Distributed clustering and interference avoidance in cognitive femtocell networks

The concept of a cellular network overlaid with Femtocell Access Points (FAPs) has emerged as a new wireless architecture that ensures high data rates and reliable coverage for indoor users. However, the spatially random installation of FAPs and lack of coordination between the FAPs and cellular tiers results in co-channel interference which limit the system performance. This paper designs an interference avoidance scheme for cognitive femtocell networks based on a hierarchical architecture that exploits clustering. Each FAP senses the licensed spectrum to find “blank spaces”, measures its mutual similarity with only its neighboring FAPs, and forms simple messages which serve as incentives for active FAPs to partition into coalition clusters. Then, the chosen cluster head coordinates its members' transmissions. Cluster formation is updated on large time scales and is not susceptible to the instantaneous channel variations, thereby reducing the overhead in real-time communication. Our proposed scheme not only minimizes cross-tier interference, but also maximizes indoor data rates. Numerical results show that the proposed distributed scheme outperforms (centralized) spectral clustering algorithm for different number of FAPs.