.png){kind=icon}
.png){kind=icon}
Return
MFFNet: Multi-Path Features Fusion Network for Source Enumeration
.png){kind=icon}
PRE
AI
DOI：10.1109/LCOMM.2021.3139676.png)
.png){kind=icon}
Original request for help.png){kind=icon}
Share.png){kind=icon}
Save
Abstract
En 中文
Source number determination is important for some communication applications. Existing source enumeration methods are sensitive to the number of snapshots, the signal-to-noise ratio and the number of sources. In the letter, we propose a multi-path features fusion network (MFFNet) to enhance the source enumeration accuracy. The inherent multi-scale scheme of Feature Pyramid Networks (FPN) and the path augmentation scheme of Path Aggregation Network (PANet) are exploited, which fuses the spatial feature of the array and the temporal feature of snapshots. The proposed method can extract sufficient information about sources from the original snapshots of array without conventional received sample covariance matrix. Experimental results illustrate that MFFNet outperforms the counterparts on the real data collected in microwave anechoic chamber in terms of detection probability. The source codes are available at https://github.com/fanrongca/MFFNet.
Keywords:
Feature extraction
Signal to noise ratio
Estimation
Training
Silicon
Sensors
Probability distribution
Source enumeration
multi-path features fusion
uniform circle array
Akaike information criterion
gerschgorin disk estimation
Journal
IF:
4.4
Papers:
1.3W
Citations:
2.2W