].

G. Demiris and H. Thompson presented the Smart Home and Ambient Assisted Living (SHAAL) systems, to integrate with home and its appliances, for capturing the data and predict the health related activities of elderly people. This system is represented that how SHAAL system can turn the sensor data into knowledge and also it improves the quality of elderly person's life [12]. Thomas kleinberger et’al presented Ambient Intelligence based elderly people to handle the future interface. This work is used to train the elderly people and how to provide the Ambient Assisted interface. It is evaluated the interface from different cases in specific situations [13].

Ivo Iliev and Ivan Dotsinsky reviewed about Assisted Living Systems for elderly and disabled People. This system goal is to improve the lifetime of elderly people and reduce expenses of healthcare [14]. Andrea Lombardi et’al presented an accelerometer based fall detection logic system. This work main objective is to detect the falls, level of confidence and enable the alarm indication [15]. Hong Shan et’al reviewed the challenges of AAL such as dynamic of service availability, service mapping, person's willingness, psychological frustration and technological frustration [16]. Holger staff et’al presented an event driven approach for recognizing the activity. The main challenge of AAL is reliable activity recognition and it is presented and process the information for detection of activity based on the Activity of Daily Living (ADL) [17].

Diego Lopez-de-Ipina et ‘al described an interactive TV based AAL platform. It is described about the architecture of AAL, Interactive TV (ITV). The interactive TV consists of OSGi middleware, RFID and NFC. It provides the better QoS on caretaking through a combination of AAL, ITV and middleware. This work is offered three different features such as i. easily deployable at elderly people home or residence, ii. It properly updates the elderly person's health status to caretaker, iii. It is low cost and easily deployable [18].

Cesta, Amedeo et ‘al described about monitoring the elderly people with the robocare domestic environment. It is described about multiple intelligent components integrated to form an artificial intelligence based elderly care system. This article presented a psychological evaluation focus on elderly peoples' attitudes; perceived utility, acceptability, emotional response and interaction modeling are considered. This article mainly focused on two aspects. The first aspect is activity monitoring and it is based on the scheduling system. It maintains the constraint based temporal knowledge in an environment. The second aspect is technology acceptability, due to the complexity of deploying this type of system [19-20].

MATERIAL AND METHODS:

Ambient Assisted Living (AAL) is a recent communication technology, which makes an intelligent object in the environment, for supporting the elderly people in living independently. An Ambient Intelligence is a technology and it aims to build a safe environment around the assisted people. The proposed system intends to perform three operations such as elder people monitoring, elderly people activity recognition and health status perdition using machine learning algorithm. Sensor and actuators are embedded with physical devices. It exchanges the data from home environment to internet through Zigbee. The hardware interface is placed in between the communication module and internet. The generated sensor data is stored in to cloud and apply the machine learning algorithm to take the decision accordingly.The proposed system architecture is represented as below.

System Architecture:

Fig 1. Proposed system architecture

The proposed system architecture is categorized into four parts such as WSN, Cloud storage, Machine learning and Application layer. In WSN part, the sensor and actuator is enabled physical devices connected with communication module. The sensor generates the data and exchange the information to cloud storage via communication module. It uses the Zigbee communication module and one of the module acts as PAN coordinator and remaining all are acting as participants. Each and every WSN nodes are connected to Application Gateway through hardware interface. The application Gateway directly connects to the internet. The sensor enabled devices instantly sending the data to the cloud. The Cloud computing is an internet based computing that provides the computer and data processing resources from on demand. It can be categorized into private, public and Hybrid cloud. The type of cloud selection is entirely based on the application. Third part of this architecture is machine learning. It gets sensor data, which is stored in the cloud. It is used to apply the tainting and testing from device generated data. Finally, it activates the actuators based on the knowledge acquired from a machine learning algorithm. The application layer monitors the elderly person's health condition.

The proposed system introduces the AAL based elderly people monitoring system. It performs three operations namely Elder people monitoring, activity recognition and Health status prediction using Internet of Things (IoT) and Support Vector Machine (SVM) algorithm. The proposed system operations are explained detailed way in below section.

Elderly People Monitoring:

Due to increase the population growth of elderly people, the monitoring is a major role in our society. The proposed system is used the sensors namely pulse rate, Accelerometer, temperature sensor, for collecting the elderly people health information. The elderly people monitoring and analyze the health condition from the elderly people heart pulse rate, blood pressure and etc. The pulse rate indicates that the number of pulsation per minute. Generally, the ordinary person heart rate ranges is from 60 to 100 beats per minute. The human body temperature is varying from person to person. The healthy adult body temperature ranges is from 97.8 degrees F to 99 degrees F.

