Sensor Networks’ Integration - Application and Multidisciplinary Aspects of Wireless Sensor Networks - page 100

Hourglass provides reliability to system dynamics by explicitly supporting a
mechanism to deal with temporary disconnections of a circuit, that is, if the con-
nectivity to a service provider that is part of a circuit becomes unavailable.
Disconnections are monitored by heart beat mechanism along the circuit, based on
explicit control messages or implicitly by data that are exchanged. Once a discon-
nection is detected appropriate actions such as buffering of data can take place in
the circuit. Thus, Hourglass offers the advantage to modify existing circuits to adapt
the services to changing conditions for continuous application queries. It allows
optimization of the delivery of same sensor information by combining transmission
between service endpoints across multiple circuits/applications. While the architec-
ture has the intention to scale well by planning for distributed operation across
multiple service providers, it leaves open how respective service registries distrib-
uted across multiple service providers interact or are managed and how the connec-
tion managers of different service providers cooperate to establish connections
across multiple domains. The overhead for establishment of circuits for each data
request by application together with the fact that state needs to be maintained at
each node that is part of the circuit is a severe scalability concern. While it may be
justified for streaming-type queries for longer periods of time, it does not suit well
one-shot queries or periodic queries with little data exchange. However, the frame-
work does not address functions for security, privacy and trust, and accounting.

Urban Sensing

In the Urban Sensing project [ 5 ] they consider three types of applications: personal,
social, and urban. A personal application uses information about the end user for
the purpose of the end user. A social application mimics Facebook and other social
networking sites, where data are shared among a set of users for free. In urban
applications the users share data with the general public, and the importance of
identity control, etc., is thus much higher. It is argued that new network architecture
is required in order to share data in a controlled way and to assure basic quality
checks of data. In this the authors see an evolution from single-domain WSNs to
collective/federated WSNs to full integration into the full global infrastructure. The
federation of WSNs is referred to as the sensor fabric.
In order to achieve full integration, the global network must know about the
abstractions used in forming the sensor fabric, and the sensor fabric must import
notions about the future global network into itself. It is hence a two-way process.
Abstractions required to form the sensor fabric could be of the following types:

• Space-time coordinates

• Policy-mediated rendezvous based on data properties and metadata

• Aggregation-based reliability

The authors argue that embedding these abstractions into the global network
changes the network from host-centric to data-centric in nature.
The authors subsequently argue that most important to solve while incorporating
sensors into the global network are the issues of verification, privacy, and dissemination.