International Journal of Research in Computer Science eISSN 2249-8265 Volume 3 Issue 1 (2013) pp. 19-25 www.ijorcs.org, A Unit of White Globe Publications doi: 10.7815/ijorcs. 31.2013.057

www.ijorcs.org

FROM PHYSICAL TO VIRTUAL WIRELESS SENSOR NETWORKS USING CLOUD COMPUTING

Maki Matandiko Rutakemwa PhD Student, Computer Science, Christ University Bangalore, INDIA

Email: makinsoft@gmail.com, makimatandiko.rutakemwa@christuniversity.in

Abstract: In the modern world, billions of physical sensors are used for various dedications: Environment Monitoring, Healthcare, Education, Defense, Manufacturing, Smart Home, Agriculture Precision and others. Nonetheless, they are frequently utilized by their own applications and thereby snubbing the significant possibilities of sharing the resources in order to ensure the availability and performance of physical sensors. This paper assumes that the immense power of the Cloud can only be fully exploited if it is impeccably integrated into our physical lives. The principal merit of this work is a novel architecture where users can share several types of physical sensors easily and consequently many new services can be provided via a virtualized structure that allows allocation of sensor resources to different users and applications under flexible usage scenarios within which users can easily collect, access, process, visualize, archive, share and search large amounts of sensor data from different applications. Moreover, an implementation has been achieved using Arduino-Atmega328 as hardware platform and Eucalyptus/Open Stack with Orchestra-Juju for Private Sensor Cloud. Then this private Cloud has been connected to some famous public clouds such as Amazon EC2, ThingSpeak, SensorCloud and Pachube. The testing was successful at 80%. The recommendation for future work would be to improve the effectiveness of virtual sensors by applying optimization techniques and other methods.

Keywords: Virtual Wireless Sensor Networks, Ubiquitous Computing, Cloud Computing

I. INTRODUCTION

Wireless sensor networks have been traditionally designed to be privately owned and used. Therefore, an increasing number of sensor networks have been deployed to monitor a variety of conditions and situations. At the same time, more and more applications are starting to rely on the data from sensor networks to provide users with (near) real-time information and conditions. This increasing demand of users for accurate information about natural and surrounding phenomena is creating a business case for application providers.

But the two hallmark features of sensor networks, namely customized network applications and the collaborative in-network processing, are not achievable beyond the boundary of the users' administrative domains (except for limited scope data sharing through Internet gateways).

We argue that even quite mature networking and duty cycling protocols for sensor networks exists, but approaches to wireless sensor network sharing and management are still immature.

In particular, typical wireless sensor networks are designed and deployed to serve a single application. Indeed, the common approach in the design of sensor networks is to deploy networks that are fit-for-purpose with the primary aim of supporting a single application that belongs to a single authority (usually the owner of the infrastructure)[13]. While this is a sensible approach for short-term and small-scale deployments, in sensor network deployments that consist of thousands of nodes with a life span of multiple years, inducing high costs of deployment and maintenance, the single application approach can lead to inefficient use of resources and low cost-benefit results. Moreover, the requirement for dedicated sensing infrastructure to support new applications belonging to different organizations can lead to unnecessary replication of sensing infrastructure. One example that illustrates this problem is the deployment of temperature sensors on a single environment by different authorities (Government, University, Social Agency, Hospital and others.

In this work we propose a departure from the notion of sensor networks aimed at supporting a single application and serving a single user. We introduce a tactic that is based on the separation of substructure and application ownership.

The primary objective of this work is to create a framework that allows sensor network infrastructures to be shared among multiple applications that can potentially belong to different authorities. By achieving this level of partaking, sensing infrastructures can be viewed as an accessible resource that can be dynamically re-purposed and re-

20 Maki Matandiko Rutakemwa

www.ijorcs.org

programmed by different authorities, in order to support multiple applications.

The main task in apprehending this journey is the design of a novel wireless sensor network architecture that supports multiple applications, dynamically uploaded by different owners and simultaneously running over a shared infrastructure.

In this work we illustrate our efforts in exploring this vision. The key characteristics of our approach are: i. A virtualization layer that is running on each sensor

node abstracts the access to sensor resources and allows the management of these resources through policies expressed by the infrastructure owner.

ii. A runtime environment on each node that allows multiple applications to run inside each node.

iii. A policy based application deployment that enables multiple applications to be deployed over the shared infrastructure.

