Elderly Personal Safety Monitoring in Smart Home Based on Host Space and Travelling Pattern Identification
The ratio of elderly has been increased in many countries, like America, China, Europe, Korea and Japan. Elderly personal safety has become a social problem in these countries. Smart home technology has been expected to settle down such problem and received much more attention globally. This study intends to provide an economic solution to monitor the movement of the accessory of elderly in the smart home, identify the hot space the elderly is likely to stay, capture the travelling pattern of the elderly who living in the smart home and then expose the possible danger when the elderly stays at a space beyond the period defined in the captured patterns. The family members, neighbor or doctor of the elderly will be informed consequently when the elderly is recognized to be unsafe. A prototype has been developed and the approaches were approved to be sufficient to check the scenarios. The pattern rules with very short duration were ignored and all the rules were customable and could be trained all the time. Simulation is introduced which validates the suggested approach as well.
January 22, 2012; Accepted: February 17, 2012;
Published: June 05, 2012
Several countries have already entered the era of aging population, including
America, China, Korea and Japan (Chang et al., 2009).
The aging population of China has increased to 1.6714 billion by the end of
2009 which is 12.5% in total Chinese population. That is, there is 1 elderly
every 8 people. Compared with the year 2009, elderly population increased 725
million, with the proportion increased by 0.5%. Most elderly suffered from chronic
illnesses such as hypertension and diabetes in 2010. Therefore, health care
for the elderly should be recognized as a social problem, reported by the Xinhua
News Agency, China (2010). Smart home technology turns out to be an efficient
approach to enable the elderly live happy with less support (Chang
et al., 2009; Yang and Helal, 2008; Alamo
et al., 2009; Ming et al., 2010; Zhou
et al., 2010; Helal et al., 2008) while
a safety enhancing mechanisms is proposed (Yang and Helal,
2008), web service is used to manage the medication (Alamo
et al., 2009), human intention change is detected and analyzed (Chang
et al., 2009; Ming et al., 2010),
ubiquitous service and home automation are suggested (Liao
and Tu, 2007; Yilmaz, 2010; Xiaohu
and Guangxi, 2006; Alam et al., 2011; Hussain
et al., 2008) and integrated mobile solution for personal health
care is provided (Zhou et al., 2010; Helal
et al., 2008).
Considering that the sudden disease, such as cardiovascular disease, cerebral
thrombosis and stroke, has an upward trend in the elderly in recent years, some
experts argued that fast rescue and timely treatment should be provided to the
sudden illness to enable the elderly survive from the emergency easily (Yang
and Helal, 2008; Alamo et al., 2009; Zhou
et al., 2010). Though there are some systems enable the elderly to
call for help when they are in trouble or emergency, these systems are not sufficient
when the elderly lose consciousness (Zhou et al.,
2010; Helal et al., 2008). To detect the emergency
as early as possible is essential in such situation, is the main purpose of
In this research, a smart home is divided into several rooms and each room
can be further divided into several spaces. It is assumed to be unsafe if the
elderly remains at a space beyond threshold period. The family members, neighbor
or doctor will be informed consequently if the elderly is recognized to be unsafe
by the system. Initial rules can be established for the safety checking, while
the living pattern of the elderly can be used to refine the rules so as to better
detect the unsafe situation. Typically the living pattern of the elderly can
be reflected as a set of movement queue (Chang et al.,
2009) but here in this paper, the spaces where the elderly stay with very
short duration are filtered out in the queue. Only those spaces where the elderly
stay with more than predefined threshold duration will be contained in the queue,
such that the pattern and related rules can be significantly reduced and simplified
and the rules can be easily defined and trained all the time.
Estimation the location of devices is import to reduce the routing and computing
in ubiquitous computing (Hightower and Borriello, 2001).
