الفهرس 
Abstract
Types of Industrial Control NetworksOnly 14 pages are availabe for public view
 |  | from | 244 |  |  |
| | | from | 244 | | |
Abstract
Modern SCADA (Supervisory Control And Data Acquisition) networks are designed and developed to provide real-time access (local or remote) to industrial control processes for management, control and supervisory purposes. They are widely adopted to manage and coordinate large and complex national critical infrastructures (i.e., power systems, oil and gas, water transportation, etc.) based on information and communication technology. The aim of this study is to propose a new paradigm for engineering modern large-scale SCADA networks based on Multi-Agent Systems (MAS) as a promising software engineering architectural style that adopts the decentralized control approach in contrast to the traditional centralized control approach which is used within most of current SCADA networks, and is widely supported, but it is more suited to centralized small-scale applications as it is lacking in flexibility and robustness and can’t handle the dynamic changes usually present in complex large-scale applications. This thesis proposes a bio-inspired organizational model called NOSHAPE (inspired from a unicellular organism called Amoeba which has no definite shape) for dynamically and transparently reorganizing large-scale MAS-based applications to adapt the work environment dynamic changes. The proposed organizational model provides a novel technique to structure and coordinate the interactions of a large number of agents distributed at large distances by exploiting the overlapping of higher order entities within large-scale MAS (i.e., agent organizations) to share and possibly exchange agents in order to reduce the scope of agents’ interactions, coordinate agents’ behaviors, and enable agents to find each other based on the services they provide. NOSHAPE MAS dynamically, transparently, and efficiently reshape their structures to provide the required technical and social contexts for individual agents to interact in order to reach their design objectives. Different from the currently used paradigms with which the designer needs more initial knowledge in design-time about the concrete application and that is usually not possible with complex applications, NOSHAPE MAS realize dynamic system reorganization using simple, predefined, and ordered set of contextualized organizational interactions.
The applicability of the proposed NOSHAPE MAS organizational model was demonstrated by developing a self-adaptive large-scale MAS-based SCADA network comprises a large number of agents. A performance evaluation has been conducted with a number of empirical evaluation experiments. The performance evaluation shows that the realized dynamic and transparent adaptation continuously improves the response times of agents’ interactions by localizing remote agents. Over time the response times of most remote agents’ interactions will seem to be as those of local interactions thanks to the transparently provided macro-level dynamic adaptation behavior. Further, two of the provided evaluation experiments have been carried out based on the traditional MAS development approaches (i.e., federation) and their results manifested the performance limitation of these traditional solutions compared to NOSHAPE MAS with which a performance improvement of about 47% was achieved.
To conclude, the proposed NOSHAPE MAS organizational model has the ability to efficiently deal with the dynamics and uncertainty usually present in large-scale complex real-life applications such as modern complex large-scale SCADA networks. Further research is recommended to give the focus to performance, resiliency, security, and agent mobility within modern SCADA networks based on NOSHAPE MAS.