Sharing Worldwide Sensor Network Lei Shu1*, Manfred Hauswirth1, Long Cheng2, Jian Ma3, Vinny Reynolds1, Lin Zhang1 1 Digital Enterprise Research Institute, National University of Ireland, Galway 1 {lei.shu, manfred.hauswirth, vinny.reynolds, lin.zhang}@deri.org 2 Beijing University of Posts & Telecommunications, email: [email protected] 3 Nokia Research Center, Beijing, email: [email protected]

Abstract We can envision that future sensor networks (WSNs) are ubiquitous, large-scale, interconnected, which we call worldwide sensor network (WWSN). Currently, most of the WSNs are working as isolated islands. Without sharing sensor data across different domains, the most important features of ubiquitous computing (e.g. context awareness) will not be easily achieved. For sharing among WWSN, the first nut to crack is to interconnect different WSNs which are spatially deployed in different locations with IP based Internet; the second one is to integrate them into a single WWSN over the Internet for publishing, sharing and searching of sensor data. In this paper, we present the challenge issues that should be addressed for sharing WWSN, by conducting a short survey on existing approaches.

1. Introduction Currently, many wireless sensor networks (WSNs) are working as isolated islands, and most of the sensed valuable data is not shared yet. Without sharing sensor data across different domains, the vision of ubiquitous computing will not easily achieve. To satisfy this key requirement, a common means to share distributed sensor data over the Internet is necessary, which requires a large scale underlying infrastructure to be designed for: 1) Easy publishing of sensor data online; 2) Easy searching of sensor data online. Researching on these two goals requires an understanding on: 1) how to connect different WSNs which are spatially deployed in different locations with the Internet and 2) how to integrate them into a single WWSN for the publishing, sharing and searching of sensor data. A worldwide sensor network (WWSN) which does not include private sensitive information will provide public services to all interested users, e.g. Fig.1. This

Figure 1. Worldwide sensor network global observation system opens up a new avenue to effective use of both local and remote sensor data, accurate analysis and decision makings [1]. It can perform as an extensive sensing and monitoring system that provides timely, continuous and multi-dimensional observations. A good example is an imaging service for tourist sites where cameras are installed. These cameras can form a WSN and provide some pictures or video streams, and allow remote users to query their interested information. However, the heterogeneity of WSNs turns the efficient collection and analysis of the various sensor data into a rather challenging issue. The main reasons for that are the lacks of: 1) uniform interface and 2) a standard representation of sensor data, which motivate that a large scale open software platform is needed for sharing heterogeneous sensor data. The purpose of this paper is to identify the challenging issues associated with sensor data sharing in the WWSN.

2. Survey on interconnection issue Different research works have been conducted for connecting WSNs with Internet: Using IP protocol for direct interconnection, which assumes that sensor nodes are powerful enough to run a u-IP protocol [2]: by running u-IP protocol in WSNs, Internet users can access any sensor node by using IP address. But, the assumption that sensor nodes always have enough resource to support the overhead that brought by IP protocol stack is not feasible. Using only IP protocol is not suitable for application-specific

WSNs, since different WSNs may have different requirements from applications. Using an overlay for indirect interconnection, the WSNs overlay IP is proposed in [3]. WSNs protocol stack is deployed over the IP layer and each Internet host is considered as a virtual sensor node. Any Internet host can directly communicate with sensor node and it processes packets exactly as sensor nodes do. However, it has to deploy an additional protocol stack into the Internet host, which brings more protocol header overhead to IP based networks. Using a bridge for indirect interconnection: In [4] authors proposed VIP Bridge to connect heterogeneous WSNs with IP based networks. Packets that come from one side will be translated into corresponding packet formats and sent to another side by this VIP Bridge. In this research different translators are used for translating different packet formats, e.g. ZigBee and Bluetooth, into IPv6 protocol packet format. Using gateway for indirect interconnection: The most common approach for connecting sensor networks with an external network is using application-level gateway, e.g. GSN [5]. Different protocols in both side networks are translated in application layer. In GSN, authors use wrappers to translate different protocols’ packet formats into unified data format. The main role of gateway is to relay packets to different networks. The advantage is: the communication protocol used in different sensor networks may be chosen freely.

server analyzes the query and finds the appropriate sensor network, and then produces a response. One main advantage of the server-client approach is its Table 1. Comparison of four typical approaches

central index which can locates sensing resources quickly and efficiently. SenseWeb [6] focuses on large scale WSNs and ease of sharing sensor data. SenseWeb requires all sensing data to register in a Web Server. GSI [7] exploits a wide range of sensing data with focus on providing a unified framework that spans both WSNs and IP networks. SensorMap [8] mainly focuses on spatial data and builds a portal that shows real-time sensor data on a map. SensorPlanet [9] is a central repository for sharing sensor data collected from large scale WSNs. SensorBase [10] develops a centralized data storage and management system that provides a uniform and consistent method for publishing and sharing data. IrisNet [11] envisions a worldwide sensor web in which users can query vast quantities of data from thousands or even millions of widely distributed heterogeneous sensors. It is noteworthy that IrisNet uses distributed databases to store sensor data to overcome the high updating rates.

