Models and methods for frequency allocation with cumulative interference constraints

We consider a realistic modelling of interferences for frequency allocation in hertzian telecommunication networks. In contrast with traditional interference models based only on binary interference constraints, this new approach considers the case of a receiver disrupted simultaneously by several senders yielding cumulative disruptions that are modelled through a unique non-binary constraint. To deal with these complex constraints, we propose extensions of classical integer linear programming formulations. On a set of realistic instances provided by the CELAR, we propose several exact and heuristic solution methods including branch and cut, constraint programming, and large neighbourhood search. We also compare the performances of our best methods with those of existing heuristics and we show how the end-user benefits from using the cumulative model instead of the traditional one.