Improving Spectrum Efficiency via In-Network Computations in Cognitive Radio Sensor Networks

To alleviate the spectrum shortage for sensor networks with tremendous sensors, cognitive radio technology enabling multi-hop opportunistic networking and concurrent transmissions overlaying the primary system suggests an attractive facilitation of large-scale wireless sensor networks (WSNs) and machine-to-machine communications. However, subsequent significant end-to-end delay in large WSN can prohibit practical applications. Leveraging the nature of traffic in sensor networks, we develop in-network computation to reduce requisite transmissions and to accommodate more concurrent transmissions under a given spectrum. Specifically, distributed source coding and broadcasting in wireless communication are exploited to build the computational framework and the achievable network capacity is examined. Furthermore, a greedy networking algorithm is adopted to justify significant improvement on end-to-end delay and further statistical QoS guarantee, while yielding considerable system throughput gain for practical deployment of WSNs. Performance evaluations confirm that we successfully demonstrate communication efficiency from in-network computations and facilitate a new paradigm for spectrum efficient cognitive radio networks, which shall be applicable in general multi-hop wireless networks and spectrum-sharing WSNs.