Understanding DBE’s Evolutionary Environment

The Evolutionary Environment (EvE) houses most of the self-organisation capability of the Digital Business Ecosystem (DBE) in the form of two market-driven optimisation approaches. A distributed optimisation enables DBE services to migrate to the SMEs that are most likely to need them based on the usage history of similar services.
This optimisation works over long time scales and leads to the formation of clusters of SMEs that share similar services.
Having prepared these local environments, a local optimisation is performed over short time scales on the local service pools of individual SMEs by means of Genetic Algorithms (GAs). This provides fast customised solutions to individual SME’s service requests.

Many of the outcomes of the DBE’s science domain and EvE need to be verified and adapted to existing networks before they are implemented into the technical infrastructure.

The network structure of the EvE reflects the real network of SMEs in the DBE. With the help of the EvESimulator, these networks of SMEs can be studied and analysed. The EvESimulator is the first step towards a rich simulation framework for lifelike simulation of the DBE.

In the EvESimulator, the focus is on the visualisation and observation of personalised networks to identify the best suited algorithms and tools from the EvE for the needs of a single SME within the ecosystem.

Using the state of the art agent-based economy simulation toolkit (repast.sourceforge.net) the EvESimulator has three aims:

Firstly, the EvESimulator comes with a visualisation capability as well as an easy-to-use input interface for social networks across the regions. This enables an application and visualisation of the underlying network behaviour of SMEs making it easier to understand the DBE concept and the advantages of, for example, information exchange for both, SMEs and researchers.

Secondly, when the EvE is up and running at several SME’s habitats, the real EvE implementation and the resulting network structure can be observed through the EvESimulator toolkit and scenarios for growth, cluster building and network dynamics can be studied in depth.

The third and most important aspect is that these networks, as well as artificial large scale networks, can be used by the science community to test and fine-tune their algorithms. These are parameters for, for example, genetic algorithms, different bootstrapping approaches, and rating engine functionalities.

Consequently, by implementing the EvESimulator, we have a toolkit in place that enables us to investigate new regions and domains with all of their singularity and, in parallel, enhance the underlying algorithms of the DBE infrastructure to better satisfy SMEs’ needs.

 (Figure 1)
Starting from a small network of SMEs, the simulator can show how the network grows over time…

 (Figure 2)
the future complexity of self-organising networks is observed, to assure the scalability capabilities of the EvE and DBE respectively.

 (Figure 3)
In order to adapt the technical parameter set, real social networks are used as a basis to simulate the behaviour of the system in a particular use case and tune the system to perform best (Figure 3 is based on the social network in Aragorn, Spain).