The reliability of traffic flows is a serious challenge to cognitive radio networks due to the interruptions from primary users. A backup path scheme is proposed to protect the secondary user traffic flows against the interruptions from primary users. In such a scheme, the traffic flow can be switched to a backup path if the working path is interrupted by primary users. When primary users leave, the traffic flow will be switched back to the working path. The problem of selecting backup paths is formulated as an integer programming problem. Given a backup path, a statistical rule is proposed for switching the traffic flow to the backup path, based on the Bayesian decision framework. The switching rule is implemented on a USRP GNU Radio node based testbed, which consists of a multi-hop cognitive radio network and a primary user. The hardware experiment shows that the system cost, defined as a weighted sum of throughput loss and average packet delay, is reduced by around 50% in a typical setup, compared with the scenario of no backup paths.
展开▼
