Towards Sustainable IoT: Space Efficiency and Serialization Speed of Data Exchange Formats

This study presents a comparison of data exchange formats in terms of space efficiency, serialization speed and usage complexity. A total of 27 data format variants were considered and divided into the categories binary, textual, schema-driven and schema-less. Further, 8 use cases were defined that represent data potentially transferred in IoT applications. In terms of space efficiency, Avro outperformed the other formats in all use cases. Cap'n Proto (packed) generally proved to be faster than the other formats when performing the serialization step. In addition, binary formats performed better than textual formats in terms of space efficiency and serialization speed. The same applies to schema-driven formats, which tend to be more efficient than schema-less formats in terms of both space efficiency and serialization speed. No clear correlations were found between usage complexity and the performance aspects investigated, further justifying the need for a comprehensive collation of data format characteristics that allows for informed decisions in the context of sustainable IoT use cases implementations.