无线传感器网络波束形成优化方法

An Optimal Beamforming Method for Wireless Sensor Networks

大规模无线传感器网络中存在大量冗余节点,为了检测特定目标,所有传感器同时使用不仅增加了电池能量消耗,不利于传感器节点长期工作,而且冗余数据加重传输和处理的负担。以无线麦克风传感器网络为例,研究了基于最小方差无失真响应(Minimum Variance Distortionless Response,MVDR)波束形成器降噪的麦克风子集选择问题。传感器的最佳子集是通过在限制输出噪声功率(或信噪比)的同时最小化传输成本来确定的,使用贪婪算法来选择最佳子集。贪婪算法的性能收敛于模型驱动方法,在动态场景和计算复杂度上都有一定的优势。实验表明,与稀疏MVDR或基于半径的波束形成器相比,所提出的方法能够在显著降低传输成本的情况下保证所需的性能。

Many nodes in large-scale wireless acoustic sensor networks,in order to detect specific targets,the use of all sensors at the same time increases the energy consumption of batteries,which is not conducive to the longterm work of sensor nodes,and the redundant data increases the burden of transmission and processing.Taking wireless microphone sensor networks as an example,this paper considers the selection of microphone subsets based on minimum variance undistorted response(Minimum Variance Distortionless Response,MVDR)beamformer for noise reduction.The optimal subset of sensors is determined by minimizing the transmission cost while limiting the output noise power or SNR.Greedy algorithm is used to select the best subset.The performance of greedy algorithm converges to modeldriven method,and has certain advantages in dynamic scene and computational complexity.Experiments show that compared with sparse MVDR or Radius-Based beamformer,the proposed method can significantly reduce the transmission cost and ensure the required performance.