机会频谱共享下任意输入的优化预编码

Optimal linear precoding for opportunistic spectrum sharing with arbitrary input

认知无线电(cognitive radio,CR)和多输入多输出(multiple input multiple output,MIMO)技术能够有效地提高无线频谱资源的利用效率。而线性预编码技术则是实现这一目的的重要手段。但是目前的预编码算法主要针对服从Gauss分布的输入信号,这一前提假设严重地限制了预编码技术在实际情况中的应用。针对这个问题,该文在分析信息论与检测理论基本关系的基础上,结合特征值分解(singular value decomposition,SVD)与水银注水法(mercury water filling,MWF)的优点,提出了一种适用于输入信号服从任意分布的线性预编码算法,有效提高了线性预编码算法的实用价值。仿真表明该算法优于现有算法。

Multiple input multiple output （MIMO） and cognitive radio （CR） are both effective methods to improve spectrum efficiencies. Linear precoding is needed to realize this goal but most existing algorithms are based on the Gaussian input assumption. This assumption limits the application of linear precoding. A linear precoding algorithm is given here which works for arbitrary input signals by combining the advantages of SVD （singular value decomposition） and MWF （mercury water filling） based on the relations between information theory and detection theory. This algorithm greatly improves the linear precoding. Simulations show that this algorithm is better than existing algorithms.