Big Sensing Data Curation in Cloud Data Center for Next Generation IoT and WSN

Scalable, IoT device responsive On-Cloud data curation will match the trend.

By Chi Yang, Deepak Puthal, Saraju P. Mohanty, and Elias Kougianos

Modern sensing devices play a pivotal role in achieving data acquisition, communication and dissemination for
the Internet-of-Things (IoT). Naturally, IoT applications and intelligent sensing systems supported by sensing
devices, such as wireless sensor networks (WSN), are closely coupled. Modern intelligent sensing systems
generate huge volumes of sensing data, well beyond the processing capabilities of common techniques and tools.
Hence, to collect, manage, and process IoT big sensing data within an acceptable time duration is a new challenge
for both research and industrial applications. The massive size, extreme complexity and high speed of big sensing
data bring new technical requirements including data collection, data storage, data organization, data analysis and
data publishing in real time when deploying real world IoT applications. To better facilitate these IoT applications,
the convergent research of WSN, big data, IoT and Cloud Computing is a natural scientific development trend. In
this paper, we concentrate on big sensing data curation and preparation issues with Cloud Computing under the
theme of IoT. There are three especially critical issues that need to be addressed, namely scalable big sensing data
cleaning, scalable big sensing data compression and Cloud based data curation response for IoT device
optimization. Viewed from the IoT side, all IoT sensing devices are integrated together in an adaptive solution and
upload their data onto the Cloud. The automatic responses from both the Cloud and intelligent sensors will change
the status or behavior of sensing devices, hence the status of the IoT itself.

1. INTRODUCTION

With fast growing attention from both academic research and industrial communities, the IoT consists of thousands
of connected devices (mostly modern sensing devices) for information monitoring, gathering, communicating,
exchanging, analyzing, decision making and finally instantly responding to acquired information to intelligently
control the behavior of physical devices or the factors of a real-world environment [1], [2], [3], [4]. In order to
achieve the above mentioned IoT applications, several pivotal technical challenges are involved including WSN,
cloud computing and big data [4], [5], [6]. It is well known that sensor devices and WSN in the IoT can generate
high reliability, variety, volume, value and velocity [7] data sets. The Cloud, with its massive computing power,
storage, and scalable provided software services, offers a promising platform to deal with the challenges brought
by IoT big data [8], [9]. Applications of big sensing data processing in the Cloud can be encountered in different
fields such as medical/health monitoring, weather forecasting, environmental monitoring, industry production,
social media analysis and business analysis. Figure 1
shows the application of integrated WSN, IoT and the
Cloud. Recently, some data curation techniques have
been developed for processing of sensing data on
cloud data centers, such as the Sensor-Cloud platform
[10]. However, those methods and platforms are far
from complete, and significant work is still needed. To
the best of our knowledge, when it comes to the
convergence of Cloud, IoT, WSN very little published
research can be found [2], [3], [12]. Hence, how to use
the massive computational power and common
platforms offered by the Cloud for processing big
sensing data from the IoT and offering response
actions to the IoT, motivates a new research direction.
This paper focuses on the data curation problems of
IoT big sensing data processing on the Cloud to
facilitate IoT applications and provides current trends FIGURE 1. Application specific integration of the
and future research scope of IoT big sensing data IoT, WSNs and Cloud.
processing.

When uploading integrated big sensing data onto a Cloud platform, a novel data cleaning technology will be
developed and deployed on the Cloud, including novel sensing data error detection and recovery algorithms to
capture errors, conflicts and missing data in big sensing data sets in real time. Simultaneously, recovery algorithms
will also be developed and deployed on the Cloud to provide compensating solutions for the detected errors and
defects. Intelligent sensing systems and WSNs can also conduct some basic error detection and recovery with the
in-network computation power and storage offered by smart sensors.

