CrossSense

Abstract

We present CrossSense, a novel system for scaling up WiFi sensing to new environments and larger problems. To reduce the cost of sensing model training data collection, CrossSense employs machine learning to train, off-line, a roaming model that generates from one set of measurements synthetic training samples for each target environment. To scale up to a larger problem size, CrossSense adopts a mixture-of-experts approach where multiple specialized sensing models, or experts, are used to capture the mapping from diverse WiFi inputs to the desired outputs. The experts are trained offline and at runtime the appropriate expert for a given input is automatically chosen. We evaluate CrossSense by applying it to two representative WiFi sensing applications, gait identification and gesture recognition, in controlled single-link environments. We show that CrossSense boosts the accuracy of state-of-the-art WiFi sensing techniques from 20% to over 80% and 90% for gait identification and gesture recognition respectively, delivering consistently good performance-particularly when the problem size is significantly greater than that current approaches can effectively handle. CCS CONCEPTS • Networks → Wireless access networks; • Human-centered computing → Ubiquitous and mobile computing systems and tools;

Low-Cost Outdoor Air Quality Monitoring and Sensor Calibration

“…It may happen that a model trained in one domain (e.g., at a specific time interval, at a specific location, or for specific sensor hardware) is not accurate in another domain and the machine learning model has to be trained separately for each domain. The idea in transfer learning is to mitigate this problem and to make it possible to use commonalities between the different domains to train better and more accurate models, with a smaller amount of data needed per domain [149,153]. Transfer learning can transform a pre-trained model so that it can be applied to a different domain.…”

MetaGanFi

“…Gate-ID [27] explored to accurately recognize individuals in different walking orientations. CrossSense [25] and TransferSense [1] proposed to use transfer learning in achieving gait recognition across different monitoring sites. In contrast, WiDIGR [31] fused multiple receivers' spectrograms as direction-independent features to realize WiFi-based HI system.…”

SenseFi: A library and benchmark on deep-learning-empowered WiFi human sensing

“…As a result, an increasing number of deep-learning models have been developed for WiFi sensing. 33 , 34 These have successfully addressed the shortcomings of traditional statistical learning methods. However, this article mainly focuses on achieving high accuracy on specific sensing tasks by customizing deep neural networks and does not explore the relationship between various deep-learning models and different WiFi-sensing data collected using different devices and CSI tools.…”