1.
A. Bhattacharyya
2.
S. Karan
3.
D Dutta Majumder
4.
V.k. Saraswat
5.
C. Mazumdar
Abstract
We present a new approach for military decision-making in
a network centric environment in perspective of warfare
information received from sensors of several networks
geographically dispersed. Sensors used across various
networks are different types, which generate data that need
to be classified for target identification in real time. We also
describe the network architecture and secure data
communication using Multi Protocol Level Switching
(MPLS)-Virtual Private Network (VPN) techniques that
enables interoperability, convergence in diverse
communication structures across land, air and sea to
protect war assets and go offensive when required, which
depends on effective communication, information
assurance and information security across a challenging
terrain and various theatres of battlefield arena. Our
proposed network architecture and classification
framework ensures packet level authentication that
enables the network to restructure itself after a large scale
or dedicated attack. It ensures scalability of the system as a
whole. It is resilient to data corrupted by enemy’s
countermeasures and can perform even if a sensor is
jammed. Also, the diversity in the classifiers of our
ensemble system allows different decision boundaries to be
generated by using slightly different training parameters,
such as different training datasets. The classification
approach primarily applies to a single target or multiple
targets that are separated sufficiently in space and/or time.
