Image Processing Reference
In-Depth Information
supports self-forwarding of capsules to reprogram a whole sensor network with new capsules. It is
thetaskofthesensornetworkoperatortoinjectcodecapsulesinordertochangethebehaviorofthe
network.
The program execution in MATÉ starts with a timer event or a packet receive event. The program
counter jumps to the first instruction of the corresponding context (Clock or Receive) and executes
until it reaches the Halt instruction. Each context can call subroutines for expanding functionality.
The Send context is invoked from the other contexts to send a message in response to a sensor reading
or to route an incoming message.
The MATÉ architecture provides separation of contexts. One context cannot access the state of
another context. There is only one single shared variable among the three contexts that can be
accessed by special instructions. he context separation qualifies MATÉ to fulfill the traditional role
of an OS. Compared to native TinyOS applications, the source code of MATÉ applications is much
shorter.
12.3.3 TinyDB
TinyDB is a query processing system for extracting information from a network of TinyOS sensor
nodes []. TinyDB provides a simple, SQL-like interface to specify the kind of data to be extracted
from the network along with additional parameters, for example, the data refresh rate. The pri-
mary goal of TinyDB is to prevent the user from writing embedded C programs for sensor nodes
or composing capsules of instructions regarding to MATÉ. The TinyDB framework allows data-
driven applications to be developed and deployed much more quickly as developing, compiling, and
deploying a TinyOS application.
Given a query specifying the data interests, TinyDB collects the data from sensor nodes in the
environment, filters and aggregates the data, and routes it to the user autonomously. The network
topology in TinyDB is a routing tree. Query messages flood down the tree and data messages flow
back up the tree participating in more complex data query processing algorithms.
The TinyDB system is divided into two subsystems: sensor node software and a Java-based client
interface on a PC. The sensor node software is the heart of TinyDB running on each sensor node.
It consists of
Sensor catalog and schema manager responsible for tracking the set of attributes, or types
of readings and properties available on each sensor
Query processor, utilizing the catalog to fetch the values of local attributes, to receive
sensor readings from neighboring nodes, to combine and aggregate the values together,
to filter, and to output the values to parents
Small, handle-based dynamic memory manager
Network topology manager to deal with the connectivity of nodes and to effectively route
data and query subresults through the network
The sensor node part of TinyDB is installed on top of TinyOS on each sensor node as an application.
The Java-based client interface is used to access the network of TinyDB nodes from a PC physically
connected to a bridging sensor node. It provides a simple graphical query-builder and a result display.
The Java API simplifies writing PC applications that query and extract data from the network.
12.3.4 SensorWare
SensorWare is a software framework for wireless sensor networks that provides querying, dis-
semination, and fusion of sensor data as well as coordination of actuators []. A SensorWare
platform has less stringent resource restrictions. he initial implementation runs on iPAQ handhelds
 
Search WWH ::




Custom Search