Tri-State Spectrum Sensing and Erasure-Injected Probabilistic Inference for Cognitive Radios
Cooperation can significantly improve the diversity order and hence the spectrum sensing accuracy in Cognitive Radio systems. Since cooperation inevitably introduces communication overhead, question arises as how and how much cooperation should be induced to attain the low-hanging fruits without being excessive and overshadowing the gain. Based on the topology graph, this paper proposes a distributed tri-state probabilistic inference mechanism for cooperative sensing. Conventional decision fusion strategies pull together all the local information (e.g. yes or no for some hypothesis) in the neighborhood, irrespective of its quality.