Description of the deliverable content and purpose
The year 2 work in WP4 of the e-SENSE project is split into three tasks, each with two deliverables, intermediate - concept and taxonomy - and final. While this task structure has provided a useful classification of the areas of work, it does not "focus on performance and functional implications resulting from the interactions and dependencies of the various functional components and their impact on the architecture" (first year reviewers’ recommendation 9).
To implement the recommendation, an integrated deliverable approach has been adopted, including all algorithmic contributions which would have been found in the 3 planned final deliverables but with specific focus on their interaction and integration into the overall architecture. This document presents the overall work performed in WP4 during the second year with this focus in mind.
It is a well-known fact that the costs associated with transportation of sensory data from the sensor network right up to the application are high and the communication between nodes introduce a large overhead. For wireless sensor networks, which in general suffer from limited resources and strive to save as much energy as possible, processing the data as close as possible to the point of action and within the network is a must. Localising and distributing the data processing and decision making task can tremendously reduce the communication overhead. Therefore, the overall objective of WP4 is to develop middleware support mechanisms to provide scalable and resilient solutions for distributed data processing and service composition/decomposition and service execution for heterogeneous wireless sensor network applications.
The main function of the e-SENSE middleware is to decouple high level context information queries from underlying WSNs and manage the services that answer a service request as well as to offer a set of support services to applications and system processes through the middleware and management subsystems. This deliverable reports on technical specifications, realisation and performance evaluation of techniques developed in the course of the e-SENSE project in the path towards achieving these objectives.
To this end, in 0we first establish the link between the distributed middleware and the overall e-SENSE architecture and will revisit the scenarios defined at an early stage of the project.
Based on these scenarios, the requirements of the distributed middleware are derived to be used as design decision parameters while developing services offered by the middleware.
Two of the requirements identified for e-SENSE middleware are the ability to (i) autonomously deploy and maintain service requests from the application and (ii) discover services offered by the e-SENSE system. Section 3 - addresses mechanisms developed to support and fulfil these requirements. These mechanisms, which we collectively call service management mechanisms, are in fact needed to compose, decompose, select, and execute services in the network from the application and to allocate tasks/services to be performed by nodes in the most optimal and resource aware manner.
The e-SENSE system is designed to be modular and reconfigurable so that many different WSN applications can be deployed using the same architecture but specific combinations of components (i.e. protocol stacks). How e-SENSE middleware provides this modularity and re-configurability is addressed in Section 4 - The remaining main sections will address various services offered by the middleware.
Section 5 - will focus on localisation techniques and will address both ranging and full 3D localisation algorithms. Additionally, a hardware acceleration engine has been developed to enhance real time tracking of resource constrained sensor nodes. Two ranging approaches are considered: first RSSI, as a ranging technique from a practical point of view where an existing RSS-based hardware localization implementation is used to provide measurements and simultaneously functions as a benchmark for comparison.
Two localisation algorithms are subsequently addressed, one appropriate to static networks, based on mass spring modelling and providing enhancements to account for the varied quality of ranging techniques, and the other for real time tracking of mobile nodes. Since no single localisation technique may be identified as the solution to all localisation applications, in this section we also provide an integrated environment to mix and match the components to suit the needs of the particular application or deployment, called “localisation awareness engine”.
The middleware also provides support services to support the operation of WSNs. Timing and synchronisation service described in Section 6 - is an example of such services. Use of high degree of spatial correlation that exists between the sensor readings of adjacent nodes in a densely deployed wireless sensor network is exploited in Section 7 - The performance of the distributed and self-organising scheduling algorithm presented in this section is evaluated in terms of energy savings compared to collecting raw data as well as network stabilisation times and message transmissions.
Section 8 - addresses low level and high level context awareness. The lower level determines simple context directly from sensory data. Algorithms determining this kind of information are typically based on very specific types of sensor data streams, which are analysed for characteristic structures that identify the context that is searched for. Work in high level context awareness aims at inferring more complex context descriptions from the low level context information.
This e-SENSE Deliverable is public and available to everybody without charge.
For internal statistics purpose we kindly ask you some personal information. You will be redirected to a webpage available for 24 hours from which you can download the document in PDF format.
(*) Mandatory fields
