Essay: Data Mining in Internet of Things

 
Abstract
Internet of things (IoT) is a network of objects such as, vehicles, buildings and
other items embedded with sensors, software and network connectivity which assist
them to collect and exchange data. As this network of objects generate massive
amount of data stream, the challenge here is to convert the generated or captured
data by IoT into information by means of data mining to provide ecient, timely
knowledge to aid decision making. Applying traditional Data mining algorithms to
the IoT platform causes several issues such as, handling the distributed nature of
IoT, resource constraints of things etc. Knowledge discovery from collected data
can be handled in two ways. The rst approach is to send the collected data to the
cloud computing platform and perform the data mining operations on the cloud.
But, this methodology increases the latency and response time which might not be
helpful in handling real time applications. So to decrease the latency and increase
the throughput, a new approach could be used to mine the data at IoT level itself,
which can be referred to as Edge Mining or Far Edge Computing. Limited battery,
processing capabilities, memory constraints etc. needs to be taken into account
during Edge Mining to achieve the expected eciency. In this project, we will
survey the existing mining algorithms in the domain of IoT, try to classify them
and, future directions to further improve the eciency of these algorithms. We will
also be reviewing the existing work done associated with the processes involved in
knowledge discovery (Data Collection, Data processing, Data transformation etc.)
and the research involved with these processes in the domain of IoT.
1
1 Introduction
IoT (Internet of Things) [1] [2] can be simply described as connecting with Internet,
dierent things that exists around us so that the formed network can collect and exchange
data. A thing can be any item in the real world that might be connected to
Internet to communicate with other things, person, server or machine. The things in
IoT can be said of possessing traits like cheap sensors, low energy communication and
limited processing capabilities. “Things,” in the IoT sense, can refer to a wide variety of
devices such as heart monitoring implants, biochip transponders on farm animals, electric
clams in coastal waters, automobiles with built-in sensors, DNA analysis devices for
environmental/food/pathogen monitoring, eld operation devices that assist reghters
in search and rescue operations, Radio Frequency Identication tags (RFID), Wireless
Sensor Networks (WSN), smartphones or wearable devices [3] [4]. IoT has the potential
to lead the next technological revolution across multiple domains including laying the
foundation for future smart cities, smart industries, health-care, disaster management,
retail, transportation, agriculture to name a few, which can improve the quality of life.
There have been plethora of good surveys presented each of which view IoT from dierent
perspective: challenges [5], applications [6], standards [7] and smartness [8] [9]. Another
study [10] described the overall design of IoT as a generic ve-layer architecture, from
bottom-up these layers are: edge-technology, access gateway, internet, middleware and
application.
It is estimated that, by 2020 there will be 50 billion things connected with Internet
against the population of 7.6 billion people around the globe, which renders the number
of 6.58 connected devices per person [11]. Enormous data generated by things will be an
important source of Big data, creating massive data streams; that will demand immediate
and context aware responses. Currently, many applications rely on data and services
hosted on remote clouds. However, more devices connected to the IoT paradigm will only
increase, augmenting the amount of data going to the cloud for processing. Pushing data
to the cloud is expensive and can cause multiple problems like data explosion. Because
of this, it is necessary to manage this enormous amount of data generated by “things”
and extract the knowledge from it in real time. This is where Knowledge Discovery from
2
Data (KDD) and data mining comes into play, as these technologies provides the means
to nd the useful hidden information collected from things, which can be used further to
enhance the performance of system or to provide new useful services this new network
can provide.
There is a great vision that all things can be easily controlled and monitored, can
be identied automatically by other things, can communicate with each other through
internet, and can even make decisions by themselves [12]. The IoT has very complex data
types and data generation requirements. So in order to make IoT smarter the mining
algorithms used in the paradigm of IoT also need to deal with 5 Vs of data generated
namely: Volume, Variety, Variability, Velocity and Veracity which all can be described
brie
y as below for the domain of IoT:
 Volume: The sheer size of data generated by dierent things, which is on a scale
never experienced before and to gather this data in a cost-ecient and energye
cient way when network bandwidth and resources may be at premium.
 Variety: Data collected from dierent things have heterogeneous data types, differences
in representations and semantic interpretations. Dierent kind of data
collected by things includes sensor readings, RFID tags, GPS readings, video, images,
social feeds (e.g. Facebook posts, tweets).