The following sections offer an overview of the architecture and a description of our implementation.

We have discussed the related work in Section 2. We present an overview of our proposed architecture in Section 3, the details of our implementation are in Section 4 and conclude while projecting future research opportunities in Section.

II. RELATED WORKS

There have been a few of studies on the virtualization of physical wireless sensor networks. Naturally, the design we propose here shares some of the design goals that are common to other research projects, while addressing unique problems that are complementary in nature.

The goal of the urban participatory sensing project at the Center for Embedded Networked Sensing (CENS) is to engage sensors built into portable devices such as mobile phones and PDAs in sharing their capabilities [10]. A tiered architecture and related protocols for people-centric urban sensing is under development as part of the MetroSense project at Dartmouth, which involves mobile devices in opportunistic interactions with wireless Internet gateways and available static sensornet infrastructure [5].

Mires [6] is a publish/subscribe architecture for WSNs. Basically sensors only publish readings if the user has subscribed to the specific sensor reading. Messages can be aggregated in cluster heads. Subscriptions are issued from the sink node (typically directly connected to a PC), which then receives all publications.

TinySIP [7] supports session semantics, publish/subscribe, and instant messaging. It offers

support for multiple gateways. Most communication is done by addressing individual devices. As device addresses are related to the gateway being used, changing the gateway on the fly is difficult.Users need to know the specifications of different kinds of physical sensors.

OGC (Open Geospatial Consortium) [9] defined Sensor Modeling Language (SensorML) [1] to provide standard models and an XML encoding for physical sensors’ description and measurement processes. SensorML can represent the metadata for any physical sensor (such as the type of physical sensor, the location, and the accuracy). We used SensorML to describe the metadata of physical sensors. We have added the mapping between physical sensors and virtual sensors to describe how to translate commands coming from users to virtual sensors into commands for the corresponding physical sensors.

Although there are many kinds of physical sensors, no application uses all of them. Each application needs sufficient physical sensors for its requirements (such as physical sensors in a certain location). A publish/subscribe mechanism [11] is used to select physical sensors in [8]. When there are multiple sensor networks, each sensor network publishes sensor data and metadata that describes the type of physical sensors. Each application subscribes to one or more sensor networks to receive a real-time data stream from their physical sensors. Such publish/subscribe mechanism allows each application to select only a particular type of physical sensors it collects data from. Sensor-Cloud infrastructure makes virtual sensors from multiple physical sensors. Because every virtual sensor is not created from a sensor network, the grouping is more flexible. Users can select groups of virtual sensors or virtual sensors.

Users should check whether the physical sensors are available and detect physical sensors’ faults for keeping the quality of the data coming from physical sensors. FIND [13] provided a novel method to detect physical sensors with data faults. FIND ranks the physical sensors based on their sensing readings as well as their physical distances from an event. FIND considers a physical sensor(s) faulty if there is a significant mismatch between the sensor data rank and the distance rank. This approach focuses on detecting physical sensor(s) faults, while we focus on monitoring the virtual sensors. Because there is a relationship between the status of a virtual sensor and the status of its sensors, the virtual sensor will also report incorrect results if the linked physical sensors are faulty. The users of the cloud computing service check the status of their virtual servers, not the status of the linked physical server. We also focus on monitoring the status of virtual sensors.

From Physical to Virtual Wireless Sensor Networks using Cloud Computing 21

www.ijorcs.org

III. VIRTUAL WIRELESS SENSOR NETWORK INFRASTRUCTURE

Let us now explore how one can temporarily “borrow” several nodes from several separate domains, virtual sensors that extend the area of coverage beyond some physical boundary.

A. Assumptions

Our motivation is to explore the possibilities of forming scalable virtual wireless sensor network under the following design assumptions:

− A group of users (private, corporate and/or Government) is willing to deploy sensor node hardware to cover areas under their power.

− There exists a level of willingness from all members of the group to share the sensing and processing capabilities of their physical sensors’ infrastructure as needed, without interruption.

− A typical user belonging to the group often borrows sensor node capabilities from parts of other sensor network infrastructure to form a virtual sensor network over a desired area of coverage.

− The rules of sharing are dictated by the owner of the resource and at no time the sensing task of a local or remote user violates the authority, privacy and security of the provider.

− Optionally, a supporting business model accounts for balancing the cost of sharing resources (such as hardware and battery costs) between private and community use.

