图共 8个 表共 4
    • 图  1  分层结构的神经网络分类器

      Figure 1. 

    • 图  2  调制识别流程图

      Figure 2. 

    • 图  3  不同${N_{\rm{t}}} \times {N_{\rm{r}}}$下的估计性能

      Figure 3. 

    • 图  4  不同调制方式下${N_{\rm{t}}}$的估计性能

      Figure 4. 

    • 图  5  特征参数T1随SNR的变化曲线

      Figure 5. 

    • 图  6  特征参数T2随SNR的变化曲线

      Figure 6. 

    • 图  7  特征参数T3随SNR的变化曲线

      Figure 7. 

    • 图  8  特征参数T4随SNR的变化曲线

      Figure 8. 

    • 累积量2PSK4PSK8PSK2ASK4ASK2FSKMSK16QAM
      $\left| {{C_{21}}} \right|$11111111
      $\left| {{C_{42}}} \right|$21121.36110.68
      $\left| {{C_{63}}} \right|$1344139.16442.08

      表 1  各调制信号的高阶累积量

    • 调制类型–2 dB0 dB2 dB4 dB6 dB8 dB10 dB
      2PSK80.685.789.594.097.098.8100
      2ASK82.487.392.095.297.099.3100
      2FSK89.093.094.897.3100100100
      4PSK89.493.096.998.298.8100100
      4ASK85.087.090.296.0100100100
      MSK88.592.496.098.399.5100100
      8PSK91.095.097.899.0100100100
      16QAM87.092.395.698.2100100100

      表 2  ${N_{\rm{t}}} \times {N_{\rm{r}}} = 4 \times 5$时的识别率(%)

    • 调制类型-2 dB0 dB2 dB4 dB6 dB8 dB10 dB
      2PSK83.788.095.098.299.4100100
      2ASK84.590.396.098.4100100100
      2FSK91.396.098.299.5100100100
      4PSK91.696.497.5100100100100
      4ASK89.094.498.099.2100100100
      MSK91.095.097.3100100100100
      8PSK93.097.299.3100100100100
      16QAM91.095.398.4100100100100

      表 3  ${N_{\rm{t}}} \times {N_{\rm{r}}} = 4 \times 6$时的识别率(%)

    • 调制类型-2 dB0 dB2 dB4 dB6 dB8 dB10 dB
      2PSK89.093.198.399.0100100100
      2ASK91.695.098.299.0100100100
      2FSK93.498.399.5100100100100
      4PSK95.898.3100100100100100
      4ASK92.097.099.6100100100100
      MSK94.097.698.8100100100100
      8PSK96.899.3100100100100100
      16QAM94.398.0100100100100100

      表 4  ${N_{\rm{t}}} \times {N_{\rm{r}}}{\rm{ = 2}} \times {\rm{5}}$时的识别率(%)