Calculating Fourier Transforms in SQL

• Published in: Advances in Databases and Information Systems Vol. 11695; pp. 151 - 166
• Main Authors: Marten, Dennis, Meyer, Holger, Heuer, Andreas
• Format: Book Chapter
• Language: English
• Published: Switzerland Springer International Publishing AG 2019; Springer International Publishing
• Series: Lecture Notes in Computer Science
• Subjects: Fourier transform; Multi-media; Performance evaluation; SQL
• ISBN: 9783030287290; 3030287297
• ISSN: 0302-9743; 1611-3349
• DOI: 10.1007/978-3-030-28730-6_10

The Fourier transform is an important tool for analyzing, transforming and searching multi-media content in databases. SQL is the lingua franca for querying structured data. Implementing the Discrete Fourier Transform (DFT) in SQL itself has several benefits. The DFT can directly be executed in the database system. It can be reused for several, different content processing steps from feature extraction to query transformation and evaluation. We not only discuss different algorithmic aspects but also do a performance evaluation on top of different database systems of different architectures, i.e. row and column stores. The SQL-based implementation is also compared to a Python-based implementation on the client side. There is no variant that always performs best.