Kernel-Based Channel Gain Estimation for Dynamic Spectrum Access in Cognitive Radio Networks

In order to achieve dynamic spectrum access in cognitive radio networks, the knowledge of all channel gains between stations is necessary. Conventional channel estimation methods require that a pair of stations tune to the same channel, and then estimate the channel gain by transmitting some pilot signals. These methods are thus time-consuming and inefficient in multi-user cognitive radio networks. Moreover, wireless channels are affected by the small-scale fading. A one-time sample of a channel gain is thus noisy, and the small-scale fading would lead to estimation errors. In this paper we propose a kernel-based channel gain estimation method. We adopt the support vector regression to build the knowledge between the location information of each station pair and the corresponding channel gain. Such a kernel-based method is noise-resistant and time-saving. We perform a real-world experiment to measure GSM signals, and use the measurement to evaluate the performance of the proposed channel gain estimation method. Experiment results show that the proposed method with sufficient training data could achieve the root mean square error of channel gain estimation as low as 2 dB.