Global Positioning System (GPS) (Yin et al., 2011),
wireless enhanced 911(E-911) are good for outdoor environment with coarse granularity
but not good for indoor or building-dense environment (Pahlavan
et al., 2002). AT&Ts Active Badge (Want
et al., 1992) which uses diffuse infrared technology and could get
room-size accuracy. Active Bats (Harter et al., 1999),
another project from AT&T, uses an ultrasound time-of-flight Lateration
technique and get more accurate (9 cm), while requiring large scale deployment
and high cost. Microsoft RADAR records the received signal strength (RSS) from
several 802.11 access points and estimate user location in office or home environment
by comparing to empirical measured data (Bahl and Padmanabhan,
2000). MIT Cricket (Priyantha et al., 2001),
Carnegie Mellon University (Small et al., 2001),
IBM (Xiang et al., 2004), Maryland University
(Youssef et al., 2003) and Pittsburgh University
(Kaemarungsi, 2005) also have related research and system
ZigBee is a low cost, low power consumption, low complexity, low data rate,
high reliability and high security two-way radio communications technologies
(Chen et al., 2011). It is remarked as a promising
open standard comparing to the other wireless networking techniques such as
IrDA, WLAN (Jabri et al., 2008), Bluetooth (Subhan
et al., 2011) and UWB (Wheeler, 2007; Dong-Mei
and Qin-Yu, 2010). Though its data rate is not high, ZigBee is still a good
choice to be used in some small range of intelligent environment such as smart
home (Chang et al., 2009).
Received Signal Strength Indication (RSSI) (Aamodt, 2006;
Wang et al., 2008) is popular used to estimate
location of devices in indoor environment. It has attractive features due to
its low device complexity and low cost of nodes and related infrastructure,
comparing to the other similar location estimation techniques such as TOA, TDOA
and AOA (Niculescu and Nath, 2003; Bahl
and Padmanabhan, 2000; Savvides et al., 2001).
ZigBee technique (Wheeler, 2007) and RSSI-based estimation
algorithms (Aamodt, 2006; Alvarez et
al., 2010) has been used in our research (Xu
et al., 2011a, b), other location techniques
(Hightower and Borriello, 2001; Pahlavan
et al., 2002; Small et al., 2001;
Xiang et al., 2004; Kaemarungsi,
2005; Bahl and Padmanabhan, 2000) could also be
used to enhance the accuracy of the location detection when there is no critical
RSSI BASED LOCATION APPROACH FOR INDOOR DETECTION
RSSI based location detection: The RSSI is a function of the transmitted power and the distance between the sender node and the receiver node. It decreases with increased distance as the Eq. 1:
where, N is the signal propagation constant, also named propagation exponent, D is the distance from sender node and A is received signal strength at a distance of one meter.
After the parameters (N and A) have been calibrated in a real environment, the distance D can be estimated as Eq. 2:
Positioning system is composed of the reference nodes and blind nodes. The reference nodes are static nodes located with known position which know its location and can tell the other nodes its location through communication among nodes. The blind nodes receive data from the reference nodes, gain the coordinates of reference nodes and distance between reference nodes and itself, finally send these data to positioning system and calculate the coordinates of blind nodes.
RSSI based location approach is used in this research to detect the location of the elderly. Several anchor nodes have been deployed around the smart home and a blind node is suggested to be attached with the elderly. Assuming the four anchor nodes have been known as Pi = (xi,yi,zi) where I = 1,2,3,4 and blind node p (x, y, z) with its distance di, where I = 1,2,3,4, then there is the following Eq. 3:
||Error sphere with the position at the center and the error
threshold as the diameter
Formula (1) can be translated into Eq. 4:
The equation set can be expressed as Eq. 5:
The final result of the Least Squares Estimation can be expressed as Eq. 6:
We can get the unknown point p according to the formula above. However, the
distance di, where I = 1,2,3,4 which we get with RSSI technology
has some error due to the environment noise or multiple reflection. Therefore,
the final results of the position also have a certain degree of error. To address
this issue, we get different results from different anchor nodes and take the
average of these results as the final data, so as to reduce the error to some
degree. Besides, we assume that the error for a position is wrapped in a sphere
which is centered at that position. Further, we may assume that the error is
similar in different directions, such that the diameter of the error sphere
can be the error threshold, as shown in Fig. 1a. Therefore,
if the distance of two positions is with the error threshold, they will be considered
as the same position. For example, having O1 as the reference position,
the position O2 and O3 will be considered as the mirror
position of O1 due to the measurement error, while O4
is another position. When using a low accuracy distance detection method, there
will be some error in location detection. Take the Fig. 1b
for instance, when one elderly stays at point O1 for a while then
moves to O4. The detection system may get the inaccurate data and
get the mirror locations O1, O2 and O3. The
elderly will be recorded to be move from O1 to O3, then
to O2 and last stop at O4. A technique shift happens in
such scenario. As there will be 20 cm distance detection error when using RSSI-based
measurement, the error sphere will be used to detect the mirror location and
reduce the technique shifts.
