A service oriented utility grid for data parallel remote sensing applications

Satellite remote sensing field involves large number of images, which are processed at data processing/reception centers to generate value-added products. This requires large number of data and compute power to map raw data into the underlying 3D-terrain model. Due to the visibility constraints of the satellites, data is received at various ground stations distributed across the globe. Grid technologies enable us to effectively harness the vast compute/data resources available across ground stations. In this paper, we discuss a high throughput utility for generating value added products, especially digital elevation models (DEMs). The key aspect is that system is modeled to take advantage of loosely coupled nature of data acquisition chain and tightly coupled nature of data processing chain to reduce, overall product generation time. We also introduce a load based cluster scheduler and a grid scheduler to optimize the product generation process.