1
2. THE STATE-OF-THE-ART

In current Internet and IoT usage, we have entered the big data era of petabytes. Traditional database management
tools or data processing technologies become powerless due to limited resources of computation, storage and
communication. Big sensing data is generated with the features of high Volume, Value, Variety, Velocity, and
Veracity from a wide range of real world applications. Since the 1980s, data generation doubles its size every forty
months [7], [8]. In only one year (2012), the everyday data generation rate was equal to 2.5 quintillion (2.5×1018)
bytes. Currently, dataset sizes are measured in exabytes (1018 bytes). In 2015, there were around 10,000 exabytes
of digital data being generated. Following that digital data explosion, the size of big data is expected to surpass
40,000 exabytes by the year 2020 [7], [8]. In many application fields, including meteorology forecast,
connectomics, physics simulation, genomics, biological science, gene analysis and environmental research [3],
[6], [7], [8], [12], big data processing can even dictate the performance of whole systems. To offer solutions for
the new challenges brought by big data, more and more research attention has been attracted by big data in terms
of both fundamental research and technical applications. Data curation, as a significant step towards big sensing
data processing technology, is commonly deployed on a Cloud platform for achieving scalability, massive resource
access, real time data analytics and behavior control for the IoT. The data flow between source sensors (in WSNs
and the IoT) and the Cloud with classification is shown in Figure 2 which provides an overview of the IoT, WSNs,
and the Cloud
integration with data
processing at
different levels.

Big Sensing Data

Modern sensing
systems significantly
change everyday life FIGURE 2. Incremental data flow between sensors and the Cloud with classifications.
by giving us with the
capability to monitor, understand and interact with the physical environment around us [1], [5], [14]. In real world
applications, sensing systems are becoming much smaller, and smarter with more connectivity and more mobile
capability [15]. The price for achieving the above functional improvement is higher data rates, more data storage
and more powerful data analysis requirements. As a result, big data sets with high speed, high dimensions and
high volume are introduced by countless sensing systems deployed in our environment [1], [7]. As important
sources of big data sets, sensing systems generate sensing data with extremely large volumes which are far beyond
the processing capability of common data processing software and tools. However, to collect, store, organize,
analyze and publish big sensing data from modern sensing systems in real time are critical and essential targets
when deploying most of real world sensing systems [16]. In addition, the privacy and security issues of big sensing
are widely important [11], [17].

Cloud Computing

The National Institute of Standards and Technology (NIST) defined the Cloud as a framework for enabling
convenient, on-demand network access to a shared pool of configurable computing resources such as servers,
storage, networks, applications, service, etc. These computational resources should be provided and accessed in a
time-efficient manner. At the same time, when acquiring these computational resources, the management effort or
interactions with service providers should be as little as possible. Based on the NIST definition, there are four
deployment models, three service distribution models and five important features for Cloud Computing. The
important five key characteristics of cloud computing include resource pooling, broad network access, measured
services, on-demand self-service and rapid elasticity. The core technologies are web service technologies,
virtualization and distributed programming models such as MapReduce [18], [19], [20] and, the much newer Spark
[21].

Intelligent Sensors and WSN

Sensors, processors and wireless communication devices are becoming much smaller, cheaper, smarter and reliable
day by day. Reliable and inexpensive sensor systems based on better computing and communication methods are
of the size of a credit card. The architecture of a node is integrated with sensing, processing and communicating.
Recently, development and research trends of WSNs are mainly influenced and propelled by the advances in

2
computing and communication [1], [2], [6], [22], [23]. In general, a wireless sensor network consists of intelligent
communicating sensors deployed across a landscape for sampling and interpreting real world phenomena, then it
forwards the results back to a base station or a user gateway [13], [24]. In addition, for current IoT applications,
some control feedback is also expected from either the inner WSN, or a third-part intelligent system such as the
Cloud to change the behavior of sensing devices in the WSN. A WSN is a typical distributed system that generates
and processes correlated temporal and spatial data from multiple data sources. Under the theme of IoT, all these
different sensing devices form different complex networks and generate huge amount of continuous heterogeneous
data sets which require to be integrated, cleaned, classified, compressed, stored, exploited and analyzed [7], [25],
[26]. These data always need third party processing such as cloud and processing results used as feedback to
change sensing device behavior for optimizing IoT component performance [5].

