Multi-Objective Portfolio Optimization for Spectrum Selection in Cognitive Radio Systems

The use of cognitive radio technology will allow for adaptive access to licensed and unlicensed portions of the spectrum. This, in turn, will improve end-user satisfaction and will partially alleviate the bandwidth scarcity problem. However, different parts of the spectrum not only have different network (e.g. radio and load) conditions, but also different licensing and/or billing agreements. Therefore, in order for the operators to obtain the major profit and the most efficient use of radio resources, their allocation must now target both technical and economic performance metrics. This problem can be conveniently formulated as a multi-objective portfolio optimization problem. This paper aims to adapt the concepts behind multi-objective portfolio optimization, commonly used in the areas of economics and finance theory, to the particular case of spectrum selection in cognitive radio. The idea is to investigate the advantages and potential interactions between technical and economic approaches. This paper proposes a framework for such coexistence and provides simulation results considering the Pareto optimal trade-off region obtained when using a multi-objective function based on the average return and risk per-Hz of allocated spectrum. The results show that network and economic metrics can be conveniently maximized inside the Pareto optimal trade-off region given an appropriate balance between average return and the allowed risk of primary and secondary transmissions, where risk is directly associated with the interference created from secondary to primary users.