The ITSIMBW environment for simulation and decision support

In this contribution, we give an overview on the military multi-agent simulation environment ITSimBw, which is developed at Fraunhofer IAIS/ART under contract for the department A5 of the IT office of the German Armed Forces (ITAmtBw), where the project is overseen by Captain Thomas Doll. Due to the growing importance of network centric capabilities in military operations, one of the main focus points for the development of ITSimBw is the faithful modelling of IT and communication aspects. This goal is achieved essentially by two means: 1. A message format for simulated communication acts between agents is provided which allows the detailed specification of communication channel, medium, and range. 2. A voxel-space representation is used to model the extension of all objects belonging to the simulated environment. This allows for the application of high performance ray tracing algorithms to precisely determine the impact of effects like radio-pockets (e.g. caused by mountains) as well as jamming by opposing forces. These two features in conjunction allow for a detailed and realistic modeling of communication chains for reporting and command both inside and across different echelons. Clearly, the modelling of communication aspects can only reasonably be carried out in an environment which is rich enough to support the simulation of a broad variety of scenarios. ITSimBw addresses this point by strictly adhering to an agent-oriented paradigm which allows for the specification of autonomous, situation-based behaviour for all entities. This extended agent concept includes environment, weather, bridges, obstacles, and the like as active elements. This means that all effects and events are handled as actions of agents. Another important issue for any simulation system is the precise and comprehensive description of the scenarios which are to be examined. To this end, ITSimBw encompasses its own LAMPS description language. Being based on high-level Petri-Nets, it can be represented graphically and by rule-sets. Moreover, due to its generality, it is equally well suited for the description of complete scenarios as for the specification of agent behaviour. An important area of ongoing research and development is the capability of LAMPS to record events. This feature enables the creation of a scenario data-base containing mission graphs from simulation runs as well as real-world manoeuvres or even actual military missions. This data-base can then form the core component of a decision support system for the military commander. Like a chess player comparing a current board position with memorized games to determine the next move, a graph-metric can be used to liken the LAMPS graph of an ongoing mission with those in the data-base. We thus envisage LAMPS to be a core factor for the application of data-mining-techniques in mission evaluation.