Database Reference
In-Depth Information
! "#$
% &
''( )( **
+ ,#! -../$&
%0)' 12 &
!3$ 4
../1.15# !$&
66('%-)/2'77"18
'' &
12+5729 8! $&
% &
!123$ 4
'' &
'' 12&
% &
% &
** 0)) : ,7, ;7 : < 0) 0('=%'
../+>12! $&
66(1/1.,.?/21.!$
!../@+::$ 4
( 1.,.+11A/1-, ,&
66('%-)/2'77"18
'' 6&
')% B 721771 &
Fig. 8. Example definition of access procedure (written in pgsql )
extract_meta_data() - Extracts the meta data from the image data and returns the
hash sum of the additional meta data (e.g., environmental data).
insert_image_into_repository() - Inserts the sensor image data with the embedded
hash sum into the sensor image table.
Unfortunately, it is not possible to embed transactions into a Postgres UDF, because
the whole function is encapsulated into a transaction itself. Hence, we have to undo all
modification manually in cases of failure instead of simply calling rollback .
5
Discussion
Although the presented approach is prototypically implemented, it is worth to shed light
on main characteristics and open issues. Hence, we want to discuss advantages and
disadvantages that such an approach may have, independent of its specific realisation.
5.1
Advantages
Reliable Provenance Information. The approach addresses authenticity and integrity of
provenance data with a certain residual risk. Nevertheless, by any means we can detect
provenance data that does not hold integrity due to unauthorised modifications (e.g
during transportation). Additionally, it is possible to identify data that is not authentic
(i.e. does not contain a valid watermark).
Resource consumption: The watermark payload is embedded directly into the image-
data, raising the entropy of the data while leaving the size of the object constant. There-
fore, there is no storage overhead for the watermarked data object in comparison to
the unwatermarked object and the watermark is tightly coupled with the original data
(see [9]). Hence, the proposed approach is efficient regarding storage requirements.
Additionally, there is no additional effort for querying the embedded provenance data
directly , since with our approach, this data is stored in separate meta data tables.
Search WWH ::




Custom Search