Optimal waveband switching in WDM networks

Switching traffic together in bundles of wavelengths called wavebands can greatly reduce the switching costs of the network. We consider the problem of partitioning wavelengths into wavebands for star networks using the minimum number of total wavebands. We provide a greedy algorithm for waveband partitioning and show that it is optimal in that it requires the minimum number of wavebands subject to using the minimum possible number of wavelengths. We also give an algorithm for allocating calls from any admissible traffic set to the wavebands in a non-blocking manner. Finally, we show that the increase in the number of wavebands required is logarithmic in the number of calls and polynomial in the size of the network.