The Internet of Things (IoT)

In brief, the Internet of Things (IoT) can be characterized as a wireless network of internet-connected smart sensing
devices ready to gather and transmit data with the support of embedded sensors or sensor networks. The IoT
consists of intelligent physical devices connected through techniques including electronics, software, sensors,
actuators, and network connectivity for data acquisition, communication and dissemination [5]. The three
components of the IOT are “Things”, the Internet, and connectivity. But the value (data) generated, transmitted
and interpreted in the IoT plays a critical role in smoothing the breach between the digital and the physical world
for setting up automated systems. In typical IoT applications, smart objects can be sensed, evaluated and controlled
in a remote manner through current popular network infrastructures. This working model of the IoT means great
opportunities and trends for integrating our physical real world with our computing systems through networks and
smart sensors. It will inevitably make significant improvement for peoples’ lives in terms of efficiency, accuracy,
economic profit and greatly reduced human intervention [1], [2], [15]. Specifically, sensors and actuators are the
backbone physical equipment for realizing IoT applications. According to the literature, it is estimated that the IoT
will contain more than 50 billion objects by 2020 [12]. Based on those smart sensed, connected and controlled
objects, heterogeneous big data sets with huge volumes are expected. To index, store and analyze those big IoT
sensing data becomes more and more important.

3. A SPECIFIC EXAMPLE FRAMEWORK

With the advance of modern smart sensing technologies, big sensing data and the IoT are emerging standards
applied to datasets which may not be digested with efficiency and effectiveness using common data processing
software, tools and techniques. Because IoT big sensing data sets are often from various structured or unstructured
physical devices with the characteristics of high volume, fast data rate and unreliable value, data curation has to
be done to guarantee the data quality and minimized size. It is also important for making decisions to change the
status of the IoT devices. Without successful data curation, it is impossible to calculate optimized strategies for
changing device behavior for the IoT. The Cloud, with its enormous power in computation and huge storage,
enables clients to deploy big data curation without requiring heavy assets. IoT big sensing data sets can be widely
encountered in industrial and scientific activities. For example, high volume of big data from body sensors is
generated by body monitoring equipment. These data sets are uploaded to the Cloud by U.S. hospitals. How to use
those data for disease analysis with acceptable accuracy and efficiency poses an interesting topic. In order to
process big sensing data efficiently on the Cloud and generate better IoT services, several critical issues should be
discussed including reduction of the big data size, data quality, fast query of big sensing data and data curation
result for interactions with smart sensing devices. To cope with the above challenges and issues, a framework is
proposed to offer a data curation solution with the convergent study of big data, WSN, Cloud Computing and the
IoT.

As shown in Figure 3, smart sensing devices in the IoT generate big sensing data which is filtered and aggregated
as soon as sensing devices perform sampling and data communication to form big sensing data streams. At this
stage, heterogeneous big sensing data sets can be integrated with some lightweight method before they are
forwarded to the Cloud for more complicated data processing.

3
When big sensing data or data streams are uploaded to cloud data centers, the first important stage of cleaning for
our proposed on-Cloud data curation framework starts. Specifically, to clean the heterogeneous data or data
streams from multiple data sources, there are three main tasks to be finalized including: (1) error detection, (2)
error recovery, and (3) data consistency and redundancy check. In terms of error detection, WSN big data set from
sensing devices or WSNs are commonly subject to corruption and error because of the low reliability of wireless
communications, signal processing inaccuracy and hardware defects in the nodes. For achieving high quality WSN
application, the first step
is to guarantee that the
data received is accurate,
logically connected and
clean. However, sensor
data error and cleaning is
still a challenging issue
and new methods are
required to solve the
problem. To the best of
our knowledge, there are
few research works on
in-network WSN error
detection recovery
techniques which are
limited by the
computation power,
storage, wireless
network energy
consumption and time
latency. With the support FIGURE 3. IoT generated big data curation steps in a cloud data center.
of Cloud computing,
scalable fast error detection and scalable fast error recovery will be possible for real time big sensing data cleaning.
In addition, in the process of error detection and recovery, the Cloud environment is helpful in analyzing the error
sources and generate feedbacks to the IoT to change sensing device behavior.

