科研之友
微信
新浪微博
Facebook
分享链接
摘要
An efficient channel selection scheme in multi-user cognitive radio networks (CRN) is supposed to address two often conflicting objectives: enhancing the network-wide performance while satisfying the individual quality of service demands of cognitive radios (CRs). In this sense, best-fit channel selection (BFC) inspired from well-known classical bin-packing algorithms achieves better performance compared to the longest-idle time channel selection (LITC). BFC facilitates each CR, with the capability of primary channel idle time estimation, select the channel that is expected to be idle for sufficiently long duration for its traffic request. Unlike BFC, LITC favors the selection of the channel with the longest idle time although the channel is not needed and will not be used for such long duration by this CR. As a generalization of these two approaches, we introduce the p - selfish scheme in which a CR selects the longest channel with probability p. Hence, we also refer to p as degree of selfishness. In [1], we evaluate the performance of BFC and show that it improves performance of the CRN in terms of spectrum opportunity utilization and CR throughput, compared to the LITC. In this work, we present an analytic model for BFC using continuous-time Markov chains (CTMC). The performance improvement achieved by BFC is due the reduced spectrum fragmentation that is achieved by best-fit allocation. BFC can be considered as an implicit solution to spectrum fragmentation in time dimension. We study the CR performance in terms of spectrum opportunity utilization and probability of success under various degree of selfishness through the presented model and compare our results with the simulation results.
- 出版日期2014-1