Algorithms for the elderly travelling pattern capture: The travelling of the elderly in the smart home can be recognized as a sequence of movement from one position to another position. In this research, we assume that there will be some danger if the elderly stay at one position beyond the duration he or she used to stay. Therefore, we are trying to solve two problems, whats the travelling pattern of the elderly in the smart home and how to identify the safety status of the elderly by monitoring the activities of the elderly. Since we dont need to care much about the positions where the elderly stayed for a very short time, these positions can be filtered out and only those positions with long residence time will be monitored and analyzed. Such positions are named as Hot Space in this paper as shown in Fig. 2.
Figure 2 demonstrates an example of the travelling pattern of the elderly in the smart home. The elderly may stay in the living room at one place (Node H in Fig. 2) watching TV for most of the daily time. When its time for the lunch or dinner, the elderly will stand up and go to washing room (Node G in Fig. 2) to wash the hands before cooking (Node A in Fig. 2). After the cooking, the elderly will stay in the dining room (Node B in Fig. 2) to enjoy the food and wash the dinnerware in the kitchen (Node C in Fig. 2) after the lunch or dinner. The elderly will return to the living room to have a rest (Node E in Fig. 2). The elderly may read book or surf the Internet in the study room (Node F in Fig. 2). The elderly may wash the clothes in the washhouse (Node D in Fig. 2) as well. The travelling network will be different for different elderly and different smart home. Therefore, its very important for us to capture the travelling pattern for each individual and monitor the personal safety accordingly.
||A travelling network of the elderly in the smart home
Definition 1: Hot Space. The hot spaces are the positions with long residence time. A hot space can be a room or part of a room when the residence time for different parts is different in the room. Hot space is defined as S = <sid, x1, y1, z1, x2, y2, z2>, where sid is the identification no of the space; x1, y1, z1 and x2, y2, z2 indicate a cube which is wrapped by the planes x = x1, y = y1, z = z1, x = x2, y = y2 and z = z2.
Definition 2: Hot Space Cluster. Hot space cluster is a group of hot spaces which have the similar residence time in the smart home. It is defined as SC = <scid, hsc>, where hsc is set of hot spaces <S>, such as s1, s2,
Definition 3: Movement record. Movement record is the record of movement of the elderly in the smart home. Its defined as MR = <T, x, y, z>, where T is the time for the record, x, y and z is the location of the elderly. All the locations will be detected by the location estimation algorithm, so the movement record contains all the movements of the elderly in the smart home. But only the hot spaces will be captured, analyzed and used to identify the personal safety of the elderly.
Definition 4: Living Pattern. Living Pattern is defined as LP = <sc,
minD, mediumD, maxD, alertD>, where sc is the space cluster, minD, mediumD
and maxD refers to the minimal, medium and maximal duration the elderly has
ever stayed at the hot spaces in the space cluster each a time. All these durations
are measured and refined in the training. The alertD is the threshold duration
for the elderly to stay at the space cluster and the system would alert the
related persons for this unsafe situation. Living pattern in this paper has
been simplified and demonstrated as some mapping rules which are defined on
the basis of space cluster. In order to calculate the medium duration, the summary
SumD and count CD of the duration for this space cluster will be calculated
at first and the value of mediumD is the factor of the SumD and CD (medium =
SumD / CD).
Definition 5: Safety checking system. Safety checking system is defined as SACHS = <S, SC, MR, LP, F, Default>, where S is a set of spaces, SC is a set of space clusters, MR is the movement record, LP is the living pattern which indicates the rules, F is a serial of algorithms including the algorithms for location estimation, hot space detection, space cluster manual merging and safety checking. Default contains the default parameters including the pre-defined inferior limit (PIL: unit s), pre-defined upper limit (PUL: unit s) and the error of location estimation (ELE: unit m).
Algorithm 1: Hot space detection algorithm. Hot space is a kind of space
where the elderly stays for a long duration more than a pre-defined inferior
limit. Hot space detection is a training algorithm.