In addition to data cleaning on the Cloud, data compression should be performed for reducing big sensing data
size, and reducing the future data processing time. Under the theme of compression, spatiotemporal features or
other sensing data correlations can be exploited. As shown in FIGURE 3, several data compression techniques
have been or will be developed. For example, by discovering spatial correlations in big data [20], multiple cluster
structures can be obtained from a graph data set, then all edges in a cluster can share similar time series of data of
a graph. Based on that partition, the workload within a cluster can be greatly reduced by the similarities of time
series based inference. Temporal data compression can be carried according to the order of data items or time
series based prediction using temporal correlation. In addition, the data prediction models can be improved and
modified according to application requirement. By compression on the Cloud, sensing data size can be
significantly reduced compared to only in-network lightweight data suppression by sensing devices themselves
[27]. Meanwhile the proposed data suppression technologies in this framework should be able to guarantee
acceptable data accuracy [27], [28].

As shown in figure 3, after the sensing data cleaning and compression stages, the third component of feedback
based IoT sensing device and Cloud optimized control follows. Specifically, in the process of big sensing data
cleaning and compression, some optimization issues have been already detected. For example, how to assign the
workload of data cleaning or compression between sensing devices and the Cloud to exploit the full potential of
the Cloud and smart sensors is an interesting optimization problem. In other words, the trade-off between using
the Cloud or using smart sensor devices for computing will be. Furthermore, because both data cleaning and
compression buffer and filter big sensing data, some data analytical functions can be performed here. From Figure
3, at the stage of curation feedback for IoT devices, the filtering process will be combined with more data analysis
techniques for generating feedbacks. Following the proposed cleaning and compression in the framework, heuristic
or game theory based algorithms could be adopted for designing adaptive mechanisms in sensing devices of the
IoT. Furthermore, data errors are an indication of device failure or network defect. To understand these errors as
a data curation result can be useful in changing IoT devices to other status in both hardware and software levels.
For example, for certain mobile sensing devices network systems, our big sensing data curation not only finds and
corrects the errors for the data set from them, it also sends feedback to mobile sensing devices in the network to
move to right places and to maintain healthy network topology.

4
4. THOUGHTS FOR FUTURE DIRECTIONS

In the process of building up the proposed On-Cloud IoT big sensing data curation framework, different research
scopes and aspects should be discussed, as shown in FIGURE 4. Specifically, the following research objectives
should be achieved.

Sensing Data Error Classification

To detect WSN sensing data errors,
a categorization is performed first to
formally define error types. With
that classification, the network
features for the cluster-head
network WSN topology are
analyzed and used for error
detection. Specifically, in big
sensing data cleaning, we use the
scale-free topology of the network
for error detection. Based on that
topology constraint, error detection
and recovery strategies can be
designed within limited spatial-
temporal data blocks rather than
FIGURE 4. On-Cloud IoT big sensing data curation framework with
traversing the complete big sensing associated properties.
data set.

Distributed Error Detection

For fast detection of data errors in big sensing data, and to make use of the cluster location feature of data, novel
data error detection techniques are designed by exploiting the storage and computing potential of Cloud. The error
detection and localization can be significantly accelerated because the detecting algorithm makes full use of
complex network system topology and isolates the searching and comparing inside a sub-structure with high
confidence. These detection and localization tasks are distributed to the Cloud with the MapReduce tool. The
trade-offs between error detection efficiency and detection accuracy will also be considered and analyzed.

