Machine-Learning-Based Positioning: A Survey and Future Directions

TLDR

Mobile intelligent terminals have expanded location‑based services, yet traditional approaches suffer from high resource demands, limited accuracy, and complex deployment, prompting interest in machine‑learning‑based positioning. This article reviews machine‑learning‑based positioning, offering a retrospective overview of research results. The review classifies location‑based services, summarizes design issues, and outlines key challenges and open questions. These insights point to promising avenues for future research in machine‑learning‑based positioning.

Abstract

Widespread use of mobile intelligent terminals has greatly boosted the application of location-based services over the past decade. However, it is known that traditional location- based services have certain limitations such as high input of manpower/material resources, unsatisfactory positioning accuracy, and complex system usage. To mitigate these issues, machinelearning- based location services are currently receiving a substantial amount of attention from both academia and industry. In this article, we provide a retrospective view of the research results, with a focus on machine-learning-based positioning. In particular, we describe the basic taxonomy of location-based services and summarize the major issues associated with the design of the related systems. Moreover, we outline the key challenges as well as the open issues in this field. These observations then shed light on the possible avenues for future directions.