A Novel Machine Learning Aided Antenna Selection Scheme for MIMO Internet of Things

Area of Science:

• Wireless Communication
• Machine Learning
• Signal Processing

Background:

• Multiple-Input Multiple-Output (MIMO) systems are crucial for high-speed wireless communication.
• Internet of Things (IoT) devices require efficient and reliable communication, especially in challenging channel conditions.
• Conventional antenna selection (AS) methods often struggle with correlated channels and large antenna arrays.

Purpose of the Study:

• To propose a novel multi-label convolution neural network (MLCNN)-aided transmit antenna selection (AS) scheme.
• To enhance the performance of end-to-end MIMO IoT communication systems operating under correlated channel conditions.
• To reduce the complexity and training data requirements of AS in multi-antenna systems.

Main Methods:

• Development of a multi-label convolution neural network (MLCNN) tailored for transmit antenna selection.
• Application of the multi-label classification concept to address multi-antenna selection challenges.
• System simulation under correlated large-scale MIMO channel conditions with imperfect Channel State Information (CSI).

Main Results:

• The proposed MLCNN-aided AS scheme significantly reduces the length of training labels required for multi-antenna selection.
• Improved prediction accuracy of the MLCNN model is achieved, even with limited training data under correlated channels.
• The scheme demonstrates near-optimal capacity performance in real-time and exhibits robustness against imperfect CSI.

Conclusions:

• The MLCNN-aided transmit AS scheme offers an effective solution for MIMO IoT systems in correlated channels.
• This approach provides a more efficient and accurate method for antenna selection compared to conventional techniques.
• The proposed method paves the way for enhanced performance in future wireless communication systems.