The sensors are connected to Arduino microcontroller. The sensor generated data will be stored in a public cloud through the ESP8266 AI Cloud module. The elderly person's temperature, heartbeat, pulse rate, Accelerometer values is shown in the ThingSpeak. If anything goes wrong, the event will be fired and it sends the SMS to the respective family member or doctor. Then, they will take the decision, to improve their health condition. Figure1 shows that the output of elderly people monitoring.

Fig 2. Elderly people monitoring

Elderly People Activity Recognition:

In this module, the proposed system uses the Gyro Accelerometer (v3) for assessing the elderly people activity. The gyro accelerometer connects with an Arduino UNO microcontroller. It converts the analog signal into digital signal and its value ranges from 0 to 1023. The proposed system finds the elder people activity like sitting, running, sleeping and falling. We have tested in different situation with number of human being and it identifies the activity of the elderly people accurately.

Table-1 Activity Recognition Gyro Accelerometer Value Ranges

Value Ranges
X Axis	Y Axis	Z Axis	Activity
520-540	540-580	740-780	Standing
350-550	300-1000	150-800	Falling
200-600	250-750	750-1023	Walking

Figure3 shows that the output of elderly people activity recognition.

Fig 3. Elderly people activity recognition

Health Status Prediction using K-SVM Algorithm:

In this module, the proposed system predicts the health status using machine learning algorithms. The sensors (pulse rate, temperature and gyro accelerometer) are connected to the Arduino micro controller. It generates the data and it will be stored in to cloud (Think Speak). Figure 4 shows that the real time sensor data represented in Thingspeak.

Figure 4. Real time data in ThingSpeak

The proposed system can download the real-time sensor data from ThingSpeak. The file format is csv or xlsx. The file contains the attributes such as pulse rate, temperature, blood pressure, timestamp, location and status. The health status prediction work flow is given below.

Fig 5. Health status prediction work flow

i. Real time input data:

The input dataset is comprised of pulse rate, temperature, blood pressure, timestamp, location and status of the elderly people. Our intention is to do this experimentation for predicting the health status. The real time data is collected about one week or one month data for this experimentation. The target attribute value can be assigned from universal truth and scientific proof. We represented the status value is normal or abnormal.

ii. Training and Testing the Real-time data:

The experiment is implemented using R-Studio. The “Sample” command is used to select the record randomly and divide the dataset into training and testing data. Training dataset is taken as 70% and remaining data for validation. The “SetDiff” command is used to validate the real time data.

iii. Classification using K-SVM algorithm:

Kernel based support vector machine (K-SVM) algorithm is used to classify the record among five attributes. It supports demographic data. SVM is a supervised machine learning algorithm, which is mostly used for classification problem. It performs the classification by finding the hyper plane and it differentiates two classes appropriately. The kernel- SVM algorithms are based on convex optimization or Eigen problems and it is statistically well-defined. The K-SVM algorithm is represented below.

K-SVM Algorithm:

Algorithm K-SVM

Candidate SV= {closest pair from opposite classes}

While there are violating points do

Find a violator

candidateSV =candidateSV U violator

if any αp < 0 due to addition of c to S then

candidateSV = candidateSV \ p

repeat till all such points are pruned

end if

end while

iv. Validate and predict the health status of elderly people:

In R studio, “predict” command is used to validate the data. Whenever we apply the new data to do the testing and it predicts the result accordingly.

RESULTS AND DISCUSSION:

The proposed system is implemented for elderly people monitoring, activity recognition and health status prediction. The health status prediction is implemented using a K - SVM algorithm. This system is taken into account about one week and monthly data from different elderly people. The K-SVM algorithm is compared with SVM algorithm and K-SVM provides the better accuracy than SVM. The accuracy of SVM and K-SVM results are represented in the below table.

Table-2 Performance Comparison

Real Time Data	Accuracy (%)
Real Time Data	SVM	K-SVM
1 week data	89	91
1 month data	84	88

Figure 6 shows that the performance comparison between SVM and K-SVM.

Fig 6. Performance comparison between SVM and K-SVM

CONCLUSION:

Ambient Assisted Living (AAL) is a recent communication technology, which makes an intelligent object in the environment to support the elderly people in living independently. Nowadays, the populations of elderly people in living alone keep on increasing. Hence it becomes a major impact in our society. The proposed system performs three operations such as Elderly people monitoring, activity recognition and Health status prediction using Internet of Things (IoT) and Support Vector Machine (SVM) Algorithm. In health status prediction, the SVM algorithm provides the accuracy rate 89% for 1 week data and 84% for 1 month data. Similarly, the K - SVM algorithm provides the accuracy rate 91% for 1 week data and 88% for 1 month data. The proposed system is introduced K-SVM algorithm that provides more prediction accuracy than SVM algorithm and also provides the cost effective solution.

