Autonomic Frameworks Deployment Using Configuration and Service Delivery Models for the Internet of Things - Interoperability and Open-Source Solutions for the Internet of Things - page 91

3.3 Plan and Processing

During the plan phase, the agent communicates with either the IoT platform or the
underlying infrastructure and ask/do necessary changes. What changes need to be done
is decided in this phase. The decision can vary from simply deploying the platform to
increasing the initial resource allocation, and might result in negative feedback from the
agent about the deployment. As an example, the de ned algorithm to do this processing

can be described as follow:
check system_resource
check IoT_rec_resource
map system_resource -> IoT_rec_resource
check performance_parameter
do PLAN

3.4 Adaptation and Execution

As the Plan phase decides what to do, the Adaptation and Execution phase follows the
decision, and executes the decision taken by the Plan and Processing phase. The
analysis phase has already resolved the dependency sources, the plan phase has set
the platform to be best suited for the IoT platform, now the Execution phase will deploy
the IoT test-bed into the infrastructure that has been prepared. Although the depen-
dency map has been created in the Analysis phase, the resolving of those de ned

dependencies are done in the Execution phase.

4 Autonomic Con guration Agent

The self-tuning or self-con guration facility of the IoT service delivery model is a novel

approach and is presented in this section. An autonomic framework for continuously
monitoring and optimizing the IoT test-bed regardless the infrastructure on which the
platform is hosted is described. The optimization can be done via several parts, but
in this work the cloud optimization [ 3 ] segment is the main focus. The autonomic-
con guration and adopted autonomic life cycle is the one described in the previous

section of this paper. Figure 2 depicts the functional architecture of the autonomic agent
and the designed features that it performs, the cycle start by monitoring the conditions or
parameters of the resources to be offered, then the next step perform an analysis of the
information and the operations de nes the way the autonomic agent will work. The third

phase denes the best performance based on conditions and nally the execution of the
changes into the system is performed.

4.1 Adaptation for the Internet of Things

For the life-cycle design, the generic autonomic agent life cycle proposed by Horn [ 1 ]
has been used as reference. In the very rst stage of the deployment of the autonomic