|
In-Motes: Overview
Wireless sensor networks
(WSNs) have been identified as a promising technology that will allow people and
machines to interact with their environment in a revolutionary way. These
networks, however, are facing limitations such as energy constraints of the
sensor and difficulties in reprogramming the actual network. To address these
limitations we propose a novel agent middleware. Namely In-Motes can be
considered as an intelligent network which is deployed with no pre-installed
application. Mobile agents are injected into the network, then migrate and clone
across it, following specific rules and performing application specific tasks.
By doing so, each mote is given a certain degree of perception, cognition and
control, forming the basis of its intelligence. Linda-like tuplespaces and
federated system architecture are proposed as the means for collaboration and
coordination of the agents. In order to make the network more robust, certain
behavioural rules are proposed taking inspiration from a community of bacterial
strains. These preserve each agents certain degree of autonomy and identifies a
highly coordinated architecture for them.
In-Motes is based
on Agilla and Mate by allowing users to inject agents inside the network and
provides a high level architecture for the given agent community based on
federated systems and behavioral rules produced by a parallelism of bacterial
strains. Mobile agents encapsulate a dynamic behavior within the network,
adopting certain degrees of perception, cognition and control of the given
environment and acting as local virtual machines with dedicated instructions and
rules. An agent can migrate or clone from one node to another and
communicate/coordinate its actions based on Linda-like tuplespaces and federated
system architecture.
The In-Motes model is shown in the billow
figure. Each node can support multiple agents and maintains a tuplespace where
specific reactions, that will be mentioned later, can be stored. This tuplespace
will be accessible by all agents and is static as agents cannot carry it
throughout the network.
This shared memory model will be enforced in
a federated system where the facilitator agent will generate specific
instructions to the analogous node that hosts it. This will be allowed since
each tuple will contain a specific set of fields with a type and a value. Types
will consist of various sensor readings. Following the Agilla model each tuple
will be accessible by any agent whose specific template matches by type. Such a
match will be available if they have the same number of fields and each field in
the tuple satisfies the analogous field in the template.
|
