Performance Evaluation of a Deep Learning-Enhanced Software-Defined Receiver for IRNSS SPS Signals in Vietnam

Abstract The Indian Regional Navigation Satellite System (IRNSS), or NavIC, provides regional satellite positioning services across India and parts of Southeast Asia. In this study, we develop and evaluate a software-defined receiver (SDR) enhanced with deep learning techniques to acquire the IRNSS Standard Positioning Service (SPS) L5-band signal. The SDR architecture incorporates data-driven improvements in acquisition decision-making while retaining compatibility with the IRNSS signal structure as specified in the official ICD. Field experiments were conducted in Hanoi, Vietnam, a location situated at the fringe of NavIC’s primary service area. Signal data were collected using a low-cost RF front-end connected to a rooftop-mounted antenna. Experimental results demonstrate that the proposed SDR is capable of reliably acquiring and tracking up to four IRNSS satellites under nominal conditions. The average C/N0 ranged from 30 to 42 dB-Hz, and successful position solutions were obtained with a horizontal accuracy of approximately 25 meters. Additionally, the deep learning-based acquisition module improved robustness in low-SNR scenarios. This work represents the first implementation of a learning-aided IRNSS receiver validated in Vietnam and offers insights into extending NavIC-based positioning services to broader Southeast Asian regions.