Next generation mobile networks have great potential in providing personalised and effcient quality of service
by using re-confgurable platforms. The foundation is the
concept of software radio where both the mobile terminal
and the serving network can be re-configurable. This approach becomes more effective when combined with
historic-based prediction strategies that enable the system
to learn about application behaviour and predict its resource consumption. We extend that concept by proposing the use of an object-oriented intelligent decision making architecture, which supports general and large-scale applications. The proposed architecture applies the principles of business intelligence and data warehousing, together with the concept of organic viable systems. The architecture is applied to the CAST (Configurable radio with Advanced Software Technology) platform.