Future Work:

The future work planned to work towards efficient elderly people monitoring system, highly cost effective, identify the diseases accurately and take decision accordingly. In order to provide the efficiency, this work planned to use the recent hardware technology and also incorporates the Nadi diagnosis techniques.

REFERENCES:

1. Al-Fuqaha A, Guizani M, Mohammadi M, Aledhari M, Ayyash M. Internet of things: A survey on enabling technologies, protocols, and applications. IEEE CommunicationsSurveys & Tutorials. 2015; 17(4):2347-76.

2. Sankar, S., and P. Srinivasan.. Internet Of Things (Iot): A Survey On Empowering Technologies, Research Opportunities And Applications. International Journal of Pharmacy & Technology 2016;8(4): 26117-26141.

3. S. Sankar and P.Srinivasan.. Composite Metric Based Energy Efficient Routing Protocol for Internet of Things, International Journal of Intelligent Engineering and Systems,2017; 10(5): 278-286.

4. Costa, R., Carneiro, D., Novais, P., Lima, L., Machado, J., Marques, A., & Neves, J. Ambient assisted living, China Communications 2016;13(5):6-7

5. Gerland, Patrick,. World population stabilization unlikely this century, Science 2014; 234-237.

6. U. Esnaola and T. Smithers,. Whistling to machines, in Ambient Intelligence in Everyday Life, Lecture Notes in Computer Science Series 2006: 198–226.

7. F. Sadri, Ambient intelligence: A survey, ACM Comput. Surv. 2011: 43(4):1–36:66.

8. E. Aarts and J. Encarnao, True Visions: The Emergence of Ambient Intelligence, 1st ed. New York: Springer-Verlag, 2008.

9. Rashidi, Parisa, and Alex Mihailidis. A survey on ambient-assisted living tools for older adults. IEEE journal of biomedical and health informatics 2013, 17(3): 579-590.

10. Ghayvat, Hemant, Wellness sensor networks: A proposal and implementation for smart home for assisted living. IEEE Sensors Journal 2015,15(12), (2015): 7341-7348.

11. Suryadevara, Nagender Kumar, and Subhas Chandra Mukhopadhyay. "Wireless sensor network based home monitoring system for wellness determination of elderly." IEEE Sensors Journal 12.6 (2012): 1965-1972.

12. Demiris, George, and Hilaire Thompson. "Smart homes and ambient assisted living applications: from data to knowledge-empowering or overwhelming older adults?. IMIA Yearbook 2011, 6(1): 51-57.

13. Kleinberger, T., Becker, M., Ras, E., Holzinger, A., & Müller, P. Ambient intelligence in assisted living: enable elderly people to handle future interfaces, International Conference on Universal Access in Human-Computer Interaction. Springer Berlin Heidelberg, 2007.

14. Iliev, Ivo, and Ivan Dotsinsky. Assisted Living Systems for elderly and disabled people: A short review. Int. J. Bioautom 15 (2011): 131-139.

15. Lombardi, A., Ferri, M., Rescio, G., Grassi, M., & Malcovati, P. Wearable wireless accelerometer with embedded fall-detection logic for multi-sensor ambient assisted living applications. Sensors, 2009 IEEE.

16. Sun, Hong, et al. Promises and challenges of ambient assisted living systems. Information Technology: New Generations, 2009. ITNG'09. Sixth International Conference on. Ieee, 2009.

17. Storf, H., Kleinberger, T., Becker, M., Schmitt, M., Bomarius, F., & Prueckner, S. An event-driven approach to activity recognition in ambient assisted living, European Conference on Ambient Intelligence. Springer Berlin Heidelberg, 2009.

18. López-de-Ipiña, D., Blanco, S., Laiseca, X., & Díaz-de-Sarralde, I. ElderCare: an interactive TV-based ambient assisted living platform, Activity recognition in pervasive intelligent environments. Atlantis Press, 2011:111-125.

19. Cesta, Amedeo, Gabriella Cortellessa, Riccardo Rasconi, Federico Pecora, Massimiliano Scopelliti, and Lorenza Tiberio.. Monitoring elderly people with the robocare domestic environment: Interaction synthesis and user evaluation,Computational Intelligence 2011,27(1) : 60-82.

20. Sharkey, Amanda, and Noel Sharkey. Granny and the robots: ethical issues in robot care for the elderly, Ethics and information technology 2012; 14.(1): 27-40.

Received on 01.03.2018 Modified on 17.07.2018

Research J. Pharm. and Tech 2018; 11(9): 3900-3904.

DOI: 10.5958/0974-360X.2018.00715.1