Hard data fusion: With the user type estimates ˆm given in(24), a straightforward (yet effective, as shown by numericalresults) strategy of the fusion center is to directly discardall the local spectrum inferences from the detected malicioususers, and apply majority voting on the data from honestusers to decide the spectrum states. The block diagram of thisapproach is illustrated in Fig. 3. The fusion center maintains a data buffer for the most recent sensing reports (of length T )from all secondary users, and adopts the proposed algorithmto estimate the corresponding HMM parameters. The resultingestimate ˆλ together with the observation history Ot−T:t−1is fed into the MAP block for malicious user identificationusing (24). Spectrum occupancy decision is made at the datafusion block, based on the current sensing reports Ot and userclassification result ˆm.