Towards application of service oriented architecture in wireless sensor networks

Sammanfattning: Wireless sensor networks (WSN) are a foreseeable source of innovation in many fields and the number of WSN deployments is expected to grow rapidly in the near future. One of the key challenges in designing, implementing, deploying and maintaining complex systems that use WSNs is provisioning and integration of sensor data with existing communication infrastructures. Service Oriented Architecture (SOA) is considered a promising distributed computing approach for providing a common communication platform that supports interoperability and scalability on the application layer. The work in this thesis investigates the application of SOA in WSNs in terms of different technological solutions, performance requirements and scalability. The focus is on efficient implementation and data representation techniques that are capable of meeting the requirements of resource constrained sensor nodes.To date, research efforts in embedded SOA solutions have been focused mainly on middleware software systems in gateway devices for interacting with the sensor network. The work presented here is directed towards deploying interoperable standard-based web services directly on the sensor nodes instead of using gateways and proprietary protocols. This strategy provides for manageable WSN provisioning and interoperability between heterogeneous sensor nodes. The biggest drawback in this approach is the overhead in network throughput, latency, memory, CPU and power consumption connected to the use of verbose data encoding formats such as extensible markup language (XML). In order to mitigate this overhead, in Paper A we investigate various resource aware implementation techniques deployed on a sensor node. The techniques include the use of application specific XML parser, lightweight TCP/IP stack with split-phase interface, parsing on-the-fly and sleep schedules. The results show that it is possible to use standard SOA implementations for resource constrained wireless sensors. Nevertheless, the use of this approach is limited to applications with low real-time requirements due to increased latency in the message exchange.In order to further lower the SOA overhead, the work in this thesis investigates the use of binary representation of the XML data. Based on an initial investigation, the newly emerged W3C standard - Efficient XML Interchange (EXI), provides the best compression and processing efficiency when it comes to XML structured information. For that reason, in Paper C we study how EXI can be applied to the state of the art SOA middlewares for factory automation in order to verify and analyze the benefits of using EXI in industrial environment. Based on this analysis and the experiments that are carried out, we conclude that EXI is a very beneficial technology for factory automation software that is capable of extending the reach of industrial SOA solutions to wireless sensor nodes.The work presented in this thesis also analyzes the current EXI implementations and shows that they are not designed to meet the high requirements of resource utilization in WSN deployments. In Paper B we also perform an analysis of the EXI specification and how it maps to the industrial requirements for wireless sensor nodes. The result of this work is an innovative software design of a new EXI processor targeted at highly resource constrained devices. Among other things, as a part of this EXI processor design is a proposal for a new Application Programming Interface (API) for working with XML Information Set items. Moreover, this work has developed an open source implementation of the new EXI processor design in order to verify the proposed architectural models. The benefits of the new API when it comes to efficiency are experimentally verified for a small embedded platform. The key aim in this work is the vertical integration between WSN and enterprise level systems. The achieved results show that the use of special implementation design as well as the EXI protocol for XML data representation provide large gains in efficiency in WSN deployments. The next step in this work is to investigate and compare these results with the new RESTful constrained protocols under development by the Internet Engineering Task Force (IETF). This analysis would pave the way towards reaching the final aim of creating a common, SOA-based communication architecture for cost-effective, flexible and reliable development of complex systems that use WSN technology.