When users request virtual sensors, the Cloud infrastructure automatically provisions them from their templates. Users can control their virtual sensors directly or via their Web browsers. Cloud infrastructure also provides the users with monitoring functions for the virtual sensors.

B. Steps

The steps to be achieved in the process are:

Virtualization: There are various kinds of scattered physical sensors. Virtual sensor enables the use of sensors without worrying about the locations and the specifications of physical sensors. Fig.2 describes the relationships between virtual sensors and physical sensors. Each virtual sensor is created from one or more physical sensors. Users can create virtual sensors and freely use them as if they owned sensors. For example, they can activate or inactivate their virtual sensors, check their status, and set the frequency of data collection from them. If multiple users freely control the physical sensors, some inconsistent commands may be issued. The users can freely control their own virtual sensors by virtualizing the physical sensors as virtual sensors.

Figure 1: Virtual vs Physical Sensors

Standardization: Different kinds of physical sensors have different specifications. Each physical sensor provides its own functions for control and data collection. Standard mechanism enables users to access sensors without concern for the differences among the physical sensors. Standard functions for virtual sensors, are used so that the users can access the virtual sensors with the standardized functions. Cloud infrastructure translates the standard functions for the virtual sensors into specific functions for the different kinds of physical sensors.

This virtualization implies the following layers of the architecture :

− A virtualization layer that is running on each sensor node, abstracts access to sensor resources and allows the management of these resources through policies expressed by the infrastructure owner.

− A runtime environment on each node that allows multiple applications to run inside each node.

− A policy based application deployment that enables multiple application to be deployed over the shared infrastructure.

Figure 2: Application – Architecture

Figure 3: Mapping Physical - Virtual Sensors

22 Maki Matandiko Rutakemwa

www.ijorcs.org

C. Major Actors

Sensor Owner: A sensor owner is an actor who owns his physical sensors. A sensor owner allows others to use those physical sensors through Cloud infrastructure. A sensor owner registers the physical sensors with their properties to the Cloud. The owner deletes the registration of them when s/he quits sharing them.

Cloud Administrator: is the actor who manages the Cloud Infrastructure service. The administrator manages the IT resources for the virtual sensors, monitoring, and the user interfaces.

Figure 4: Federation of Clouds - Cloud Administrator

End User: An end user is an actor with one or more applications or services that use the sensor data. An end user requests the use of virtual sensor that satisfies the requirements from the templates. The user can control her/his virtual sensors directly or via a Web browser. The user can monitor the status of the virtual sensors. When they become unnecessary, the user can release them. The end users can use the virtual sensors by paying for usage and with no detailed knowledge about the physical sensors.

IV. IMPLEMENTATION AND DEVELOPMENT

− Using Arduino Uno Atmega 328 and Arduino Shield along with an LDR sensor and a LM35 sensor, a prototype of the architecture has been implemented and tested as shown in Figure 6.

− Firstly Ubuntu 11.10 has been installed on 2 systems to constitute a private cloud. To this cloud the two physical sensors have been connected using the Arduino Shield, Figure 7.

− Next, the private cloud based mainly on Eucalyptus structure was connected to several public clouds such as Pachube, ThingSpeak, Amazon EC2, SensorCloud and Github shown in Figure 8.

Figure 5: Arduino Uno Atmega 328, Arduino Ethernet

Shield, LDR Sensor, LM35 Sensor

Figure 6: Arduino Uno, Arduino Ethernet Shield and

sensors connected to the System running Ubuntu 11.10

Figure 7: Ubuntu 11.10 with Eucalyptus connected to the

hardware for Private Cloud with virtual sensors

Figure 8: Connection to Pachube from Private Cloud

From Physical to Virtual Wireless Sensor Networks using Cloud Computing 23

www.ijorcs.org

Figure 9: Temperature and Light Graphs on Pachube after sending readings from physical sensors

Figure 10: Connection to ThingSpeak Public Cloud with LDR readings graph generation

Figure 11: Connection to SensorCloud Public Cloud

24 Maki Matandiko Rutakemwa

www.ijorcs.org

Figure 12: Temperature readings sent to ThingSpeak Public Cloud

V. CONCLUSION

We present Cloud infrastructure which virtualizes physical sensors so that end users can share them with no concerns about the details of them (i.e. location and specification). This infrastructure enables end users to create virtual sensors dynamically by selecting the templates of virtual sensors with IT resources.

