QUICK FACTS
- Developing software to predict the best coverage for antennas in their cellular network to save the company money and provide the best possible service to customers
- Needed to allow development resources to focus on modeling
- Uses IMSL C Library functions in their software for installation testing, propagation planning, cell calculation and call planning
THE
PROBLEM
The development group at E-PLUS is producing software for use by
the company's team of 70 network planners. This group works to ensure
that E-PLUS gets the best coverage from their antennae, which saves
the company money while providing the best service to their customers.
In
practice, these network planners will choose the area to be covered
by the transmitter by calculating a typical radius of transmission
and will use an off-the-shelf radio propagation package to select
possible sites for the antenna. They will then use the software
developed by Dr. Kurner's team to predict the area actually covered,
choose frequencies to be used and predict the call traffic for the
site.
THE SOLUTION
Dr.
Kurner’s Software Development Team, which programs in C and
C++, relies on the IMSL C Mathematics and Statistics Libraries,
together with the IMSL ObjectSuiteTM Math Module for C++ to speed
their application development.
The network-planning software developed by the team currently covers
"frequency assignment" and "cell calculation".
The textbook examples all use hexagonal cells, but these are too
simplistic for practical application; therefore, the group must
produce cells of arbitrary size taking into account the topology
of the land and the type of ground coverage (such as buildings and
forests).
E-PLUS currently uses the IMSL linear solvers for testing the installations,
statistical analysis for test and measurement, numerical integration
for propagation planning and cell calculation and regression routines
for call planning.
RETURN
ON INVESTMENT
The algorithms in the IMSL libraries were chosen because of the
time saved by using predeveloped numerical libraries compared with
the labor-intensive and costly process of developing and testing
home-grown routines, thereby allowing the developers to concentrate
on the actual modeling of the problem rather than the coding of
the algorithms.
