Policy-Enabled Traffic Engineering in Maritime Tactical Networks
Carleton University, Ontario, Canada. August 2008.

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.