Distributed Error Recovery

A novel approach can be developed based on the prediction of recovery replacement data by making multiple data
source approximations. The approximation process will use coverage information carried by data units to limit the
algorithm in a small cluster of sensing data instead of the whole data spectrum. Specifically, in each sensing data
cluster, a Euclidean distance based approximation is proposed to calculate a time series prediction curve. With the
calculated time series, a detected error can be recovered with a predicted data value approximately. The proposed
error recovery approach should achieve high accuracy in data approximation to replace the original data error. At
the same time, with MapReduce based implementation for scalability, the experimental results also show
significant efficiency on time saving.

Scalable and Distributed Compression

To offer novel compression methods on the Cloud for processing big sensor data sets, two important factors should
be taken into consideration. Firstly, because of the sensor data strong topology and graph features, the compression
should exploit them. Secondly, traditional compression is not scalable in terms of the Cloud environment. How to
make it more scalable becomes critical. In our research, we combine our graph based compression and MapReduce
to achieve the objective of distributed compression. Different data compression models will be developed and
adopted. The compression can happen at data unit level, time series level, or even compound data blocks level.

Time Efficiency
Many big data processing applications have time limitation or high time efficiency requirement. In this paper, we
highlighted that there is need of lightweight algorithms including compression and cleaning. So, under this theme,
optimization, approximation and fast processing techniques are discussed.

5
Data Similarity
The clustering algorithm is developed for partitioning node sets and can compute the similarity between two time
series. If two time series are similar to each other, they can be used for operations including mutual replacement
and recovery. etc. So, to define appropriate similarity models is critical. Traditional similarity definitions and
models should be calibrated before deployment. In our framework, distance calculation can evaluate similarity,
hence to carry out further operations among time series, temporal and spatial prediction models should be designed.
Currently, we have proposed some temporal prediction models based on data trend method and regression based
method. However, there is still some disadvantages for our provided prediction models when applying them for
cluster topologies. For instance, under the situation where two time series have similarity with sinusoidal functions,
the regression based prediction models may totally miss effects. To offer a solution for the problem caused by the
prediction models based on standard regression, we plan to offer a novel prediction model based on improved data
regression with historical time series record. Suppose that there exist two time series, denoted as T1{t11, t12, …,
t1m}, and T2{t21, t22, …, t2m} being involved in the similarity computation. m is the data sampling time stamp. The
solution is designed for predicting the average dissimilarity of data trend for T1 and T2 in the next m rounds. A
dissimilarity vector V(d1, d2, …, dm) is calculated, where di=t1i-t2i. With the trends vector V, we redesign a
regression model with a novel weight assignment to calculate the average dissimilarity from T1 and T2 [29].

Data Accuracy
Based on previous work of sensor data processing on the Cloud [23], [25], [26], fast detection of data errors and
recovery in big sensing are quite challenging. For example, it is still an open and challenging topic to quickly
locate and find sensor errors in a WSN by using cloud computational power. In our work, we initiate the process
of error detection by using error models. If we compare with error detection in WSN in-network, our approach
utilizes the massive data processing ability of Cloud to accelerate speed of error detection [28]. Furthermore, in
more effective manner the topology features of complex networks are also analyzed in combination with the Cloud.
Our proposed solution mainly focuses to achieve significance in time performance gains in detection of error with
high accuracy with more consideration for optimized control of sensing devices in the IoT [30].

Scalability
As analyzed above, we can improve the efficiency in compression, cleaning and evaluation of big sensor data by
deploying our data processing lifecycle on cloud data centers. However, to implement those techniques and make
them scalable on the Cloud, our potential new design should offer better scalability. As a matter of fact, by the
implementation with MapReduce and Hadoop, we can guarantee the scalability of the techniques in the given
lifecycle.

Heuristic and Game Theory based Methods

WSN applications are quite customer oriented applications [31]. In other words, the understanding of sensor data
is different from user to user. So, it is possible to get knowledge from domain experts during the data processing.
In other words, heuristic methods can be developed for guiding the sensing device behavior in the IoT.
Furthermore, when changing the sensing device behavior using our proposed data curation feedbacks, game theory
can be involved for selecting among multiple candidate devices or systems in the IoT.

