Random signal detection method based on non-reconstruction sequential compression in cognitive network

The invention provides a random signal detection method integrating an individual cognition user and multiple users in a distributive manner on the basis of non-reconstruction sequential compression. The method comprises the steps of sampling and projection transforming a signal of a broadband main user in a cognitive radio network, calculating the detection statistic quantity and the likelihood ratio by directly utilizing a low-speed observation sequence, and comparing the likelihood ratio with a judgment threshold; resampling if the detection precision cannot meet the requirement of the user, forming a novel low-speed observation vector through the preliminary low-speed observation vector, and rejudging; repeating the steps until the precision meets the requirement of the user. By adopting the method, not only can the advantage that the compression sampling data processing quantity is low be maintained, but also the signal reconstruction can be completely avoided, more importantly any prior information of a main user signal is not needed, and the compression ratio can be self-adaptively adjusted. On the premise of guaranteeing the detection precision, the observation number can be reduced as far as possible, and the calculation quantity is reduced, so that the time expenditure can be saved, and the detection real-time property can be improved. The effectiveness and accuracy of the method are verified in a simulation manner.