3. Discussion on interconnection issue These four typical approaches for interconnection issue are compared in Table 1 in terms of two aspects: 1) freely choose routing protocol in WSNs; 2) easy to integrate heterogeneous WSNs. It is not difficult to see that it is easier to integrate different heterogeneous WSNs by adopting wrappers/translators.

4. Survey on integration issue Some research works have been invested on integrating heterogeneous WSNs into an open shared system, which can be categorized into two classes: 1) Server-Client Approach, 2) Peer to Peer Approach. Server-Client Approach: employing a central system which requires data owners to register their data sources to one central server [6-11], e.g. Fig.2. These sensing resources are updated at intervals to let the server know the availability. When an application submits a query to search for a service, the central

Figure 2. Centralized system architecture

Figure 3. P2P system architecture

Peer-to-Peer Approach: adopting P2P techniques, each WSN with a gateway act as a peer, e.g. Fig.3. The main goal of P2P overlay is to treat the underlying heterogeneous WSNs as a single unified network, in which users can send queries without considering the details of the network, e.g. topology and architecture. User peers communicate with gateway peers in a P2P approach. Typical representatives are [5, 12-15]. GSN [5] is a P2P platform which provides a scalable infrastructure for integrating heterogeneous WSNs using a small set of powerful abstractions. In [12], the authors propose a generic communication structure to discover and access widely distributed wireless sensor nodes from various types of devices by

using P2P technology. In [13], an architecture for P2P mobile WSNs is proposed, which allows P2P WSNs overlay on 3G mobile networks to connect with each other. Sharing sensor networks [14] places more emphasizes on sharing services provided by existing WSNs in the area of distributed manufacturing by using P2P technology. ShareSense [15] attempts to provide a general, extensible and easy to handle overlay software platform for WSNs.

5. Discussion on integration issue The server-client approach allows the most efficient services discovery but with the risk of a single point of failure. The network can collapse if the central server is incapacitated. The central server may also provide out of date information, as it is only refreshed periodically. Another drawback is that the server-client approach is not suitable for sharing real time live streaming data that gathered from WSNs, e.g. multimedia streams. Concerning peers participating in a P2P WWSN, each WSN with a gateway is a peer and each user also acts as a peer. All participating peers are equal. For individual sensing data contributors who would like to make their sensing services available on the Internet, central server places greater demands on time and effort than that of P2P WWSN, in which contributors simply configure their sensing services in a standard way designated to be publicly available for sharing.

6. Challenging issues Sharing WWSN along with its benefits brings out many new challenging issues. In this section, we try to articulate these challenges. z Heterogeneity Components of a WWSN are highly heterogeneous along two dimensions. In the sensor node dimension, the types of sensors (small embedded devices, mobile phones, wireless cameras), and the capabilities for processing data are different. In the sensor network dimension, the number of nodes and the rate for producing data are different. These heterogeneities demand a standardized abstraction to achieve a better understanding of different sensor information and a uniform solution for gathering different sensor data. z Scalability A WWSN becomes more useful as the number of participant WSNs grows. Aiming at a very large number of data producers and consumers with a variety of application requirements, this introduces a significant challenge for scalability. To keep the

resource usage scalable and to avoid unnecessarily denying access to more applications, it’s essential for the system to reuse common data and sensing resources among overlapping application needs. This scalability challenge is further compounded by the need for maintaining extensibility to unknown applications and domain-specific data processing & aggregation. z Publishing and discovering sensor resources If publishing even a single stream of sensing data requires too much effort, this will discourage data owners to contribute sensing data. So, the effort requested for publishing sensing data is a significant issue, which has direct impact on participant numbers. Not only is the publishing very important, but also the discovering sensor data is a significant issue. The shared WWSN should have the ability to discover and invoke services from heterogeneous WSNs and fuse related data to cater for different applications. z Query aggregation There will be a large number of applications over this WWSN, although different applications may seek different sensor data, a subset of data collection tasks is common to some services. Wasting computation and limited energy resource should be avoided if two sensing applications overlap. An arbitrary number of user queries can be merged into a single query, which can avoid the overlapping queries on the same data source and decrease the processing load in the gateway. Therefore, how to aggregate a large number of queries, optimize the processing load in the gateway and efficiently share sensing streams among multiple applications is a big challenge issue. z Interconnection To achieve sensing resources sharing in the WWSN, an appropriate interconnection approach must be introduced to answer: how to interconnect different sensor networks which are spatially deployed in different locations with IP based Internet; how will the designed interconnection solution support the integration issues in both network layer and data layer. The design of this interconnection solution will have strong impact on all the solutions for other issues, which actually requests the designers to take the crosslayer interaction into consideration. z Integration Integration follows the challenge of interconnection. Sensor networks which are physically located in different locations may use totally different routing protocols for their specific applications. How to integrate these heterogeneous WSNs into one shared system over Internet to provide comprehensive services for users is the critical issue after connecting sensor networks to IP based Internet.

z

Data collection and data storage The system employs numerous globally distributed sensors that observe the physical world. Because the sensors would collect a vast volume of data, and because we would want to retain both the most recent observations and a historical record, the system should collect observations and store them. How to efficiently collect and store global sensor data is a critical issue. z APIs for high-level applications As the WSN participant number grows, a variety of new kinds of applications over existing data networks can be deployed by third party. Appropriated open APIs should be provided to ease the development of these kinds of applications. This programming paradigm can conversely incentive more data producers and consumers to participate.

