The computational limitations in on-board sensing devices
highlight the importance of IDs-based DPS for Local Analytics
(LA). The IDs collect data from local sensors and near-by
on-board sensing devices and process it locally. For example,
OMM [4] and CAROMM [5] enable LA by executing
complete KD process inside mobile phones. Similarly, StreamAR
[6] process data streams locally inside mobile phones
and uncover activity patterns from on-board accelerometers.
Although, IDs are successfully adopted as LA platforms
but resource-constraints bound to the development of Lightweight
algorithms in adaptive environments. Hence, either the
accuracy or timeliness of the discovered patterns has to be
compromised. We are presenting a feasibility study in this
paper to find the opportunities of executing full algorithms in
addition with light-weight algorithms.