 Variability: Things senses and collects data from dierent environments like crowd
sensing data on the streets or movement of objects inside the home to detect the
falling of old persons. This kind of data changes rapidly so along this V we need to
take care of the rate at which data is changing.
 Velocity: The mining algorithms need to perform the task of knowledge discovery
from data in real time where millions of data points are being generated by things
in seconds such as by the sensors to detect fuel leakage in airplanes.
 Veracity: The data generated by things may be noisy, inaccurate or unreliable
and the mining algorithm should take care of this kind of data eciently in order
to discover helpful knowledge. Often in WSN only some of the sensors are selected
to sense the environment and collect data to preserve the lifetime of entire network.
3
This project tries to examine the dierent mining algorithm proposed in the domain
of IoT. According to [13] the data mining can be categorized by dierent views
such as: (i) knowledge view or data mining functions view which includes classication,
clustering, association analysis, discrimination, time series analysis, characterization and
outlier analysis (ii) utilized techniques view which includes machine learning, statistics,
pattern recognition, big data, support vector machine, rough set, neural networks and
evolutionary algorithms (iii) application view which includes industry, telecommunication,
banking, fraud analysis, biodata mining, stock market analysis, web mining, text
mining, social network and e-commerce.The authors in [14] tries to classify dierent mining
algorithms in IoT for data mining functions view and sub classify it further to examine
whether they are used for infrastructure of IoT(i.e. to enhance the performance of IoT
network) or for the applications of IoT. We will rst try to generalize the existing data
mining models and then will try to divide the dierent mining algorithms developed so
far according to the 5-dimensional metric we created. The rest of the paper is organized
as follows. We present the related work in section 2. Section 3 tries to generalize the
existing data mining models for the IoT and gives dierent categories of it. Section 4
denes the 5-Dimensional space in which the existing mining algorithms for IoT can be
classied. We present our conclusions in Section 5 of the paper.
2 Related Work
There has been signicant amount of work done to study dierent architectures, models
and algorithms for the IoT. The term Data Mining does not really present all the components
of KDD. Data mining is merely a step in the process of KDD. The entire data
mining process can be divided into dierent parts with major parts consisting of data
preprocessing (data cleaning, data integration, data selection and data transformation),
data mining step itself which discovers interesting knowledge about data in order to yield
fruitful results and data post-processing (pattern evaluation and visualization) to aid
proper decision making. All these steps of data mining is presented in gure 1 below
properly.
Starting with data pre-processing, as “things” in IoT have resource constraints within
4
Figure 1: Data mining as a step in knowledge discovery
them and they can be used in real time analysis the work done for this step can be
assessed in terms of energy saving and/or latency improvements. Authors at [15] proposed
an automatic time series modeling based data aggregation scheme in wireless sensor
networks to decrease the number of transmitted data values between sensor nodes and
aggregator by using time series prediction model and to save energy of the WSN. The
big data collection framework for water industry has been suggested in [16] to provide
useful insights consumers to proactively manage their water supplies and to utilities
management to achieve higher levels of sustainability in water supply. Extensible and
exible architecture for integrating data collected from WSN has been proposed using
REST based Web services as an inter-operable application layer that can be directly
integrated into other application domains for remote monitoring such as e-health care
services, smart homes, or even vehicular area networks (VAN) in [17]. Apart from that,
data collection and pre-processing frameworks and architectures has been proposed in
dierent application domains as well such as in medical Alarm-Net [18] or CodeBlue [19]
5
for environment monitoring.
Based on the data post-processing step there has been many interesting and innovative
applications built on top of data mining in IoT from the domains ranging from smart
home, smart cities, health-care, agriculture, industry monitoring, environment monitoring,
disaster management, vehicular systems etc. Several possible applications of the IoT
have been presented or are about to be presented in the near future as are mentioned in
several technical reports [20], [21], research papers [6], [10], and books [22], [23], [24] and
by international companies [25], [26].
