Progress & ToDo

Things that we found out so far

General:

Format seems to be Little Endian Binary
The tag can only be read by the reader device that started the tag. However, this could be related to a legal issue (avoidance patient misidentification) rather than a technical issue.

Layout:

Bytes 317 and 316 contain the number of minutes since the sensor was started (as short value). This value will stop to count up when the sensor reaches end-of-life.
Data is separated in two parts: First part contains data of the last 15 minutes (probably for calculating the trend direction of the glucose level). The second part contains historic data of the last 8 hours.
Byte 26 contains the block number that will be written next for the first part (15 minutes trend)
Byte 27 contains the block number that will be written next for the second part (historic data)
15 Minutes trend data is found on bytes 28 to 123
Historic trend data is found on bytes 124 to 315
Each data sample is of size 6 bytes
Trend data is written as a ring buffer of fixed size. The next block to be written can be found in byte 26/27
After byte 317 to the end there is unused memory on the tag (that's why the reader from Abbott is so fast compared to the app). Each tag contains the same memory here...

It could be that each data sample contains of two values with each 24 bits. I suppose these are raw data from the sensor that need calculation to fit to blood glucose levels.