Port Selection for Fluid Antenna Systems

TL;DR

This paper introduces fast port selection algorithms for fluid antenna systems that leverage machine learning and analytical methods to significantly reduce outage probability with minimal port observations.

Contribution

It proposes novel port selection algorithms combining machine learning and analytical approximation, enabling efficient operation with limited port observations.

Findings

Over 10x reduction in outage probability with only 10% port observations.

Significant performance improvements even when observing only a single port.

Algorithms outperform traditional methods in simulation results.

Abstract

Fluid antenna system promises to obtain enormous diversity in the small space of a mobile device by switching the position of the radiating element to the most desirable position from a large number of prescribed locations of the given space. Previous researches have revealed the promising performance of fluid antenna systems if the position with the maximum received signal-to-noise ratio (SNR) is chosen. However, selecting the best position, referred to as port selection, requires a huge number of SNR observations from the ports and may prove to be infeasible. This letter tackles this problem by devising a number of fast port selection algorithms utilizing a combination of machine learning methods and analytical approximation when the system observes only a few ports. Simulation results illustrate that with only 10% of the ports observed, more than an order of magnitude reduction in the outage probability can be achieved. Even in the extreme cases where only one port is observed, considerable performance improvements are possible using the proposed algorithms.