There has been a number of surveys done to analyze dierent mining algorithms in
the context of IoT, which all view data mining in IoT from data mining functions view,
application view or utilized technique view. Some survey exists to incorporate the new
notion of edge mining or far-edge mining where many of the mining functionalities are
achieved by processing and discovering knowledge at IoT layer itself. But to our best
knowledge there exist no survey so far that gives useful insights for mining algorithms used
in both traditional mining in IoT and edge mining. Through this project we are trying
to integrate emerging edge mining algorithms in the algorithms proposed for traditional
IoT. In [14] authors examine dierent mining algorithms from mining functions view (i.e.
clustering, classication, frequent pattern mining). Dierent data mining algorithms and
challenges faced to integrate those algorithms to IoT has been discussed in [13]. A novel
distributed data-mining model to realize the seamless access between cloud computing
and distributed data mining has been proposed in [27]. Authors also analyzed dierent
edge mining algorithms in [28].
3 Data Mining Models for IoT
Data mining is no longer considered as an approach for traditional data analysis and
statistics. It has become an essential tool in Internet of things (IoT). Data mining faces a
number of challenges and technical issues in the rapidly changing real-time environment
of IoT. First of all, because of the real time environment of IoT, Data mining processes
should be really quick and ecient to support the need of real-time data analysis and
decision making. Second, the nature of the data is heterogeneous and it is distributed. It
6
is essential that Data mining processes should be capable enough to adopt the existing
distribution and heterogeneity of IoT. Third is the data quality control. It is important
to store and manage data properly to guarantee real time results. Last is the decision
making control. The characteristics such as rules used for decision making etc. need
to be calculated carefully. It is said that, RFID technology used in the supply chain
by supermarket, recording Electronic Product Code (EPC), location and time will be
generating 12.6 GB of data stream in one second and 544 TB of data per day [29]. The
challenges mentioned above can be concisely presented as shown in the gure-2.
Figure 2: Challenges in Internet of Things
Now we will be discussing dierent data mining models that can be objected to IoT.
3.1 Multi-layer data mining model in Internet of Things:
Internet of Things communicate and exchange information through equipment such as
sensors, GPS (global positioning system), radio frequency identication (RFID) etc.
Multi-layer data mining model is based on the RFID data mining framework and hierarchical
architecture of IoT and is divided into four layers and can be seen in Figure-3:
 Data collection layer: Data collection layer is the basis on which IoT is formed
and acts like the source of IoT as its main functionality is to recognize things and
collection information. RFID reader, receiver etc. are used as the components to
collect all kinds of data from objects such as GPS data, RFID stream data, location
and sensor data, satellite data, and so on. Collecting information from these
7
dierent kind of sources require dierent strategies. The components mentioned
earlier collect the data and transmit it to the upper gateway access points and so
on in the hierarchy. multiple issues such as energy consumption, fault tolerance,
data ltering etc. come into picture during the data collection process and they
must be handled carefully.
 Data management layer: The data collected from the homogeneous and heterogeneous
environments in the data collection layer is managed in this layer. According
to the RFID data framework[reference], an RFID cube has three tables: the
information table, the pause table and the graph table. These table have dierent
functionalities and in data management layer, records from the pause table (used
to store the position information of the data) are obtained to store and manage the
cleansed data.
 Event process layer: An event can be dened as a combination of multiple factors
such as integration of data, time etc. and it provides a high level of processing
mechanism in Internet of Things. Event ltering is the most important part of this
layer. Contents are collected, organized and analyzed in this layer through various
data mining strategies and passed to the upper layer based on the requirements of
the events.
 Data mining service layer: This layer depends deeply on the functionality of
lower layers. It provides services to the users based on the analyzed and processed
data. It mainly consists of three parts: 1. Data 2. Data Mining and 3. Knowledge.
3.2 Distributed data mining model in Internet of Things:
Unlike the traditional data, data in Internet of things is huge, distributive in nature, time
& location-related etc. These characteristics create several issues for central data mining
and hence, a new mining model known as Distributed data mining model was designed.
Using this model, high-performance requirements can be met reducing the computing
power and high storage capacity required in central data mining process.
8
Figure 3: Distributed Data Mining Model
The global control node is the core of the data mining system in this model. It selects
data mining algorithm and organizes mining data sets, then guides these sets to secondary
nodes, which will collect original data from various kinds of smart objects. The original
data will be stored in the local database after preprocess of data lter, abstraction and
compression. A local model is obtained with event lter, complex event detection and
data mining of local node. When needed, the local model will be controlled by the global
control node, and all set of local models will form the global model. Data of objects,
preprocessed data and information can be exchanged among secondary nodes. Multilayer
of agents based on union administrative mechanism controls the whole process.
