Wireless cognitive network is regarded as one of the fundamental architectures of next generation wireless network. It can solve the key problem of increasing demand for spectrum and low spectrum utilization. By opportunistically accessing the temporal available spectrum, the spectrum utility can be increased tremendously. As the spectrum utility is mainly determined by the spectrum allocation scheme, how to efficiently allocate the spectrum is always one of the fundamental aspects of wireless cognitive research domain. We have proved that the optimal spectrum allocation problem under heterogeneous spectrum availability is NP hard. To solve the spectrum allocation problem efficiently, a new spectrum access approach was proposed which is based on the distributed maximum weighted independent set algorithm——DMWIS. The time complexity of this algorithm is O(V²/2). By detailed simulation, the algorithm can be finished within 3 rounds under 90% of the random network scenarios. The performance of this algorithm can reach 90% performance of optimal algorithm.