Cognitive radio sensor networks (CRSNs) have been widely considered to monitor the smart microgrid environment. In such an environment, different applications with various characteristics may coexist in the network. As a consequence, the sensor nodes deployed to collect the generated data need to access the available spectrum resources with different priorities. Thus, achieving a fair channel allocation in the microgrid driven CRSN is a very challenging task. In this paper, we propose a new Distributed Unselfish Spectrum Assignment approach (DUSA) for One-Hop smart microgrid communication network. The proposed approach exploits the promising cognitive radio technology and the heterogeneity of the traffic generated by sensor nodes to satisfy the requirements of the smart microgrid applications. It achieves channel allocation without relying on a predefined common control channel (CCC) to exchange control messages. As DUSA is a distributed approach, it aims to dynamically predict the primary channel availability. Thereafter, each sensor node estimates the buffer occupancies of its neighbors through discrete Markov chains. Unselfish distributed channel allocation is achieved based on a Partially Observable Markov Decision Process (POMDP) formulation. Performance evaluation using the OMNeT#x002B;#x002B; simulator, reveals that DUSA achieves a fair sharing of the spectrum resources and improves the spectrum utilization compared to the CCC-based resource allocation scheme.