[4] L. Shu, J. Cho, S. Lee, M. Hauswirth, L. Zhang, "VIP Bridge: Leading Ubiquitous Sensor Networks to the Next Generation", Journal of Internet Technology, vol.8, 2007. [5] K. Aberer, M. Hauswirth, Salehi, A., “Infrastructure for Data Processing in Large-Scale Interconnected Sensor Networks”, in Proceedings of the 8th International Conference on Mobile Data Management, Mannheim, Germany, May 7-11, 2007. [6] A. Kansal, S. Nath, J. Liu, and F. Zhao, "SenseWeb: An Infrastructure for Shared Sensing", IEEE Multimedia, 2007, vol. 14, pp. 8-13. [7] C.L. Fok, G.C. Roman, and C. Lu, "Towards A Flexible Global Sensing Infrastructure", in ACM SIGBED Review, Special issue on the workshop on wireless sensor network architecture 2007, vol. 4, pp. 1-6.

7. Conclusion Sharing data in worldwide sensor networks poses many challenging issues, which require more research effort and new solutions to be proposed. The interconnection issue should be solved by adopting the wrappers or translators approach to allow sensor networks to freely choose application-specific routing protocols, and the integration issue should be solved by adopting the P2P approach to: avoid the single point of failure, support live sensor stream and decrease the effort for publishing sensor data.

8. Acknowledgement The work presented in this paper was supported by the Lion project supported by Science Foundation Ireland under grant no. SFI/02/CE1/I131. Thank Dr. Christian Decker’s kind guidance.

9. References [1] X. Chu, R. Buyya, “Service Oriented Sensor Web”, Sensor Network and Configuration: Fundamentals, Standards, Platforms, and Applications, N. P. Mahalik (ed), pp.51-74, Springer-Verlag, Berlin, Germany, Jan. 2007. [2] A. Dunkels, J. Alonso, T. Voigt, H. Ritter, J. Schiller, "Connecting Wireless Sensornets with TCP/IP Networks", in Proceedings of the Second International Conference on Wired/Wireless Internet Communications (WWIC2004), Frankfurt(Oder), Germany, February 2004. [3] D. Hui and R. Han, "Unifying Micro Sensor Networks with the Internet via Overlay Networking", in Proceeding of the 29th Annual IEEE International Conference on Local Computer Networks, R. Han, Ed., 2004, pp. 571-572.

[8] S. Nath, J. Liu, J. Miller, Z. Feng, and A. Santanche, "SensorMap: A Web Site for Sensors World-Wide", in Proceedings of the 4th International Conference on Embedded Networked Sensor Systems, Boulder, USA, 2006. [9] Sensorplanet http://www.sensorplanet.org/. [10] K. Chang, N. Yau, M. Hansen, D. Estrin, "A Centralized Repository to Slog Sensor Network Data", in Proceedings of International Conference on Distributed computing in Sensor Systems, San Francisco, USA, June 17, 2006. [11] P. B. Gibbons, B. Karp, Y. Ke, S. Seshan, "IrisNet: an architecture for a worldwide sensor Web," in Pervasive Computing, vol. 2, pp. 22-33, 2003. [12] M. Isomura, C. Decker, M. Beigl, "Generic Communication Structure to Integrate Widely Distributed Wireless Sensor Nodes by P2P Technology," in Proceedings of the seventh International conference on Ubiquitous Computing (Ubicomp 2005), September, 2005, Tokyo, Japan. [13] S. Krco, D. Cleary, and D. Parker, "P2P Mobile Sensor Networks," in Proceedings of the 38th Hawaii International Conference on System Sciences, 2005, pp. 324-324. [14] M. Isomura, T. Riedel, C. Decker, M. Beigl, H. A. H. H. Horiuchi, "Sharing sensor networks," in Proceedings of the 26th IEEE International Conference Workshops on Distributed Computing Systems, Lisboa, Portugal, 2006. [15] A. Antoniou, I. Chatzigiannakis, A. Kinalis, G. Mylonas, S. Nikoletseas, and A. Papageorgiou, "A Peer-toPeer Environment for Monitoring Multiple Wireless Sensor Networks," in Proceedings of the 10th ACM/IEEE Symposium on Modeling, Analysis and Simulation of Wireless and Mobile Systems, Crete Island, Greece. 2007.