3.3 Data mining model based on grid in Internet of Things:
Stankovski et al. [30] proposed a data mining grid based on which this model is designed.
Data mining model based on grid in Internet of Things consists of ve layers:
 IoT Resource Layer: Resource layer is formed with the combination of dierent
software and hardware modules.
 IoT Service Layer: Service layer consists of dierent divisions of services provided
by the function modules.
 Grid Middleware Layer: The main functionality of this layer is to solve the
problems generated in the network.
9
 Grid Mining Layer: This layer is responsible for data fusion.
 Grid Application Layer: This layer provides the interface services to the users
and consists of 4 modules.
Figure 4: Data Mining Model Based on Grid in Internet of Things.
Grid computing, similar to IoT is receiving a growing attention from research and industrial
communities. The basic idea of Grid is that users can make use of the computation
resources of Grid as the same as power resources. Various computing resources, data
resources and devices resources can be accessed or used conveniently. The basic idea
of IoT is to connect various smart objects via internet. Thus smart objects become intelligent,
context-awareness, and long-range operable. Therefore, we may regard smart
objects of IoT as a kind of resources for Grid computing, and then use data mining services
of Grid to implement the data mining operations for IoT. The dierences between
DataMiningGrid-based data mining model for IoT and DataMiningGrid is the part of
software and hardware resources. IoT provides more types of hardware, e.g., RFID tags,
RFID Readers, WSN, WSAN and Sensor networks etc. It also oers various software resources,
e.g., event processing algorithms, data warehouse and data mining applications
etc.
10
3.4 Data mining model for IoT from multi-technology integra-
tion perspective:
The Internet of Things is one of the most important development directions of the nextgeneration
Internet. At the same time, there are still a number of new directions, e.g.,
trusted network, ubiquitous network, grid computing, cloud computing etc. Therefore,
from the perspective of multi-technology integration, a new data-mining model for IoT
has been proposed by Shen Bin et al. [31]. In this model, data comes from the contextawareness
of individuals, smart objects or the environment. 128-bit IPV6 address is
Figure 5: Data mining model for IoT from multi-technology integration perspective.
adopted, and a variety of ubiquitous ways are provided for accessing to the future Internet,
such as: Intranet/Internet, FTTx/xDSL, sensor devices, RFID, WLAN/WiMAX,
2.5/3/4G mobile access and so on. Trusted control plane is able to ensure credibility and
controllability of data transmission. Data mining tools and algorithms submit gained
knowledge to various service-oriented applications, such as intelligent transportation, intelligent
logistics etc.
11
4 Data Mining Algorithms for IoT
In this section we will try to distinguish dierent mining algorithms for the IoT based on
dierent parameters. We consider ve dierent parameters and we call it as 5-dimensional
system to classify a mining algorithm. These ve parameters are:
 Data mining algorithm functionality: Along this dimension we can classify
algorithm according to the data mining function that is used. The dierent mining
functions are clustering, classication, frequent pattern mining.
 Goal of algorithm: Along this dimension we can discuss about whether the
mining used is for the purpose of enhancing the performance of infrastructure of
IoT system or to discover useful knowledge for the application built on top of
IoT infrastructure. So the value for this dimension can be infrastructure if mining
algorithm is enhancing the performance of system or service if it discovers knowledge
for the application used.
 Place of mining: This is to distinguish whether the selected mining algorithm
is being processed at the edge or in the Internet based Cloud services (IBC)/data
servers.
 Mining model used: Along this dimension we will distinguish which mining
model is being used from section 3 for the algorithm in focus.
 “Things” involved: This parameter will be used to specify which devices are
considered for algorithm. The devices or things includes RFID, WSN, smartphones,
cars, wearable devices, etc.
We present below the table describing dierent mining algorithms in IoT and where do
they stand according to our 5-D system.
