NASA AIST Proposal

From Federation of Earth Science Information Partners

MIDAS: Middleware for Dynamic and Adaptive Service Oriented Sensorwebs

Observations from sensors spanning multiple domains, locations and timescales must be dynamically integrated and coordinated into distributed sensorweb system providing multi-scale information services and response planning to address the grand challenges of NASA Decadal (space and earth) missions. Sensing, computing and networking resources must be managed for optimally robust and adaptive performance in context of dynamic situations. Building on the success of recent NASA AIST funded VSICS (Virtual Sensorweb for Intelligent and Collaborative Sensing) work, and ongoing Space Science IRAD efforts, MIDAS will result in robust and smart middleware infrastructure that supports flexible integration of data across domains, locations and timescales. In particular MIDAS will scale and extend the distributed service coordination support to enable: i) Workflow-model based flexible and adaptive integration of services spanning streaming physical and sensing layer and decision support layer, ii) Coordination across multiple workflow instances for cross science domain fusion involving bayesian inference, and iii) adaptive self-configuration of workflow models driven by dynamic context and user utilities. A key objective of MIDAS is demonstrating the middleware in context of real and complex data analysis and prediction applications for maturing the technology to TRL of 5 and higher.