Relational Database for PNT Data

Navigation filter researchers often deal with multiple sets of data collected from different sources. Over time it can be difficult to identify how data was collected and how to model it without sufficient data storage techniques and documentation. A navigation database would allow storage of sets of Position, Navigation and Timing data, along with designated metadata, which would enable future researchers to access and understand historical navigation data. This paper proposes three approaches for a PostgreSQL relational database designed to store navigation test data based on the Scorpion Data Model. Each approach uses different schema for storing navigation data, and identical schema for storing sensor and non-sensor metadata. Using queries designed to be of interest to filter researchers, the authors designed test scripts to rank all three designs according to how quickly the original files could be recreated, and how quickly queries based on sensor, non-sensor, and SDM data returned correct information. In order to test how the different approaches scaled when the databases became larger, these test scripts were used with six databases (two for each approach) with 100 and 1000 logs of repeated navigation test data and randomized metadata. This paper presents the results of these tests, along with a background of relational and NoSQL databases, schema details for each approach, query and testing details, and an analysis of how each approach performed across all tests. Finally, we identify the navigational database schema with the best overall performance based on the data and analysis.