Integrated Management of Hybrid Communication Networks
Over the last eight years I have undertaken intensive research on many aspects of hybrid (seamlessly interconnected terrestrial and satellite) communication networks. In collaboration with Hughes Network Systems we developed a next generation integrated network management system which relies heavily on sophisticated techniques from object oriented databases, rule databases and graphical user interfaces. We have developed new object oriented data models for network management. We have developed also distributed object oriented and dynamic graphical user interfaces, as well as distributed middleware systems for integrating heterogeneous management information bases (MIBs) and heterogeneous network management applications.
We have developed dynamic routing and access control algorithms for multimedia hybrid networks. We are also investigating intelligent systems for fault and performance management of large hybrid networks. We have developed with Loral Space a system for learning and self-healing network fault management. In particular, we are developing adaptive and efficient Operations and Maintenance (OAM) cell strategies for broadband network monitoring. Sensing and monitoring results are use for pro-active fault management via adaptation of routing and access control strategies. Belief networks are used to learn fault patterns and to guide on-line probing for fault monitoring. Decision theory is used to design probing and monitoring strategies. More recently we initiated research on the use of JAVA applets that can be sent to network devices and compute specific statistics for monitoring, including altering the monitoring computations designed into the network element. This technique is more efficient than OAM cell based monitoring.