分布式接收宽带多目标信号盲检测迭代处理方法

Iterative Blind Detection Algorithm for Wideband Multi-Target Signals in Distributed Receiving Systems

摘要:

随着软硬件技术的飞速发展和宽带接收机的广泛使用,频谱检测向着高瞬时带宽的方向发展,传统基于信道化处理的频谱检测方法存在搜索速度慢、处理效率低下的问题。文章提出了 1种新的分布式接收宽带多目标信号盲检测迭代处理方法,在无须预先知道信号数目及信号频谱位置的情况下,能够实现特定虚警概率多信号盲检测,具备较高的灵活性和稳健性。首先,在对信号特征进行分析的基础上,通过构造线性模型,将分布式接收多目标信号检测转化为线性模型求解问题进行处理；然后,基于贝叶斯多参数联合求解模型,在对未知参数先验分布进行合理假设的基础上,推导了各未知参数变分分布及信号检测门限的解析表达式,采用变分分布软信息迭代的方式实现多传感器信号、多参数联合估计,并利用每次迭代参数估计结果,对信号检测门限进行更新,通过置零操作实现预设虚警概率下的多信号盲检测；最后,通过仿真实验对所提方法性能进行了分析,并与相关方法进行了对比。仿真结果表明,所提方法能够有效利用多路接收信号信息,实现宽带未知多目标信号的盲检测,有效提升短数据下的算法处理效能,与现有方法相比,在接收单元数目较多以及信噪比较低时具有明显优势。

Abstract:

With the rapid development of software and hardware technology and the widespread use of broadband receivers, spectrum detection is moving towards high instantaneous bandwidth. Traditional channelization-based spectrum detection methods have the problems of slow search speed and low processing efficiency. A new iterative blind detection algorithm for wideband multi-target signals in distributed receiving systems is proposed. The proposed method can realize blind detection of multiple signals under specific false alarm probability without knowing the number and spectrum position of signals in advance, having relative high flexibility and robustness. Firstly, on the basis of analyzing the signal characteristics, the multi-signal detection is transformed into a linear model solving problem for processing. Secondly, based on the Bayesian multi-parameter joint solving model and the reasonable assumption of the prior distribution of the unknown parameters, the analytic variation distribution of the unknown parameters and the signal detection threshold are derived, the multi-antenna multi-parameter joint estimation is realized by using the soft information iteration of the variation distribution. In each iteration, the signal detection threshold is updated using the newly obtained parameter estimation results, and the blind detection of multiple signals under the preset false alarm probability is realized through the pruning operation. Finally, the performance of the proposed method is analyzed through simulation experiments and compared with relevant methods. It is shown that the proposed method can effectively utilize the information of multiple received signals, achieve blind detection of wideband unknown multi-target signals, and improve the processing efficiency under short data lengths. Compared with existing methods, the proposed method has significant advantages when the number of receiving units is large and the signal-to-noise ratio is low.