A Framework for Self-adaptive Collaborative Computing on Reconfigurable Platforms

As the number and complexity of computing devices in the environment around us increases, it is interesting to see how we could exploit that and glue them together to create larger co-operative distributed systems. This paper describes a framework for dynamically aggregating and configuring processing resources in order to meet local requirements and constraints. The capability of this framework is demonstrated by a case study using an adaptive least mean squares filter (ALMS) application. ALMS improves convergence of least mean squares filters at the cost of more resources, and allows us to demonstrate abilities of the framework such as task offloading and run-time adaptation to available resources.