Hierarchical Fair Spectrum Sharing in CRSNs for Smart Grid Monitoring

Cluster-based cognitive radio sensor networks (CRSNs) are envisioned as a strong driver in smart grid (SG) network development. Added to the sensors' capability to exploit the temporally available licensed spectrum, a hierarchical CRSN allows a better network organization in such a harsh electrical environment. However, in CRSNs, one common control channel (CCC) is usually assumed to exist allowing sensors to exchange control messages before every data transmission. This assumption is not always possible given the licensed signal random activities. Furthermore, SG sensors have to be aware of their neighboring nodes' priorities. Indeed, sensors monitor different applications that have heterogeneous impacts on the SG. Accordingly, in this paper, we design a new solution that allows hierarchical data transmission in CRSNs without being dependent on a CCC and while considering the prioritization in the SG network. We first propose a new clustering algorithm. Then, channels are assigned to the sensors based on local estimates of other nodes' spectrum availability and priority. Given the opportunistic CRSN, we use a Partially Observable Markov Decision Process (POMDP) to allocate channels distributively. Performance evaluation reveals that the spectrum resources are fairly shared between sensors and that our solution outperforms existing work in terms of spectrum utilization.