Optimized Workload Allocation between Cloud and IoT devices

Based on different processing capability and process models, data curation tasks can be assigned to the Cloud or
smart sensing devices dynamically. Specifically, the trade-off between Cloud and device curation should be
studied with the consideration of sensing data stream changes. Then, an adaptive approach will be developed to
dynamically evaluate and schedule different tasks of sensing data curation. An optimized scheduling result will be
generated to benefit both data curation targets and future IoT services.

Device Behavior Changes based on Data Curation

Sensing device behavior changes can happen in both the physical behavior of IoT devices such as location, and
software level behavior such as data rate, communication topology and other adaptive algorithms. For example, in
a sensing data curation process, if our Cloud-based error detection locates a sensing data source (device) reporting
continuous errors, it can send feedback to the IoT to actuate some mobile sensing device to replace that data source
to sustain a healthy IoT sensing data service.

5. CONCLUSIONS

This article presented a general roadmap for IoT big sensing data curation on the Cloud with background studies.
This roadmap was designed according to current research trends, limitations and future directions. It combines
different technologies from several fields including WSN, IoT, big data and the Cloud. The interaction and

6
correlation between those four technologies was analysed. Especially the correlation between big data and WSN
in IoT, the correlation between big data and Cloud, and the correlation between Cloud and WSN in IoT were
investigated with a logical connected integrity. In addition, the interactions between Cloud and IoT were discussed.
To the best of our knowledge, it was first proposed in this roadmap how to use Cloud big data curation results for
sensing device manipulation, including mobile sensors and actuating equipment in the IoT. The design and
construction of this roadmap would benefit current real world IoT big data processing applications such as
medical/health monitoring, weather forecasting, environmental monitoring, industry production, social media
analysis and business analysis where the users could access seamless IoT data services and realize remote physical
device control through their mobile devices such as cell phones in a pervasive environment. In other words, all the
complicated computation and communication of big data curation among sensing devices, IoT and Cloud platform,
could be hidden from end users.

ABOUT THE AUTHORS

Chi Yang ([email protected]) received his Master of Science by Research degree (in computer science) from
Swinburne University of Technology, Melbourne, Australia, and received his Ph.D in computer science at the
University of Technology, Sydney (UTS), Australia. He is also a doctorate student in the University of Western
Australia. Currently, Dr. Chi Yang is working for the Stratus project (https://stratus.org.nz/), as a full-time Postdoc.
Research Fellow in the Unitec Institution of Technology, Auckland, New Zealand. Dr. Chi Yang has published
research papers in different high-quality research conferences, journals and transactions. He is also PC member of
different international conferences, such as TrustCom, ACISP and BDSE. His major research interests include
WSN, IoT, Big Data Processing, Could Computing, parallel & distributed computing, privacy & security and
XML data streams.
Deepak Puthal ([email protected]) is a Lecturer in the School of Computing and Communications at
University of Technology Sydney (UTS), Australia. His research interests include cyber security, Internet of
Things, distributed computing, and wireless communications. Puthal has a PhD in Computer Science from UTS,
Australia. He has published in several international conferences and journals including IEEE and ACM
transactions.
Saraju P. Mohanty ([email protected]) is a Professor at the Department of Computer Science and
Engineering, University of North Texas. He is an inventor of 4 US patents. He is an author of 220 peer-reviewed
research articles and 3 books. He is currently the Editor-in-Chief (EiC) of the IEEE Consumer Electronics
Magazine. He currently serves on the editorial board of 6 peer-reviewed international journals including IEEE
Transactions on Computer-Aided Design of Integrated Circuits and Systems and ACM Journal on Emerging
Technologies in Computing Systems. Prof. Mohanty has been the Chair of Technical Committee on Very Large
Scale Integration (TCVLSI), IEEE Computer Society (IEEE-CS) to oversee a dozen of IEEE conferences. He
serves on the steering, organizing, and program committees of several international conferences. More about him
can be available from: http://www.smohanty.org.
Elias Kougianos ([email protected]) is Professor in Engineering Technology at the University of North Texas. He
obtained his Ph.D. in electrical engineering from Louisiana State University in 1997. He is author or co-author of
over 120 peer-reviewed journal and conference publications. He is a Senior Member of IEEE.

