ETD

• data parallel
• hpc
• pareto
• skyline
• stream

In this thesis we apply high-performance Parallel Data Stream Processing
methodologies to approach the problem of computing the skyline over a stream
of d-dimensional points. Since the stream is possibly unbounded, we adopt the
sliding window specifications in order to maintain the skyline over the most
recent received points. We propose a parallel implementation of a module that
given as input a stream of points, produces skyline updates.