Scalability became an extremely important func-tionality 
of video codecs employed in error-prone communi-cation 
environments like wireless transmission systems. The paper 
describes a multi-layer video coder based on spatio-temporal 
scalability and data partitioning. The proposed coder 
combines spatial and temporal scalability with FGS (Fine 
Granularity Scalability). The proposed solution intro-duces 
minor modifications of the bitstream semantics and syntax. 
The coder consists of two independently motion-compensated 
sub-coders that encode a video sequence and produce two 
bitstreams corresponding to two different levels of spatial 
and temporal resolution. The functionality of FGS is 
related to some drift in the enhancement layer. This drift 
can be limited by excluding temporal prediction in 
some enhancement layer frames.