Abstract: In classical spectrum sensing performance analysis methods,primary user's( PU) signal is usually assumed to occupy the entire sensing slots. However,the PU is likely to access or leave at any time of the sensing slot. In response to this situation, the random access or leave behavioral characteristics of the PU was modeled,based on which the spectrum sensing performance was analyzed.First,according to the PU's access or leave moment within the sensing slot, four scenarios were discussed,and the probability expression of PU's signal duration within the sensing slot was obtained.Then by adopting energy detection method,the SU's average sensing probability function expression was derived theoretically. Simulation results verify the rationality of theoretical derivation,and show that the PU's random access and leave behavior not only significantly affect the trend of sensing performance,but also decrease the SU's detection probability by 50% maximally.