Autonomic Aspects of IoT Based Systems: A Logistics Domain Scheduling Example - Interoperability and Open-Source Solutions for the Internet of Things - page 156

(2) the depot administrator sends the list of materials to be transported (for all
the construction sites) to the transport company;
(3) the transport company sends the transport plans to the minivan drivers;
(4) the minivan drivers request the materials from the depot;
(5) the minivan drivers requests approval to unload the construction materials;
(6) the transport company monitors the state of the trac; reconfigures the
plans if an unexpected event (e.g. trac jam) occurs and sends the new
transportation plans to the minivan drivers.
4 System Architecture
4.1 Autonomic Route Scheduler
The complex environment required to evaluate the proposed autonomic aspects
of the IoT system has been recreated using the Constrains Responsive Mobility
(CoReMo) software platform, which is based on Repast Symphony multi-agent
system [ 7 ].
CoReMo has been developed as an extensible emulator platform used for
testing a variety of concepts and solutions. A dedicated extension of the platform
is focused on autonomic systems involving mobile as well as fixed agents. While
the former represent vehicles the latter are exposing fixed infrastructure elements
such as trac lights, loop detectors, crossings or road signs. Different trac
load conditions are simulated using a wide range of dummy agents whereas the
weather conditions are replicated using the integrated environments simulator.
The version of the emulator deployed for this scenario, whose architecture of
is depicted in Fig. 3 , consists of CoReMo Core and an updated version of the
CoReMo Citadel extension.
CoReMo Core represents the simulator kernel providing generic simulation
functionality and is build on top of the RepastSymphony simulation framework.
Asetof contexts and projections are used to build the agent space. Contexts are
used as containers for different agents, grouping them based on functionality or
other user defined criteria. Projections are used to relate agents to the Cartesian
or 3D space, or to a network structure supporting graph processing.
Complex agent behaviors are modeled using the Petri Nets formalism and the
translation of these nets into executable code is achieved by using PetriNetExec
[ 13 ]. PetriNetExec is Java library we have developed in order to allow embedding
of Petri Nets into executable code. This functionality is critical especially when
complex behavior is modeled using Petri Nets thus leading to a large number
of places, transitions or inhibitor arcs whose translation into code without this
dedicated library would be dicult and prone to errors. A set of callback func-
tions are attached to events relevant for the network evolution at runtime and
their description is documented in [ 14 ].
The Simulation Control component provides the simulation time keeping
and advance mechanism whereas the Event Messaging provides support for the
interconnection of different components of the simulator.