As a smart combination of cognitive radio networks and wireless sensor networks, recently introduced cognitive radio sensor network (CRSN) poses new challenges to the design of topology maintenance techniques for dynamic primary-user activities. This paper aims to provide a solution to the energy-efficient spectrum-aware CRSN clustering problem. Specifically, we design the clustered structure, establish a network-wide energy consumption model and determine the optimal number of clusters. We then employ the ideas from constrained clustering and propose both a centralized spectrum-aware clustering algorithm and a distributed spectrum-aware clustering (DSAC) protocol. Through extensive simulations, we demonstrate that DSAC can effectively form clusters under a dynamic spectrum-aware constraint. Moreover, DSAC exhibits preferable scalability and stability with its low complexity and quick convergence under dynamic spectrum variation.cognitive radio sensor network, constrained clustering, energy efficient, distributed spectrum-aware clustering
Citation:Zhang H Z, Zhang Z Y, Yuen C. Energy-efficient spectrum-aware clustering for cognitive radio sensor networks. Chin Sci Bull, 2012, 57: 37313739, doi: 10.1007/s11434-012-5254-4As a candidate for next-generation communication networks[1], cognitive radio sensor network (CRSN) [2] has attracted worldwide attention. The dynamic virtue of such network poses new technical challenges on the design of its routing algorithm. Existing cluster-based routing schemes for cognitive radio networks aim to provide joint spectrum and routing decisions, but do not consider CRSN's problem of energy scarcity inherited from traditional wireless sensor networks. Research work [3] has aimed to minimize the number of clusters in a cognitive mesh network while ensuring the connectivity of all users. Distributed clustering for cognitive radio ad hoc networks, in which a clustered structure is established according to the local control channel, has been proposed in [4]. By contrast, clustering schemes for non-cognitive WSNs are developed with the main objective of minimizing energy consumption. However, they are not designed to deal with spectrum-aware communications. In pioneering work [5], an energy-efficient low-energy adaptive clustering hierarchy protocol was proposed. The hybrid, energy-efficient, distributed clustering approach has been developed [6] to cluster ad hoc sensor networks. Recently, centralized clustering schemes, such as k-means clustering [7], have been developed and found to further improve the network performance in terms of energy efficiency.The novelty of our proposed clustered structure is twofold. On the one hand, the structure is aware of the radio environment. On the other hand, the structure should be energy efficient.The basic assumptions are listed as follows: (i) Spectrum-sensing capability: possessing spectrumsensing capability, each CRSN node can correctly determine the available channels at its location.(ii) Spectrum-aware constraint: CRSN nodes that belong