General Spectrum Sensing in Cognitive Radio Networks

It is well known that the successful operation of cognitive radio (CR) between CR transmitter and CR receiver (CR link) relies on reliable spectrum sensing. To network CRs requires more information from spectrum sensing beyond traditional techniques, executing at CR transmitter and further information regarding the spectrum availability at CR receiver. Redefining the spectrum sensing along with statistical inference suitable for cognitive radio networks (CRN), we mathematically derive conditions to allow CR transmitter forwarding packets to CR receiver under guaranteed outage probability, and prove that the correlation of localized spectrum availability between a cooperative node and CR receiver determines effectiveness of the cooperative scheme. Applying our novel mathematical model to potential hidden terminals in CRN, we illustrate that the allowable transmission region of a CR, defined as neighborhood, is no longer circular shape even in a pure path loss channel model. This results in asymmetric CR links to make bidirectional links generally inappropriate in CRN, though this challenge can be alleviated with the aid of cooperative sensing. Therefore, spectrum sensing capability determines CRN topology. For multiple cooperative nodes, to fully utilize spectrum availability, the selection methodology of cooperative nodes is developed due to limited overhead of information exchange. Defining reliability as information of spectrum availability at CR receiver provided by a cooperative node and by applying neighborhood area, we can compare sensing capability of cooperative nodes from both link and network perspectives. In addition, due to dynamic network topology lack of centralized coordination in CRN, CRs can only acquire local and partial information in limited sensing duration, robust spectrum sensing is therefore proposed to ensure successful CRN operation. Limits of cooperative schemes and their impacts on network operation are also derived. Index Terms Spectrum sensing, cognitive radio networks, link availability, network tomography, statistical inference, reliability, neighborhood.