12
Functionality Goal Place Model “Things” References
Clustering Infrastructure Cloud
Distributed
Grid
WSN
[32], [33], [34],
[35], [36]
Frequent Pattern-
Mining
Clustering
Service Cloud
Multi-layer
Multi-Technology
WSN [37], [54]
Clustering Service Cloud Multi-layer GPS [38]
Clustering Service Cloud
Multi-layer
Multi-Technology
GPS
Sensors
[39], [40]
Clustering Service Cloud
Distributed
Grid
Multi-layer
GPS
Smart phone
PDA
[41], [42]
Clustering Service Cloud
Multi-layer
Multi-Technology
Smart phone [43]
Classication Infrastructure Cloud Multi-layer RFID [44]
Classication Service Cloud
Multi-layer
Multi-Technology
GPS
Smart phone
Sensors
[45], [46], [47]
Classication Service Edge
Multi-layer
Multi-Technology
Distributed
Infraredsensors
WSN
[48]
Frequent Pattern-
Mining
Service Cloud
Multi-layer
Grid
RFID [49], [50]
Frequent Pattern-
Mining
Service Cloud
Multi-layer
Grid
Multi-Technology
RFID
GPS
[51], [52], [53]
Sequential Mining
Clustering
Service Cloud
Multi-layer
Multi-Technology
RFID
GPS
[55]
13
Frequent Pattern-
Mining
Infrastructure Cloud
Multi-layer
Multi-Technology
Distributed
RFID
GPS
Smart phone
[56], [57], [58]
Frequent Pattern-
Mining
Clustering
Classication
Service Cloud
Multi-layer
Multi-Technology
GPS
Smart phone
Sensors
[59]
Clustering
Classication
Infrastructure Edge Multi-layer WSN [60], [62]
Classication Service Edge
Multi-layer
Grid
WSN [61]
Table 1: Classication of dierent algorithms according to 5-D system
5 Conclusion
The Internet of Things concept arises from the need to manage, automate, and explore
all devices, instruments, and sensors in the world. In order to make wise decisions both
for people and for the things in IoT, data mining technologies are integrated with IoT
technologies for decision making support and system optimization. Data mining involves
discovering novel, interesting, and potentially useful patterns from data and applying
algorithms to the extraction of hidden information. In this paper, aiming at the characteristic
of massive data in Internet of Things, we discuss about the dierent data mining
models for Internet of Things, which when considered during the real time implementation
will be helpful for the applications working on IoT and setting up the environment
as needed. During this survey we tried to include more parameters while examining the
existing mining algorithms in IoT. We also tried to include the notion of Edge Computing
[63], [64] and Fog computing [65] which can be used to reduce the bandwidth
consumption, latency by mining the collected data at the edge of the network. There
also exist several challenges to focus on for the domain of data mining in IoT such as:
capturing and processing the massive amount of data streams generated by things. Converting
the traditional mining algorithms to t the distributed nature of IoT and that
14
too generating massive amount of data (Big Data) is a big challenge. Secondly, data collected
from things contain very personal and private data like daily activity of a person,
medical-records, banking transactions to name a few. The solutions provided for mining
these data should also focus on privacy and security of data and results.
References
[1] Ashton, K., 2009. That `internet of things\’ thing. RFiD Journal, 22(7), pp.97-114.
[2] Auto-ID Labs, Massachusetts Institute of Technology, 2012, available at
http://www.autoidlabs.org/.
[3] Erlich, Y., 2015. A vision for ubiquitous sequencing. Genome research, 25(10),
pp.1411-1416.
[4] I. Wigmore: “Internet of Things (IoT)”. TechTarget, June 2014
[5] Miorandi, D., Sicari, S., De Pellegrini, F. and Chlamtac, I., 2012. Internet of things:
Vision, applications and research challenges. Ad Hoc Networks, 10(7), pp.1497-1516.
[6] Domingo, M.C., 2012. An overview of the Internet of Things for people with disabilities.
Journal of Network and Computer Applications, 35(2), pp.584-596.
[7] Palattella, M.R., Accettura, N., Vilajosana, X., Watteyne, T., Grieco, L.A., Boggia,
G. and Dohler, M., 2013. Standardized protocol stack for the internet of (important)
things. Communications Surveys & Tutorials, IEEE, 15(3), pp.1389-1406.
[8] Kulkarni, R.V., Forster, A. and Venayagamoorthy, G.K., 2011. Computational intelligence
in wireless sensor networks: a survey. Communications Surveys & Tutorials,
IEEE, 13(1), pp.68-96.
[9] Lopez, T.S., Ranasinghe, D.C., Harrison, M. and McFarlane, D., 2012. Adding sense
to the Internet of Things. Personal and Ubiquitous Computing, 16(3), pp.291-308.
[10] Bandyopadhyay, D. and Sen, J., 2011. Internet of things: Applications and challenges
in technology and standardization. Wireless Personal Communications, 58(1), pp.49-
69.
