Our team develops and applies new techniques for compressing data from various IoT data sources in order to reduce the number of bytes transmitted using wireless interfaces as well as the number of bytes stored in data centers or the Cloud.
We study both lossless and lossy techniques that exploit data similarities and correlations to compress across data from a large number of sensors, i.e., we do not compress data for individual sensors, as well as over the history of all sensor transmissions. We have pioneered a new concept called generalized deduplication, which allows not only for spatial (multiple devices) and temporal compression but that is also online in nature. This means that compression takes place as data is coming in, without waiting for all data to be collected. Data can also be accessed immediately after reception, in any order, and without decompressing any additional information. This is ideal for operating on IoT and time-series data.
Current and potential future partners include IoT companies developing sensor devices, operating wireless (mesh) networks of sensors, and those providing Cloud support. Verticals include companies developing smart meters, wind turbine companies and operators, precision agriculture, biomedical applications, amongst others.