REFERENCES
[1] Y. Sun, H. Song, A. J. Jara and R. Bie, “Internet of Things and Big Data Analytics for Smart and Connected Communities”, IEEE Access,
vol(4), pp. 766-773, 2016.
[2] D. Tracey, “A Holistic Architecture for the Internet of Things, Sensing Services and Big Data”, in Proceeding of the 13rd IEEE/ACM
International Symposium on Cluster, Cloud and Grid Computing, pp. 546-553, 2013.
[3] T. Zhu, S. Xiao, Q. Zhang, Y. Gu, P. Yi and Y. Li, “Emergent Technologies in Big Data Sensing: A Survey”, International Journal of
Distributed Sensor Networks, 2015(8), 2015.
[4] S. P. Mohanty, U. Choppali, and E. Kougianos, “Everything You wanted to Know about Smart Cities”, IEEE Consumer Electronics
Magazine, Volume 6, Issue 3, July 2016, pp. 60-70.
[5] IEC White Paper for “Internet of Things: Wireless Sensor Networks” http://www.iec.ch/whitepaper/pdf/iecWP-internetofthings-LR-
en.pdf, accessed on March, 01, 2017.
[6] The Continuum: Big Data, Cloud & Internet of Things, https://www.ibm.com/blogs/internet-of-things/big-data-cloud-iot/, accessed on
March, 01, 2017.
[7] Editor Summary, “Big data: science in the petabyte era: Community cleverness Required”, Nature 455 (7209):1, 2008.
[8] S. Tsuchiya, Y. Sakamoto, Y. Tsuchimoto, V. Lee, “Big Data Processing in Cloud Environments,” FUJITSU Science and Technology
Journal, 48(2), pp. 159-168, 2012.
[9] M. Armbrust, A. Fox, R. Griffith, A.D. Joseph, R. Katz, A. Konwinski, G. Lee, D. Patterson, A. Rabkin, I. Stoica, M. Zaharia, “A view
of cloud computing,” Communications of the ACM, 53(4) (2010), 50-58.
[10] M. Yuriyama and T. Kushida, “Sensor Cloud Infrastructure,” in Proceedings of the 13th International Conference on Netowrk-Based
Information Systems, pp.1-8, 2010.
[11] D. Puthal, S. Nepal, R. Ranjan, and J. Chen, “Threats to Networking Cloud and Edge Datacenters in the Internet of Things,” IEEE Cloud
Computing, vol. 3, no. 3, pp. 64-71, 2016.