15
[11] Mukhopadhyay, S.C. and Suryadevara, N.K., 2014. Internet of Things: Challenges
and Opportunities (pp. 1-17). Springer International Publishing.
[12] Tsai, C.W., Lai, C.F. and Vasilakos, A.V., 2014. Future Internet of Things: open
issues and challenges. Wireless Networks, 20(8), pp.2201-2217.
[13] Chen, F., Deng, P., Wan, J., Zhang, D., Vasilakos, A.V. and Rong, X., 2015. Data
mining for the internet of things: literature review and challenges. International Journal
of Distributed Sensor Networks, 2015, p.12.
[14] Tsai, C.W., Lai, C.F., Chiang, M.C. and Yang, L.T., 2014. Data mining for internet
of things: a survey. Communications Surveys & Tutorials, IEEE, 16(1), pp.77-97.
[15] Wang, Y. and Li, G., 2012. An Ecient Data Aggregation Scheme in Wireless Sensor
Networks. In Internet of Things (pp. 25-32). Springer Berlin Heidelberg.
[16] Koo, D., Piratla, K. and Matthews, C.J., 2015. Towards Sustainable Water Supply:
Schematic Development of Big Data Collection Using Internet of Things (IoT).
Procedia Engineering, 118, pp.489-497.
[17] Piyare, R. and Lee, S.R., 2013. Towards internet of things (iots): Integration of
wireless sensor network to cloud services for data collection and sharing. arXiv preprint
arXiv:1310.2095.
[18] Wood, A., Virone, G., Doan, T., Cao, Q., Selavo, L., Wu, Y., Fang, L., He, Z.,
Lin, S. and Stankovic, J., 2006. ALARM-NET: Wireless sensor networks for assistedliving
and residential monitoring. University of Virginia Computer Science Department
Technical Report, 2.
[19] Malan, D., Fulford-Jones, T., Welsh, M. and Moulton, S., 2004, April. Codeblue:
An ad hoc sensor network infrastructure for emergency medical care. In International
workshop on wearable and implantable body sensor networks (Vol. 5).
[20] Asin, A.L.I.C.I.A. and Gascon, D.A.V.I.D., 2012. 50 sensor applications for a smarter
world. Libelium Comunicaciones Distribuidas, Tech. Rep.
16
[21] R. Macmanus, \\Top 10 internet of things products of 2009,” 2009, available
at http://www.readwriteweb.com/archives/top 10 internet of things products of
2009.php.
[22] Gusmeroli, S., Haller, S., Harrison, M., Kalaboukas, K., Tomasella, M., Vermesan,
O., Vogt, H. and Wouters, K., 2010. Vision and challenges for realising the internet
of things.
[23] Bassi, A. and Horn, G., 2008. Internet of Things in 2020: A Roadmap for the Future.
European Commission: Information Society and Media.
[24] Uckelmann, D., Harrison, M. and Michahelles, F., 2011. An architectural approach
towards the future internet of things (pp. 1-24). Springer Berlin Heidelberg.
[25] Smarter Planet, 2012, available at http://www.ibm.com/smarterplanet/th/en/overview/ideas/index.[26] Eye on Earth, 2012, available at http://watch.eyeonearth.org.
[27] Guan, W., Lu, H.J., Chen, J.J. and Wu, J., 2014, October. Research on the Distributed
Data Mining Cloud Framework Oriented Internet of Things. In Applied Mechanics
and Materials (Vol. 686, pp. 306-310). Trans Tech Publications.
[28] Gaura, E.I., Brusey, J., Allen, M., Wilkins, R., Goldsmith, D. and Rednic, R., 2013.
Edge mining the internet of things. Sensors Journal, IEEE, 13(10), pp.3816-3825.
[29] Zhang, C., Zeng, G., Wang, H. and Tu, X., 2012. Analysis on Data Mining Model
Objected to Internet of Things. International Journal of Advancements in Computing
Technology, 4(21).
[30] Stankovski, V., Swain, M., Kravtsov, V., Niessen, T., Wegener, D., Rohm, M.,
Trnkoczy, J., May, M., Franke, J., Schuster, A. and Dubitzky, W., 2008. Digging deep
into the data mine with DataMiningGrid. Internet Computing, IEEE, 12(6), pp.69-76.
