We present MiHazeFree3D , a system that leverages millimeter-wave (mmWave) radar signals to predict 3D bounding boxes of vehicles and pedestrians in real-world traffic scenarios. While current 3D object detection methods rely primarily on cameras and LiDARs, their performance degrades significantly in rain, fog, or poor lighting conditions. Our system exploits mmWave radar’s ability to operate reliably in these challenging conditions, offering a complement to existing sensors without increasing computational costs. The key challenge in using mmWave for 3D detection lies in handling motion-induced errors and the specular reflection of mmWave signals. To address these issues, we developed a deep learning architecture with multiple feature fusion layers and trained it on diverse real-world scenarios. We evaluated MiHazeFree3D using data collected from mmWave radars mounted on the dashboard of an ego-vehicle driving through urban environments. Our results show that MiHazeFree3D detects and bounds both vehicles and pedestrians in tested conditions, including fog and low-light scenarios, highlighting the potential of mmWave radar for 3D object detection in autonomous driving systems.

Gait speed is an important indicator of human health. Monitoring patients' gait speed can help doctors assess the recovery process, but traditional clinician observation fails to track in home scenarios. Compared to vision-based and wearable approaches, radio frequency signals offer an easily deployable and light free solution protecting user privacy in home scenarios. Therefore, we proposed a millimeter-wave (mmWave) system to accurately extract walking periods from collected trials and calculate gait speeds. To evaluate the robustness and reliability of our system and determine the optimal mounting position, we collected data from 5 volunteers with normal walking speeds and imitated various abnormal gait patterns and walking speeds. The results show that the mmWave device mounted near the ground outperforms across all volunteers than the one mounted near the ceiling, achieving an average estimation error of 0.02 m/s in abnormal gait evaluations.

We propose AutoSense, a millimeter-wave (mmWave) wireless signal-based system for predicting 3D bounding boxes of vehicles. While cameras and LiDAR can be adversely affected by challenging weather conditions such as heavy rain, fog, or snow, mmWave signals are less susceptible to these environmental factors, making them more resilient. As a result, AutoSense can complement other sensors for accurate 3D bounding box predictions in all weather conditions.

Object detection plays a pivotal role in various fields, for example, a smart traffic system relies on the detected results for decision-making. However, existing studies predominately utilize optical camera and LiDAR, which exhibit limitations in adverse outdoor environments, such as foggy weather. To address these challenges, millimeter-waves (mmWaves) attract researchers’ attention to detect objects in severe conditions since they can work effectively in low-visibility conditions and overcome small obstacles. Yet, previous mmWave-based works have shown limited performance, such as no shape information for objects. Therefore, we design and implement a two-stage system, mmBox , to accurately predict bounding boxes with depth for vehicles and pedestrians, which first generates heatmaps in different dimensions and then leverages a deep learning model to extract features for predictions. To evaluate the performance of mmBox , we collected real-world mmWave reflections from urban traffic intersections and dense-fog environments. The extensive evaluation metrics show remarkable accuracy and the low latency of our model.

We present CoSense , a system that enables coexistence of networking and sensing on next-generation millimeter-wave (mmWave) picocells for traffic monitoring and pedestrian safety at intersections in all weather conditions. Although existing wireless signal-based object detection systems are available, they suffer from limited resolution and their outputs may not provide sufficient discriminatory information in complex scenes, such as traffic intersections. CoSense proposes using 5G picocells, which operate at mmWave frequency bands and provide higher data rates and higher sensing resolution than traditional wireless technology. However, it is difficult to run sensing applications and data transfer simultaneously on mmWave devices due to potential interference, and using special-purpose sensing hardware can prohibit deployment of sensing applications to a large number of existing and future inexpensive mmWave devices. Additionally, mmWave devices are vulnerable to weak reflectivity and specularity challenges, which may result in loss of information about objects and pedestrians. To overcome these challenges, CoSense design customized deep learning models that not only can recover missing information about the target scene but also enable coexistence of networking and sensing. We evaluate CoSense on diverse data samples captured at traffic intersections and demonstrate that it can detect and locate pedestrians and vehicles, both qualitatively and quantitatively, without significantly affecting the networking throughput.

Millimeter-Wave (mmWave) networks rely on carefully placed small base stations called “picocells” for optimal network performance. However, the process of conducting site surveys to identify suitable picocell locations is both expensive and time-consuming. The current low-cost approaches for indoor surveying are often unsuitable for outdoor environments due to the presence of various environmental factors. To address this issue, we present Theia, a drone-based system that predicts outdoor mmWave Signal Reflection Profiles (SRPs) and facilitates picocell placement for optimal network coverage. The drone platform integrates optical systems and a mmWave transceiver to collect depth images and mmWave SRPs of the environment. These datasets are fed into a machine learning model that maps the depth data to SRPs, allowing SRPs to be predicted at previously unseen parts of the environment. Theia then leverages these predictions to identify optimal picocell locations that maximize network coverage and minimize link outages. We evaluate Theia in three large-scale outdoor environments and demonstrate that the proposed design can generalize the deployment method with a little refinement of the model.

Millimeter-wave technology's unique advantages, in-cluding low-light functionality, cost-effectiveness, and penetration of small objects, make it perfect for outdoor object detection. But traditional methods like likelihood clustering have faced challenges in determining target objects' extent and distance. This work presents mmBox, a two-stage system tailored for precise bounding boxes of vehicles and pedestrians outdoors. We assess mmBox's effectiveness through extensive testing in outdoor street scenes using multiple metrics.

We propose ReSense, a residual network for spectrum sensing high-frequency signals with low-frequency samplers. ReSense first transforms the aliased signal from low-frequency samplers into image-like inputs and uses multiple convolution layers and skip connections to predict the signal’s frequency components to enable spectrum sensing. We evaluate ReSense on the signal dataset with single and double frequencies and achieve 95% and 40% accuracy in detecting modulation type on respective datasets, indicating accurate spectrum sensing.

We propose D3PicoNet, which allows network deployers to quickly complete realistic indoor site surveys at D-band (mmWave) frequency. D3PicoNet models the mmWave reflection profile of a given environment, considering the primary reflecting objects. It then utilizes this model to identify places that optimize the efficiency of the reflectors. D3PicoNet understands an environment and deploys D-band picocells at such locations that picocells provide coverage with Non-Line-of-Sight (NLoS) paths when Line-of-Sight (LoS) is obstructed. The core module of D3PicoNet is a deep learning network that learns the relationship between the visual depth images to the mmWave signal reflection profiles and can accurately predict signal reflection profiles at any unobserved location, which allows D3PicoNet to find the best deployment locations maximizing the coverage and data rate with a minimum number of picocells in an environment. We implement and evaluate D3PicoNet on two buildings with multiple indoor environments. D3PicoNet can adapt to new environments, allowing it to be used in other indoor environments with minimal adjustments.