In the training model, when a hot space hs is detected by the algorithm, a
related space cluster will be generated which only contain the item hs. Such
space cluster can be refined or merged to some similar space cluster late manually.
SACHS. Default.newSpaceID is used to generate a new space id if there not exists
any space contain the input location, considering the error ele. SACHS.LP.add
is used to add the tempscid into the living pattern, init the minD, mediumD,
maxD, SumD with the input duration and init the CD with 1. SACHS.LP.update will
update the related living pattern with input duration accordingly.
Algorithm for the elderly safety monitoring: Algorithm 2. Elderly Safety
SACHS. S. find by position at line 10 is used to fetch the hot space from the pre-defined hot space list which contains the position. The warning or alert message will be sent out according to the pre-defined rule. If a position doesnt have any rule pre-defined but the resistance time is larger than the threshold, a warning message will be sent out.
VALIDATION IN THE SIMULATION
Movement record was simulated by a program which provided random duration, random 3-D movement (x, y and z). The program generated the record in 5 sec period.
Table 1 lists the hot spaces detected by the algorithm. AlertD is assigned with the maxD plus 300 sec.
For each space, it is recognized as safe if the elderly stay within the maxD,
unsafe is the elderly stay beyond the maxD but no exceed alertD.
|| Hot spaces detected by the algorithm
|| Elderly personal safety checking
If the elderly stay more time than the alertD, the system assumes that there
is a danger. Any observation on non-hot space but with duration more than the
PIL will be recognized as abnormal as no hot space could be detected. Because
the data volume is small, manual calibration is easy and enabled all the time.
The space cluster can be manually merged when there are of similar rules. For
example, in Table 1, space cluster 2, 7 and 8 could be merged
to one space cluster. In such way, the rules could be reduced though the spaces
remain the same.
Table 2 shows the checking results on some test cases. All these test cases were designed after the hot spaces had been detected. For the observation 1, 4 and 6, the elderly stayed in space cluster 1, 6 and 5. Since the durations (300, 3900 and 2400 sec) were no more than the related maxD (500, 28800 and 2400 sec), these three observations are safe. For the observation 2, the elderly was in space cluster 3, however the duration is 1800s which is between the maxD (1740s) and alertD (2040 sec), the situation is unsafe. For the observation 7, the elderly stayed in space cluster 2 for 2400 sec which is larger than alerD (1500 sec), it is assumed to be danger. Observation 5 indicated that the elderly stay in some other places other than available hot spaces and is recognized as abnormal.
The suggested personal safety monitoring approach may identify the possible danger in the smart home in a simple and straight way. Such method may detect the danger when the elderly is narcoses and doesnt move for a relative long time. There is significant advancement when comparing it to the current available manual danger reporting systems.
However, the suggested approach couldnt detect the danger very quick for some scenarios where the elderly is used to stay for a very long period. For example, if the elder used to sleep for most than 5 h and dont have any significant movement (some movements may be ignored due to the detection error). The suggested approach can only detect the danger after 5 h. To solve such issue, more accurate location detection method can be used so as to detect the small movement. In such way, the elder will be detected with significant movement from time to time and the danger will be detected in a very short duration. Here in this research, we assume that there is only one elderly in the smart home. When there are several elderly live together, the individual identification should be enhanced so as to distinguish different elderly. Otherwise, there will be some conflicts when analyzing the movement record which is the mixture of different movement sequence of the elderly. he communication protocol should be implemented so as to achieve such identification.
Elderly personal safety has become a social problem globally due to the increasing ratio of elderly. Smart home has been suggested to solve such problem. Different with other research, here in this paper, an economic solution for elderly personal safety checking has been proposed in this paper. ZigBee and RSSI based location estimation technique have been suggested; however, other research may choose some expensive solutions to get accurate location. Space, space cluster, movement record, living pattern and SACHS system have been defined while living pattern has been simplified so as to reduce the data volume of the rules and enable better calibration of the rules. Hot space has been used to indicate the locations where the elderly prefer to stay and checking rules have been defined as 4 levels of duration including minimal, medium, maximal and alert duration. Different status could then be recognized, such as safe, unsafe, abnormal and danger.
