Naval at sea (maritime tactical) networks are
characterised by a dynamic, heterogeneous, and low-bandwidth environment. The
effective management of communication resources in this domain is critical, but
is hampered by constraints imposed by a hierarchical
command structure and dynamic mission requirements. The most critical network
management issue in maritime networks is the limited bandwidth connecting each
node (ship) is often insufficient to support the network traffic generated
locally. This leads to very poor perceived Quality of Service (QoS) for all traffic. A solution common in fixed networks
is to use Traffic Engineering (TE) techniques. The goal of TE is to facilitate
efficient and reliable network operations while simultaneously optimizing
network resource utilization and traffic performance.
This concept has been applied
in the maritime environment by developing four Policy-Enabled Traffic
Engineering (PETE) services for this environment: traffic monitoring, traffic
prioritisation, adaptive routing, and resource reservation. The flow-based
resource reservation service is a novel mechanism we developed to provide
robust and efficient end-to-end bandwidth reservations.
To evaluate the PETE services, our methodology focused on modelling and
performance evaluation of the management services using simulation. Since
maritime networks have not been described in depth in
the literature, the modelling exercise provided valuable insight into the
challenges of operating such networks. Simulation allowed us to evaluate our
management solutions for different networks sizes, node mobility, and traffic
types.
Results from these simulations are encouraging. With policy control, the
traffic monitoring service was able to adapt to dynamic network conditions and
provide global traffic statistics within a policy defined delay. The traffic
prioritisation service was able to achieve an improvement in delay from
approximately 17% in a saturated network to 52% in an overloaded network.
Similarly, adaptive routing greatly improved the QoS
achieved by sending some traffic over under-used links while simultaneously
improving the QoS of other traffic by spreading the
load over multiple links. The resource reservation service was
found to be particularly effective in the maritime environment with an
acceptance rate 19-76% better than RSVP and 86-1095% better than INSIGNIA, two
alternative reservation protocols. The resource reservation
service model was validated against results from a prototype
implementation.