This thesis considers a logical network design problem for a lightwave network for which subcarrier and wavelength division multiplexing are used. An assignment of transmitting-wavelengths, receiving-wavelengths and subcarrier frequency to each node characterizes a logical connectivity among users in the network. The lightwave network permits logical connectivity to be updated in response to dynamic traffic patterns, which induces a reconfigurable network environment. The total achievable throughput is a critical performance measure for networks, for which design objective consists of finding network connectivity such that the largest traffic load among the links is minimized. So as to accommodate nonuniform node-to-node traffic flows. The problem is expressed in a mixed integer programming for minimizing the maximum link load for which heuristic algorithms are proposed by using simulated annealing(SA), tabu(TB) and genetic(GA) techniques. These algorithms are tested and compared one against another for effectiveness and efficiency by use of several different numerical examples.