[31] Bin, S., Yuan, L. and Xiaoyi, W., 2010, April. Research on data mining models
for the internet of things. In Image Analysis and Signal Processing (IASP), 2010
International Conference on (pp. 127-132). IEEE.
17
[32] Younis, O. and Fahmy, S., 2004, March. Distributed clustering in ad-hoc sensor
networks: A hybrid, energy-ecient approach. In INFOCOM 2004. Twenty-third
AnnualJoint Conference of the IEEE Computer and Communications Societies (Vol.
1). IEEE.
[33] Heinzelman, W.R., Chandrakasan, A. and Balakrishnan, H., 2000, January. Energye
cient communication protocol for wireless microsensor networks. In System sciences,
2000. Proceedings of the 33rd annual Hawaii international conference on (pp.
10-pp). IEEE.
[34] Ghiasi, S., Srivastava, A., Yang, X. and Sarrafzadeh, M., 2002. Optimal energy aware
clustering in sensor networks. Sensors, 2(7), pp.258-269.
[35] Choi, W., Shah, P. and Das, S.K., 2004, August. A framework for energy-saving
data gathering using two-phase clustering in wireless sensor networks. In Mobile and
Ubiquitous Systems: Networking and Services, 2004. MOBIQUITOUS 2004. The
First Annual International Conference on (pp. 203-212). IEEE.
[36] Kardeby, V., 2011. Automatic sensor clustering: connectivity for the internet of
things.
[37] Rashidi, P., Cook, D.J., Holder, L.B. and Schmitter-Edgecombe, M., 2011. Discovering
activities to recognize and track in a smart environment. Knowledge and Data
Engineering, IEEE Transactions on, 23(4), pp.527-539.
[38] Anastasiou, N., Horng, T.C. and Knottenbelt, W., 2011, May. Deriving generalised
stochastic Petri net performance models from high-precision location tracking data.
In Proceedings of the 5th International ICST Conference on Performance Evaluation
Methodologies and Tools (pp. 91-100). ICST (Institute for Computer Sciences, Social-
Informatics and Telecommunications Engineering).
[39] Schuster, E.W., Kumar, S., Sarma, S.E., Willers, J.L. and Milliken, G.A., 2011,
November. Infrastructure for data-driven agriculture: identifying management zones
for cotton using statistical modeling and machine learning techniques. In Emerging
18
Technologies for a Smarter World (CEWIT), 2011 8th International Conference &
Expo on. IEEE.
[40] Min, Z., Bei, W. and Chunyuan, G., 2011. Application study of precision agriculture
based on ontology in the internet of things environment. In Applied Informatics and
Communication (pp. 374-380). Springer Berlin Heidelberg.
[41] White, S. and Smyth, P., 2005, April. A Spectral Clustering Approach To Finding
Communities in Graph. In SDM (Vol. 5, pp. 76-84).
[42] Wang, L., 2011, October. Using the relationship of shared neighbors to nd hierarchical
overlapping communities for eective connectivity in IoT. In Pervasive Computing
and Applications (ICPCA), 2011 6th International Conference on (pp. 400-406).
IEEE.
[43] Jian, A., Xiaolin, G., Wendong, Z. and Jian, A., 2011, October. Nodes social relations
cognition for mobility-aware in the internet of things. In Internet of Things
(iThings/CPSCom), 2011 International Conference on and 4th International Conference
on Cyber, Physical and Social Computing (pp. 687-691). IEEE.
[44] Lee, Y.H., Kim, H.J., Roh, B.H., Yoo, S.W. and Oh, Y.C., 2005, December. Treebased
classication algorithm for heterogeneous unique item id schemes. In Embedded
and Ubiquitous Computing{EUC 2005 Workshops (pp. 1078-1087). Springer Berlin
Heidelberg.
[45] Horvitz, E.J., Apacible, J., Sarin, R. and Liao, L., 2012. Prediction, expectation,
and surprise: Methods, designs, and study of a deployed trac forecasting service.
arXiv preprint arXiv:1207.1352.
[46] Thompson, C., White, J., Dougherty, B., Albright, A. and Schmidt, D.C., 2010.
Using smartphones to detect car accidents and provide situational awareness to emergency
responders. In Mobile Wireless Middleware, Operating Systems, and Applications
(pp. 29-42). Springer Berlin Heidelberg.