7
[12] Big data and the Internet of Things: Two sides of the same coin? https://www.sas.com/en_au/insights/articles/big-data/big-data-and-iot-
two-sides-of-the-same-coin.html, accessed on March, 01, 2017.
[13] M. L. Rajaram, E. Kougianos, S. P. Mohanty, and U. Choppali, “Wireless Sensor Network Simulation Frameworks: A Tutorial Review”,
IEEE Consumer Electronics Magazine, Volume 6, Issue 2, April 2016, pp. 63-69.
[14] S. Mukhopadhyay, D. Panigrahi and S. Dey, “Data aware, low cost error correction for wireless sensor networks,” in Proceedings of the
IEEE Wireless Communications and Networking Conference, pp.2494-2497, 2004.
[15] S. P. Mohanty, "iVAMS: A Paradigm Shift System Simulation Framework for the IoT Era", Keynote Presentation, 17th IEEE
International Conference on Thermal, Mechanical and Multi-Physics Simulation and Experiments in Microelectronics and Microsystems,
2016.
[16] X. L. Dong and D. Srivastava, “Big data integration,” in Proceedings of the 29th IEEE International Conference on Data Engineering,
pp. 1245-1248, 2013.
[17] D. Puthal, S. Nepal, R. Ranjan, and J. Chen, “DLSeF: A Dynamic Key-Length-Based Efficient Real-Time Security Verification Model
for Big Data Stream,” ACM Transactions on Embedded Computing Systems, vol 16, no. 2, PP. 51, 2017.
[18] B. Li, E. Mazur, Y. Diao, A. McGregor, P. Shenoy, “A platform for scalable one-pass analytics using mapreduce,” in Proceedings of the
ACM SIGMOD International Conference on Management of Data, 2011, pp. 985-996.
[19] R. Kienzler, R. Bruggmann, A. Ranganathan, N. Tatbul, “Stream as you go: The case for incremental data access and processing in the
cloud,” IEEE ICDE International Workshop on Data Management in the Cloud, 2012.
[20] K. Shim, “MapReduce Algorithms for Big Data Analysis,” in Proceedings of the VLDB Endowment, 5(12), pp. 2016-2017, 2012.
[21] Spark, http://spark.apache.org/, accessed on February 28, 2017.
[22] S. Mukhopadhyay, D. Panigrahi, and S. Dey, “Model Based Error Correction for Wireless Sensor Networks,” IEEE Transaction on
Mobile Computing, vol. 8(4), pp. 528-543, September, 2008.
[23] S. Slijepcevic, S. Megerian, and M. Potkonjak, “Charaterization of Lacation Error in Wireless Sensor Networks: Analysis and
Application,” in Proceeding of the 2nd International Conference on Information Processing in Sensor Networks, pp. 593-608, 2003.
[24] M. L. Rajaram, E. Kougianos, S. P. Mohanty, and P. Sundaravadivel, “A Wireless Sensor Network Simulation Framework for Structural
Health Monitoring in Smart Cities”, in Proceedings of the 6th IEEE International Conference on Consumer Electronics - Berlin, 2016,
pp. 78-82.
[25] D. J. Wang, X. Shi, D. A. Mcfarland, and J. Leskovec, “Measurement Error in Network Data: A re-classification,” Social Networks, vol.
34(4), pp. 396-409, October, 2012.
[26] K. Ni, N. Ramanathan, M. N. H. Chehade, L. Balzano, S. Nair, S. Zahedi, G. Pottie, M. Hansen, M. Srivastava, and E. Kohler, “Sensor
Network Data Fault Types,” ACM Transactions on Sensor Networks, vol. 5(3), May, 2009.
[27] C. Yang and J. Chen, “A Scalable Data Chunk Similarity based Compression Approach for Efficient Big Sensing Data Processing on
Cloud,” IEEE Transactions on Knowledge and Data Engineering, 2016, DOI: 10.1109/TKDE.2016.2531684.
[28] C. Yang, C. Liu, X. Zhang, S. Nepal and J. Chen, “A Time Efficient Approach for Detecting Errors in Big Sensor Data on Cloud,” IEEE
Transactions on Parallel and Distributed Systems, vol. 26, no. 2, pp. 329-339, 2015.
[29] C. Yang, X. Zhang, C. Liu, J. Pei, K. Ramamohanarao and J. Chen, “A Spatiotemporal Compression based Approach for Efficient Big
Data Processing on Cloud,” Journal of Computer and System Sciences, vol. 80, no. 8, pp.1563-1583, 2014.
[30] N. Laptev, K. Zeng and C. Zaniolo, “Very fast estimation for result and accuracy of big data analytics: The EARL system,” in Proceedings
of the 29th IEEE International Conference on Data Engineering, pp. 1296-1299, 2013.
[31] S. Sharma et al., “Rendezvous based Routing Protocol for Wireless Sensor Networks with Mobile Sink,” The Journal of
Supercomputing, vol. 73, no. 3, pp.1168-1188, 2017.