A Survey on Federated Learning in Human Sensing

TL;DR

This survey systematically chronicles Federated Learning in the Human Sensing domain, where privacy-sensitive sensor data from wearables, audio-visual, and location sources are increasingly abundant. It introduces an eight-dimension assessment and an application-driven taxonomy to evaluate how FL addresses privacy, communication, heterogeneity, and unlabeled data across six sensing domains. By compiling 211 studies, the work exposes where FL is effective and where real-world constraints—such as device heterogeneity and non-IID data—remain challenging, and it proposes five urgent research directions to close these gaps. The paper thereby offers a comprehensive corpus and practical guidance for FL practitioners aiming to deploy privacy-preserving, scalable sensing solutions in real-world settings.

Abstract

Human Sensing, a field that leverages technology to monitor human activities, psycho-physiological states, and interactions with the environment, enhances our understanding of human behavior and drives the development of advanced services that improve overall quality of life. However, its reliance on detailed and often privacy-sensitive data as the basis for its machine learning (ML) models raises significant legal and ethical concerns. The recently proposed ML approach of Federated Learning (FL) promises to alleviate many of these concerns, as it is able to create accurate ML models without sending raw user data to a central server. While FL has demonstrated its usefulness across a variety of areas, such as text prediction and cyber security, its benefits in Human Sensing are under-explored, given the particular challenges in this domain. This survey conducts a comprehensive analysis of the current state-of-the-art studies on FL in Human Sensing, and proposes a taxonomy and an eight-dimensional assessment for FL approaches. Through the eight-dimensional assessment, we then evaluate whether the surveyed studies consider a specific FL-in-Human-Sensing challenge or not. Finally, based on the overall analysis, we discuss open challenges and highlight five research aspects related to FL in Human Sensing that require urgent research attention. Our work provides a comprehensive corpus of FL studies and aims to assist FL practitioners in developing and evaluating solutions that effectively address the real-world complexities of Human Sensing.

Figures (4)

• Figure 1: A comparison between Centralized Learning and Federated Learning
• Figure 3: An Overall Workflow of PRISMA in This Survey
• Figure 4: An overview of the proposed taxonomy in this survey. Physio: Physiological Signal. CSI/RSS: Channel State Information/Received Signal Strength (Indicator). The numbers represent the count of studies within the corresponding domain.
• Figure 5: A summary pie chart. For each section, a histogram illustrates the percentage of studies that covered a specific FL characteristic. Dark color represents Consideration, moderate color represents Not Applicable, and light color represents No Consideration. Priv.: Privacy (and Security). Comm.C.: Communication Cost. Sys.H.: System Heterogeneity. Stat.H.: Statistical Heterogeneity. U.D.U.: Unlabeled Data Usage. Ser.O.: Server-optimized Federated Learning. Cli.O.: Client-optimized Federated Learning.