Distributed diagnosis of network faults and performance problems for tactical military networks

This paper describes the Tackle system and its model-based approach for analysis and correlation of fault- and performance-related problems in wireless networks. The Tackle system is intended to operate over tactical networks to help uphold communication requirements in the network that are critical to the success of a mission. It does this by producing and reasoning over executable models that adapt to changing network topologies and communication requirements. The Tackle modeling approach allows reasoning about network state to be distributed throughout the nodes in the network, and to be able to operate at different perspectives at multiple levels of aggregation based on the layers in the network hierarchy. Each perspective at each layer of the hierarchy is represented in a correlation domain. During operation, each Tackle Correlator reasons over its own correlation domain. Tackle Clearinghouses reconcile network state that is shared by multiple correlation domains. This approach supports the eventual goal of consistent and informed root cause analysis across the network. We have implemented and rigorously tested a correlation engine that executes the models in a simulated environment with multiple nodes in multiple subnets. We present our testing results and analysis at the end of this paper.