19
[47] Perera, K. and Dias, D., 2011, June. An intelligent driver guidance tool using location
based services. In Spatial Data Mining and Geographical Knowledge Services
(ICSDM), 2011 IEEE International Conference on (pp. 246-251). IEEE.
[48] Sudhaakar, R., Sanzgiri, A., Demirbas, M. and Qiao, C., 2009, June. A plant-andplay
wireless sensor network system for gate monitoring. In World of Wireless, Mobile
and Multimedia Networks & Workshops, 2009. WoWMoM 2009. IEEE International
Symposium on a (pp. 1-9). IEEE.
[49] Li, X., Liu, J., Sheng, Q.Z. and Zhong, W., 2008. Time to live: temporal management
of large-scale RFID applications (No. SSE-2008-02, pp. 1-12).
[50] Li, X., Liu, J., Sheng, Q.Z., Zeadally, S. and Zhong, W., 2011. TMS-RFID: Temporal
management of large-scale RFID applications. Information Systems Frontiers, 13(4),
pp.481-500.
[51] Koperski, K. and Han, J., 1995, August. Discovery of spatial association rules in geographic
information databases. In Advances in spatial databases (pp. 47-66). Springer
Berlin Heidelberg.
[52] Yoo, J.S. and Shekhar, S., 2006. A joinless approach for mining spatial colocation
patterns. Knowledge and Data Engineering, IEEE Transactions on, 18(10), pp.1323-
1337.
[53] Kim, S.K., Lee, J.H., Ryu, K.H. and Kim, U., 2014. A framework of spatial colocation
pattern mining for ubiquitous GIS. Multimedia tools and applications, 71(1),
pp.199-218.
[54] Rashidi, P. and Cook, D.J., 2010. An adaptive sensor mining framework for pervasive
computing applications. In Knowledge Discovery from Sensor Data (pp. 154-174).
Springer Berlin Heidelberg.
[55] Nazerfard, E., Rashidi, P. and Cook, D.J., 2010, December. Discovering temporal
features and relations of activity patterns. In Data Mining Workshops (ICDMW),
2010 IEEE International Conference on (pp. 1069-1075). IEEE.
20
[56] Galushka, M., Patterson, D. and Rooney, N., 2006. Temporal data mining for smart
homes. In Designing Smart Homes (pp. 85-108). Springer Berlin Heidelberg.
[57] Chen, L. and Ren, G., 2012. The research of data mining technology of privacy
preserving in sharing platform of internet of things. In Internet of Things (pp. 481-
485). Springer Berlin Heidelberg.
[58] Bekri, F.E. and Govardhan, A., 2011. Association of Data Mining and healthcare
domain: Issues and current state of the art. Global Journal of Computer Science and
Technology, 11(21).
[59] Choi, J., Shin, D. and Shin, D., 2006. Ubiquitous intelligent sensing system for a
smart home. In Structural, Syntactic, and Statistical Pattern Recognition (pp. 322-
330). Springer Berlin Heidelberg.
[60] Goldsmith, D. and Brusey, J., 2010, November. The Spanish Inquisition Protocol|
model based transmission reduction for wireless sensor networks. In Sensors,
2010 IEEE (pp. 2043-2048). IEEE.
[61] Brusey, J., Rednic, R., Gaura, E.I., Kemp, J. and Poole, N., 2009. Postural activity
monitoring for increasing safety in bomb disposal missions. Measurement Science and
Technology, 20(7), p.075204.
[62] Gaura, E.I., Brusey, J. and Wilkins, R., 2011, October. Bare necessities|Knowledgedriven
WSN design. In Sensors, 2011 IEEE (pp. 66-70). IEEE.
[63] M. T. B. M. W. S. eld Thomas Schimper, \\Mobile computing edge: A taxonomy,”
in The Sixth International Conference on Advances in Future Internet (AFIN 2014),
2014.
[64] P. N. N. Laboratory, \\Edge computing,” 2013, [Online; accessed 10-May-2015]. [Online].
Available: http://vis.pnnl.gov/pdf/
iers/EdgeComputing.pdf
[65] A. Froehlich, \\Iot: Out of the cloud & into the fog,” 2014, [Online; accessed
11-May-2015]. [Online]. Available: http://www.networkcomputing.com/cloudinfrastructure/
iot-out-of-the-cloud-and-into-the-fog/a/d-id/1298101
21