A Top-Down Approach for a Generic Programming Abstraction in Real Space-Time

Abstract

Today’s widely used programming approach for
mobile distributed systems, e.g., swarms, is bottom-up. I.e.,
the programmer has to be aware of the system’s distribution.
Such a kind of programming corrupts the principle Separation
of Concerns and turns out to be complicated. This
article proposes a top-down approach for the programming
of mobile distributed systems. It should be incorporated in a
new operating system to be developed by our research group.

Classical distributed systems often are intended to hide
their distribution from the user but not stringently from the
programmer. Moreover, most applications don’t consider the
executing system’s location and mobility, respectively. However,
state-of-the-art mobile distributed systems’ applications
are widely based on location and motion data. So, the programming
approach of classical distributed systems which
abstracts from location and motion might no longer be convenient.
We suggest raising the level of abstraction in order
to hide the system’s distribution from the user and programmer,
in contrast to the bottom-up programming approach of,
e.g., swarms. This distribution transparency within the executing
system is intended to be combined with location/-
motion awareness within the application so that the requirements
of modern applications can be met. The programming
of such a distributed mobile system will be separated from
the complex and error-prone application partitioning and assignment
a bottom-up approach would impose. This offers
further benefits like scalability and robustness with regard to
scheduling of sub-activities and sub-systems, respectively.
We promote the use of spatiotemporal constraints to realize
such a top-down approach. These constraints will be introduced
and explained using two examples.