Spectral based solutions for (near) optimum channel/frequency allocation

Optimum channel/frequency allocation problem in a general mobile radio networks is well-known to be NP-complete. The optimum general solution for a mobile radio network for even 2-channel case is not known. In this paper, we focus on the case L=2^k where L is the number of channels/frequencies and k is a positive integer. In order to minimize the sum of the network level interference, we show that a hybrid solution (first finding the maximum eigenvector centrally and then running the well-known distributed standard minimum-interference-channel-allocation algorithm) outperforms both the spectral solution and the standard distributive algorithm. The proposed solution can be adopted by any cellular, cognitive, ad-hoc or mesh type radio networks.