Reflecting the increased complexity and globalization of today's world, the importance of communication networks is apparent and growing in many aspects of everyday life. New networking architectures (e.g. optical networks, ATM Virtual Path networking) and the requirements of nimbleness have created a large number of network design/configuration optimization problems, many of which fall in the category of NP-hard combinatorial optimization problems. Optimizing some design and usage characteristics of communication networks leads to considerable cost savings. The main objective of the proposed research is to improve the design and usage of optical networks. This will be achieved via identifying possible design problems and/or opportunities, proposing mathematical models, and developing efficient algorithms to optimally (or suboptimally) solve these models. Specific objectives and applications include (but are not limited to):
(1) dynamic reconfiguration of multiwavelength multihop broadcast lightwave networks with variable number of transmitters/receivers per node, in order to minimize maximal congestion on its links; (also applicable to distributed memory architectures in order to decrease the time spent in communication); (2) reallocating capacity among links in a network to maximize the network utilization and minimize the total capacity required (applicable to rearrangeable T1/T3 telecommunication networks with digital cross-connect nodes, ATM networks); (3) sequencing the arrangement of links and/or reallocation of capacity to minimally disrupt network operations (applicable to networks mentioned in (1) and (2)); (4) design optimization of non-broadcast networks with optical cross-connect equipment to allow wavelength routing and re-usage (applicable to Wide Area Networks); (5) Designing minimum cost hub and spoke network by optimally (or suboptimally) locating hubs and allocating non-hub nodes subject to given restrictions (applicable to telephone networks, airline services and parcel delivery networks, distributed database systems, VLSI design). (6) Controlling user's access to common communication medium by assigning time slots for variable bandwidth switching systems (applicable to satellite-switched time division multiple access system with different rates).
The primary direction of research includes the development, enhancement, and efficient implementation of heuristic search algorithms for some NP-hard combinatorial optimization problems arising in the design of optical networks. The work on heuristic methods includes development of improved criterions for measuring the quality of the proposed heuristic solution. This leads to research on development of efficient algorithms for obtaining tight lower bounds (for minimization problems). Having a certain measure of the quality of a solution, it is interesting to investigate under which conditions the quality deteriorates (or improves). The secondary direction of research is concerned with the identification of special cases of NP-hard problems arising in communication networks design that (due to restricting assumptions) could be efficiently solved to optimality. The related work includes the design of algorithms for more general classes of problems that make use of efficient subroutines to solve subproblems.
Increasing the solvability of networking optimization problems and placing them in the dynamic environment by studying the impact of data changes on the proposed solution, adds to the applicability of such mathematical models to an even broader number of problems arising in applications of modern technology. This will, in turn, advance our knowledge regarding solvability of relevant optimization problems facing contemporary engineering.