The main contributions of this research are two folders. First, an error tolerant location method is suggested to reduce the impact of detection error. In such way, the low accuracy but economic detection is enabled to be used in indoor environment while there is less technique shifts. Second, an efficient and effective method for personal safety monitoring method is suggested to report the possible danger in a timely mode. Its the enhancement to the currently available manual reporting mechanism.
Though the error of the location detection has been considered in this research and a parameter name ele has been introduced in the hot space algorithm, the impact of error is still there. The more accurate location detection results in more accurate safety checking. Due to the cost of infrastructure, smart home may not deploy with some expensive location detection solution. Our group will try to design better accurate location detection solution with reasonable cost in the future research.
This research is supported by Zhejiang Gongshang University, China with No. Xgz1102 to Dr. Weigang Chen (High Education Research Project). Thanks to Zheng Yu and Zepeng Hu for their effort in developing the algorithms for the prototype.
This research was also financial supported by National Natural Science Foundation of China with No. 60873022 to Dr. Hua Hu, The Science and Technology Department of Zhejiang Province, China with No. 2008C11009 to Dr. Bin Xu and Education Department of Zhejiang Province with No. 20061085 to Dr. Bin Xu.
Aamodt, K., 2006. Application Note AN042. CC2431 Location Engine, Chipcon products from Texas Instruments. http://focus.ti.com/lit/an/swra095/swra095.pdf.
Alam, M.R., M.B.I. Reaz and M.A.M. Ali, 2011. Statistical modeling of the resident's activity interval in smart homes. J. Applied Sci., 11: 3058-3061.
Direct Link |
Alamo, J.M.R., J. Wong, R. Babbitt, H.I. Yang and C.K. Chang, 2009. Using web services for medication management in a smart home environment. ICOST. Lecture Notes Comput. Sci., 5597: 265-268.
Alvarez, Y., M.E. de Cos, J. Lorenzo, and F. Las-Heras, 2010. Novel received signal Strength-based indoor location system: Development and testing. EURASIP J. Wireless Commun. Networking, 10.1155/2010/254345
Bahl, P. and V.N. Padmanabhan, 2000. RADAR: An in-building RF-based user location and tracking system. Comput. Comm. Soc., 2: 775-784.
Chang, C.K., H.Y. Jiang, H. Ming and K. Oyama, 2009. Situ: A Situation-theoretic approach to Context-aware service evolution. Services Comput., 2: 261-275.
Chen, T.S., J. Chen and Y.R. Tu, 2011. A study of bidirectional antenna for indoor localization using zigbee wireless sensor network. Inform. Technol. J., 10: 1836-1841.
Dong-Mei, H. and Z. Qin-Yu, 2010. Impulse radio ultra-wide-band through wall imaging radar based on multiple-input multiple-output antenna arrays. Inform. Technol. J., 9: 782-789.
CrossRef | Direct Link |
Harter, A., A. Hopper, P. Steggles, A. Ward and P. Webster, 1999. The anatomy of a context-aware application. Proceedings of the 5th Annual ACM/IEEE International Conference on Mobile Computing and Networking, August 15-19, 1999, Seattle, WA, USA., pp: 59-68.
Helal, S., S. Mitra, J.S. Wong, C.K. Chang and M. Mokhtari, 2008. Smart Homes and Health Telematics. Springer, New York.
Hightower, J. and G. Borriello, 2001. Location systems for ubiquitous computing. IEEE Comput., 34: 57-66.
CrossRef | Direct Link |
Hussain, C.S., C.S. Ahmed, A.H. Akbar, A.K. Bashir, K.H. Kim and W.S. Yoon, 2008. Ubiquitous service discovery in pervasive computing environment. Inform. Technol. J., 7: 533-536.
CrossRef | Direct Link |
Jabri, I., A. Soudani, N. Krommenacker, T. Divoux and S. Nasri, 2008. QoS protocol specification for IEEE 802.11 WLAN. Inform. Technol. J., 7: 549-559.
CrossRef | Direct Link |
Kaemarungsi, K., 2005. Design of indoor positioning systems based on location fingerprinting technique. Ph.D.Thesis, University of Pittsburgh, USA.
Liao, H.C. and C.C. Tu, 2007. A RDF and owl-based temporal context reasoning model for smart home. Inform. Technol. J., 6: 1130-1138.
