Distributed individual-based environmental simulation

This paper describes the implementation of a distributed modelling environment which allows the simulation of a large number of individuals. Particular attention will be paid to modelling an individual's behaviour, communication and interaction with the population and a shared environment. Individual based modelling is not a new concept, nor is the idea of distributed simulations, the system detailed here offers a means of combining these two paradigms into one large-scale modelling environment. A key concept in this system is that each individual being modelled is implemented as a separate entity. This atomisation of the model allows the simulation a greater flexibility, individuals can be rapidly developed and the simulation can be spread over a number of machines of varying architectures. In an attempt to produce a flexible, extensible, individual-based model of a large number of individual subjects the client-server paradigm has been employed. Combining the individual-based modelling techniques with a client-server network architecture has been found to be straightforward with the added bonus of having communication between individuals included for free. The idea of considering the problem as one of interaction between an individual and the environment means that the problems normally associated with distributed simulations, those of continuity of world-views for different clients and of communication between clients, are easily solved. Although this system has been developed originally to allow simulations of the mountain gorilla population, the modelling methods employed have meant that almost any entity can be simulated with very little change to the basic simulation processes described here. As such details of how the gorilla behaviour or learning will be implemented are not covered except for how they are facilitated by the system See Scahill (1996, 1997) for details of the system not covered in this paper.