Online expert-based prediction for cognitive radio secondary markets.

Abstract

The growing importance of wireless communications drives an increasing interest in dynamic access to spectrum resources. This requires efficient management policies that allow spectrum sharing between licensed primary users (PU) and unlicensed secondary users (SU). On such scenario, PUs shall preserve their usage priority right over any SU. Also, no SU shall interfere on any PU. Technical viability can be achieved through Cognitive Radio devices that adjust their operating parameters adaptively. After discussing several economic and technical models to achieve efficient spectrum sharing, we propose an on-demand secondary market model regulated by a spectrum broker who controls resource allocation. This model provides economic incentives for both kind of users to cooperate: SUs are charged by the broker on behalf of PUs for resource utilization but are indemnified if expelled to ensure PU priority. We describe the main characteristics of such a system and address the question of what allocation decisions should the broker take in order to achieve economic benefit regardless of users behavior. Several online expert-based no-regret algorithms are proposed to guide the decision taking process and evaluated under different user behavior patterns. Their results are compared with the ones achieved by dynamic programming to assess its convenience.