| |
| Research |
Since the very beginning of our consulting
business we were anxious to adapt the most recent research results to practical
concerns. Over more than 20 years we had the chance to participate in findings
of AT&T Bell Laboratories, IBM Watson Research Labs and several
universities. One of our utmost concerns has been the architecture,
installation and operation of call and contact centres. Insofar our consulting
services in that field date back to the times of incumbent telecomm operators
and monopolies. Several years ago we decided to give something back to the
scientific community and started our own research on call centres.
From a business point of view, a call centre is an entity that combines voice
and data communications technology to enable organizations to implement
critical business strategies or tactics aimed at reducing costs or increasing
revenues. At an organizational level, cost are highly dependent on capacity
management of human and technical resources. By utilizing methods of operations
research and especially of queueing theory, a suitable description in
mathematical terms is to be found. A survey on queueing theoretic results
geared towards the solution of problems occuring in telephony is provided
here. A much simpler
introduction covering also relevant tools such as transforms, difference and
differential equations may be downloaded from
here.
Current models are often restricted to a very simple set of operations, e.g.
call centre agents toggle only between an idle and a talking state. It turns
out, that these simplifications do not reflect reality in an appropriate way.
Considering the given example, an after-call work time may be introduced
following talk time. Obviously there is a high impact on capacity of system
resources, because in after-call work time the agent phone is not occupied,
while the agent indeed is. Accordingly a single view is not sufficient and
models have to be adapted to reflect a technical (system oriented) and a
(human-)resource layer. We have developed methods based on matrix exponential
queueing systems, which are capable to cope with this new requirements. The
final document is provided
here. Related
computer programs are available for
R and as
Classpad 300 Library.
Furthermore we have negotiated an industry partnership, which allows for
validation of the new methods in an applied setting. The results will be
published as a case study in the final document. In addition to that we will
develop new methods for parameter estimation in matrix exponential queueing
systems. One target is to create algorithms, which are capable to estimate
parameters in real-time and are ready to be implemented as part of a management
information system. The models and techniques of parameter estimation will be
evaluated theoretically and empirically. By combining our parameter estimation
algorithms with a suitable matrix exponential model, the prediction of call
centre load in real-time will become a reality also for complex
configurations.
Due to personal reasons we had to cease researching
call centre related topics. We have decided to publish the
entire work under
a noncommercial Creative Commons license. On about 500 pages it covers call
centre technology as well as modeling and performance analysis. All chapters
except the one discussing call centre architecture in public networks are
complete. As mentioned above data have been collected to support use and
validation of the proposed methods in a more general setting, but the
corresponding analysis has not been carried out yet. Consequently the chapter
describing this field study has been omitted completely. We had the hope to
provide a holistic and complete document on call centres, where topics are put
in context instead of being issued in the form of unrelated scientific papers.
In releasing it to the public we hope, that researchers will carry on what we
have started.
Occasionally we will also publish documents, which do
not relate to traffic engineering of call centres. These will be found in the
resources section only, so it may
be worth a visit.
|
|
