Noise cancellation with adaptive filters on the public radio spectrum

Abstract

Along with the radios invented with the beginning of wireless communication studies, radio communication has been one of the preferred communication methods in every period. The main motivations in its preference are that it does not need high infrastructure installation and maintenance costs, and its mobility is high. Radio usage areas have expanded over time and today it has become the choice of disaster relief agencies, firefighters, and law enforcement. Noise may occur in the public radio spectrum for various reasons such as weather, reflection, refraction, diffraction, absorption, polarization, and scattering which adversely affects the communication quality. Recent studies have shown that noise canceling adaptive filter algorithms can be used to eliminate noise and improve communication quality. The advantages and disadvantages of various adaptive filter algorithms can directly affect their noise canceling performance. In previous studies, it was emphasized that the parameters of the filters affect variables such as iteration rate, convergence performance, computational complexity. In this thesis, the parameters of five adaptive filter algorithms used in noise cancellation are examined and determined. These algorithms are noise canceling performances in public radios are analyzed and their suitability is discussed and compared. The public radio communication systems that have been examined are based on digital systems, which are a newer system compared to the analog radio communication system. The advantages of digital systems can be listed as providing reliable and fast data communication, communication of all public groups over a single infrastructure, and more effective use of frequency resources. Digital systems are actively used throughout the world and in our country. In Europe, TETRA (Terrestrial Trunked Radio) and TETRAPOL are used as Project-25 (APCO-25) in the United States and EDACS (Enhanced Digital Access Communication System) in Canada. In Turkey, on the other hand, since the radio systems used by law enforcement officers used in the public are created independently from each other, even if there are coordination problems, ease of use has been provided with the dissemination efforts in recent years. Aselsan brand 4700 handheld radio model with APCO support was used for the experimental environment and the same transmission frequency was chosen for fair performance analysis in all adaptive filters. Open source HackRf One software defined radio is used as the receiver circuit. Among the filters used, the LMS algorithm is widely used due to its simplicity. However, it has a very slow convergence rate. The NLMS algorithm offers better convergence than LMS and the noise removal performance is better. The RLS algorithm offers a faster convergence scheme. RLS algorithms create less errors than LMS algorithms. However, the computational cost and CPU usage are higher. QRD-RLS and IQRD-RLS algorithms optimize the instability in RLS as they perform QR decomposition directly on the correlation matrix of the input signal compared to RLS.