This document translates the vision of the e-SENSE project, which has been derived in deliverable D.1.1.2, into an early reference model for the e-SENSE architecture. The early reference model defines elementary building blocks of an architecture that satisfies the e-SENSE architectural vision and identifies interfaces between those building blocks. It defines a minimum set of constraints and identifies research challenges in different aspects of the e-SENSE system. The early reference model further identifies a set of evaluation metrics and possible evaluation methodologies to investigate the system on a common basis.

The purpose of the early reference model is to present a general framework for all other work packages by providing a common set of assumptions and a consistent view of the overall e-SENSE system. The early reference model does not specify the overall e-SENSE system architecture or detailed requirement specification. The system architecture will be developed considering the research effort of different work packages and described in a later deliverable D.2.2.1. The early reference model should be seen as guidance for other work packages to align their contributions in the overall e-SENSE system and to ensure a basic level of compatibility between the work packages. It does not aim to dictate the form of research contributions of other work packages, nor essentially limit the scope of their contributions.

Section 2 of this document presents a high level definition of the e-SENSE network reference model and describes the logical architecture of the e-SENSE protocol stack. A sensor node is the elementary building block of a wireless sensor network, capturing the ambient intelligence surrounding users and objects in the targeted application spaces. The captured information is provided to or accessible by beyond 3G nodes, which are consuming entities of the ambient intelligence in mobile communication systems e.g. application servers or service platforms.

The e-SENSE protocol stack architecture is decomposed into four logical subsystems, providing the functionality for the architecture of the system: application, middleware, management and connectivity subsystem. Within the e-SENSE system, a sensor node can take up one or several different roles, which determines the subsystem configuration. Both, the connectivity subsystem as well as the middleware subsystem, are envisioned as highly modular toolboxes, catering for cross-layer optimisation and configuration to specific application scenarios and deployment environments. Suitable protocol elements within the subsystems can be configured as required at time of deployment or even dynamically reconfigured after deployment.

Section 3 presents a reference model for the sensor node hardware architecture. Initial assumptions and constraints are identified. Different building blocks of the hardware platform of a sensor node are presented. Each of the building blocks is described in more detail.

Section 4 presents a reference model for the architecture inside and in-between wireless sensor networks in e-SENSE system. Assumptions and constraints for the network architecture are identified. The reference model considers three types of sensor network architectures based on the application spaces identified in D.1.1.2: body sensor networks, object sensor network and environment sensor networks. Architectural aspects that are covered include addressing, topologies, and interworking of sensor networks. Researchissues for the three network architectures are identified and adequate evaluation metrics and methodologies are presented.

Section 5 then details a reference model for the middleware architecture in the e-SENSE system. The model is based on the assumption that processing of the information captured by the sensor nodes is handled by the middleware in a totally distributed fashion. The reference model differentiates middleware functionality into semantic transport, time synchronization and localization, service discovery and context awareness, management functions as well as application dependent processing. Details of these functional blocks are given and expected research challenges in each of the blocks are identified.

Section 6 derives a reference model for the gateway architecture. The gateway enables the integration of wireless sensor networks to beyond 3G systems and provides to access the context information captured by the sensor networks. Initial assumptions and architectural consideration are provided and possible data flows to and from the e-SENSE system are examined. Based on this analysis, generic functional requirements for a gateway are identified. A reference model for a gateway is then developed and research challenges specific for a gateway are identified.

Section 7 identifies possible approaches for providing context information, captured by the e-SENSE system, to the beyond 3G service platforms. Discussed possibilities include the integration of e-SENSE system into the IP Multimedia Subsystem (IMS) through the Generic User Profile (GUP) or as a service capability function in the Open Services Access (OSA) architecture. Respective advantages and disadvantages are presented. Finally the possibility of interfacing e-SENSE with IST-SPICE is briefly explored.

Section 8 concludes the document with a discussion of security, privacy and trust implications on the overall reference model. Assumptions and constraints of wireless sensor networks are identified and resulting research challenges are outlined. Different security implications on the reference model are analysed, encompassing node, gateway and network architecture. The effects of different application environments are considered, motivating the need for a highly adaptive security framework.