Image Processing Reference
In-Depth Information
Software for sensor networks
Programming aspect
Behavioral aspect
Systemwide API
Splitting the complexity of APIs
Hiding heterogeneity of distributed systems
Separation of interface and implementation
Optimization of interfaces
Access to remote resources without
previous knowledge
Adaptation of software to dynamical
changes
Task changes
Evolution of network over time
FIGURE .
Two aspects of software for wireless sensor networks.
quantity, and programming language. All approaches to achieve an identical API on different systems
arecoveredbytheprogrammingaspect(Figure.).
The programming aspect enables the developer to easily create programs on different hardware
and software platforms. But an identical API on all platforms does not necessarily take the dynamics
of the distributed system into account. Ideally, the application does not notice any dynamic system
changes.hisdecouplingistermedas behavioral aspect and covers:
Access to remote resources without previous knowledge, for example, remote procedure
calls (RPCs) and discovered services
Adaptations within the middleware layer to dynamic changes in the behavior of a dis-
tributed system, caused by incoming or leaving resources, mobility of nodes, or changes
of the environment
Ability of the network to evolve over time including modifications of the system's task,
exchange or adaptation of running software parts, and mobile agents
12.3 Software Architectures
This chapter presents five important software solutions for sensor networks. It starts with the
most mature development TinyOS and depending software packages. It continues with Sensor-
Ware followed by two promising concepts, middleware linking applications and networks (MiLAN)
and EnviroTrack. The section finalizes with an introduction to SeNeTs that features interface
optimization.
12.3.1 TinyOS
TinyOS is a component-based OS for sensor networks developed at UC Berkeley. TinyOS can be
seen as an advanced software framework [] that has a large user community due to its open-source
character and its promising design. he framework contains numerous prebuilt sensor applications
and algorithms, for example, multihop ad hoc routing and supports different sensor node platforms.
Originally it was developed for Berkeley's Mica Motes. Programmers experienced with the C pro-
gramming language can easily develop TinyOS applications written in a proprietary language called
NesC [].
 
Search WWH ::




Custom Search