CrossRef | Direct Link |
Ming, H., C.K. Chang, K. Oyama and H.I. Yang, 2010. Reasoning about human intention change for individualized runtime software service evolution. Proceedings of the IEEE 34th Annual Computer Software and Applications Conference, July 19-23, 2010, Seoul, pp: 289-296.
Niculescu, D. and B. Nath, 2003. Ad hoc positioning systems (APS) using AoA. Proceedings of the 22nd Annual Joint Conference of the IEEE Computer and Communications Society, March 30-April 3, 2003, IEEE Xplore London, pp: 1734-1743.
Pahlavan, K., X. Li and J.P. Makela, 2002. Indoor geolocation science and technology. IEEE Commun. Mag, 40: 112-118.
CrossRef | Direct Link |
Priyantha, N., A. Miu, H. Balakrishnan and S. Teller, 2001. The cricket compass for context-aware mobile applications. Proceedings of the 7th Annual International Conference on Mobile Computing and Networking, July 2007, ACM New York, pp: 1-14.
Savvides, A.. C.C. Han and M. Srivastava, 2001. Dynamic fine-gained localization in ad-hoc networks of sensors. Proceedings of the 7th International Conference on Mobile Computing and Networking, July 16-21, 2001, ACM Press, Rome, Italy, pp: 166-179.
Small, J., A. Smailagic and D.P. Siewiorek, 2001. Determining user location for context-aware computing through the use of a wireless LAN infrastructure. ACM Mobile Networks Appli., 6: 1-8.
Direct Link |
Subhan, F., H. Hasbullah, A. Rozyyev and S.T. Bakhsh, 2011. Handover in bluetooth networks using signal parameters. Inform. Technol. J., 10: 965-973.
Direct Link |
Wang, Y., T. Liang, X. Yang and D. Zhang, 2008. Scalable and effective cluster based routing algorithm using nodes location for mobile Ad Hoc networks. Inform. Technol. J., 7: 958-971.
CrossRef | Direct Link |
Want, R., A. Hopper, V. Falcao and J. Gibbons, 1992. The active badge location system. ACM Trans. Inform. Syst., 10: 91-102.
Wheeler, A., 2007. Commercial applications of wireless sensor networks using zigbee. IEEE Comm. Mag., 45: 70-77.
Xiang, Z., S. Song, J. Chen, H. Wang, J. Huang and X. Gao, 2004. A wireless LAN-based indoor positioning technology. IBM J. Res. Dev., 48: 617-626.
Xiaohu, G. and Z. Guangxi, 2006. Empowering ubiquitous services in next-generation smart homes. Inform. Technol. J., 5: 64-69.
CrossRef | Direct Link |
Xu, B., Z. Yu, Y.J. Ge and P.Y. Han, 2011. A light RSSI-based three dimensions location estimation algorithm for indoor zigbee network. Appl. Mech. Mater., 88/89: 427-431.
Xu, B., Z.P. Hu, P.Y. Han and Y.J. Ge, 2011. Location-aware elderly personal safety checking in smart home. Lecture Notes Electr. Eng., 107: 1583-1590.
Yang, H.I. and A. Helal, 2008. Safety enhancing mechanisms for pervasive computing systems in intelligent environments. Proceedings of 6th Annual IEEE International Conference on Pervasive Computing and Communications, March 17-21, 2008, Hong Kong, pp: 525-530.
Yilmaz, C., 2010. Implementation of programmable logic controller-based home automation. J. Applied Sci., 10: 1449-1454.
CrossRef | Direct Link |
Yin, J., M. Gu and J. Zhang, 2011. The expanded state space kalman filter for GPS navigation. Inform. Tech. J., 10: 2091-2097.
Youssef, M.A., A. Agrawala and A.U. Shankar, 2003. WLAN location determination via clustering and probability distributions. Proceedings of the 1st IEEE International Conference on Pervasive Computing and Communications, March 23-26, IEEE Computer Society, Washington, pp: 143-150.
Zhou, F., H.I. Yang, J.M.R. Alamo, J.S. Wong, and C.K. Chang, 2010. Mobile personal health care system for patients with diabetes. Aging Friendly Technol. Health Independence, 6159: 94-101.