Within this novel architecture users can share several types of physical sensors easily and consequently many new services can be provided via a virtualized structure which allows allocation of sensor resources to different users and applications under flexible usage scenarios within which users can easily collect, access, process, visualize, archive, share and search large amounts of sensor data from different applications.

Moreover, an implementation has been achieved using Arduino-Atmega328 as hardware platform and Eucalyptus/Open Stack with Orchestra-Juju for Private Sensor Cloud. Then this private Cloud has been connected to some famous public clouds such as Amazon EC2, ThingSpeak, SensorCloud and Pachube. The testing was successful at 80%.

The recommendation for future work would be to improve the effectiveness of virtual sensors by applying optimization techniques and other methods.

VI. REFERENCES

[1] C. Lenzen, P. Sommer, R. Wattenhofer, “Optimal Clock Synchronization in Networks,” The 7th ACM Conference on Embedded Networked Sensor Systems (SenSys 2009), 2009. doi: 10.1145/1644038.1644061

[2] J. Hill, R. Szewczyk, A. Woo, S. Hollar, D. Culler, K. Pister, “System architecture directions for networked sensors,” International Conference on Architectural Support for Programming Languages and Operating Systems, 2000. doi: 10.1145/356989.356998

[3] J. Koo, R. K. Panta, S. Bagchi, L. Montestruque, “A Tale of Two Synchronizing Clocks,” The 7th ACM Conference on Embedded Networked Sensor Systems (SenSys 2009). doi: 10.1145/1644038. 1644062

[4] J. Scott Miller, Peter A. Dinda, Robert P. Dick, “Evaluating A BASIC Approach To Sensor Network Node Programming,” The 7th ACM Conference on Embedded Networked Sensor Systems (SenSys 2009). doi: 10.1145/1644038. 1644054

[5] J. Shneidman, P. Pietzuch, J. Ledlie, M. Roussopoulos, M. Seltzer, M. Welsh, "Hourglass: An Infrastructure for Connecting Sensor Networks and Applications," Harvard Technical Report TR-21-04, 2004.

[6] Kevin Klues, Chieh-Jan Mike Liang, Jeongyeup Paek, Razvan Musaloiu-E, Philip Levis, Andreas Terzis, Ramesh Govindan, “TOSThreads: Thread-Safe and Non-Invasive Preemption in TinyOS,” The 7th ACM Conference on Embedded Networked Sensor Systems (SenSys 2009). doi: 10.1145/1644038.1644052

[7] Keiji Matsumoto, Ryo Katsuma, Naoki Shibata, Keiichi Yasumoto, Minoru Ito, “Extended Abstract: Minimizing Localization Cost with Mobile Anchor in Underwater Sensor Networks,” The Fourth ACM International Workshop on UnderWater Networks (WUWNet), 2009. doi: 10.1145/1654130.1654144

[8] M. Gaynor, M. Welsh, S. Moulton, A. Rowan, E. LaCombe, and J. Wynne, “Integrating Wireless Sensor Networks with the Grid,” IEEE Internet Computing, Special Issue on Wireless Grids, 2004. doi: 10.1109/MIC.2004.18

From Physical to Virtual Wireless Sensor Networks using Cloud Computing 25

www.ijorcs.org

[9] Open Geospatial Consortium. http://www.opengeospatial.org/

[10] Shuo Guo, Ziguo Zhong, Tian He, “FIND: Faulty Node Detection for Wireless Sensor Networks,” Proc. The 7th ACM Conference on Embedded Networked Sensor Systems (SenSys 2009), pp. 253-266. doi: 10.1145 /1644038.1644064

[11] SensorML. http://vast.uah.edu/SensorML/ [12] T. I. Sookoor, T. W. Hnat, P. Hooimeijer, W. Weimer

and K. Whitehouse, “Macrodebugging: Global Views of Distributed Program Execution,” The 7th ACM Conference on Embedded Networked Sensor Systems (SenSys 2009), 2009.

[13] The Jython Project. http://www.jython.org/ [14] Ziguo Zhong, Tian He, “Achieving Range-Free

Localization Beyond Connectivity,” The 7th ACM Conference on Embedded Networked Sensor Systems (SenSys 2009). doi: 10.1145/1644038.1644066

How to cite

Maki Matandiko Rutakemwa, "From Physical to Virtual Wireless Sensor Networks using Cloud Computing". International Journal of Research in Computer Science, 3 (1): pp. 19-25, January 2013. doi: 10.7815/ijorcs. 31.2013.057