Eye motion tracking plays a vital role in many applications such as human-computer interaction (HCI), virtual reality, and disease detection.Camera-based eye tracking, albeit accurate and easy to use, may raise privacy concerns and appear to be unreliable in poor lighting conditions.In this paper, we present RadEye, a radar system capable of detecting fine-grained human eye motions from a distance.RadEye is realized through an integrated hardware and software design.It customizes a sub-6GHz FMCW radar so as to detect millimeter-level eye movement while extending its detection range using low frequency.It further employs a deep neural network (DNN) to refine the detection accuracy through camera-guided supervisory training.We have built a prototype of RadEye.Extensive experimental results show that it achieves 90% accuracy when detecting human eye rotation directions (up, down, left, and right) in various scenarios.

Sub-6GHz radio sensing offers several compelling advantages, such as resilience to poor lighting conditions, privacy preservation, and the ability to see through walls. However, in indoor environments, the sub-6GHz ISM spectrum is heavily occupied by WiFi devices, leaving little available spectrum for sensing purposes. In this paper, we introduce SiWiS, a new approach to integrate radio sensing capabilities into individual WiFi devices for fine-grained human activity detection. SiWiS comprises two main components: (i) a new hardware component that can be easily installed on an off-the-shelf WiFi device, and (ii) a dual-branch deep neural network (DNN) optimized for concurrent human mask segmentation and pose estimation. We have built a prototype of SiWiS and installed it on a commercial WiFi router for evaluation. Extensive experimental results demonstrate a significant performance improvement over WiFi channel state information (CSI) based sensing methods. More importantly, zero-shot experiments confirm that SiWiS can be directly transferred to unseen real-world environments.

Beam selection is a fundamental problem in millimeter-wave (mmWave) communication systems. Yet, most existing beam selection techniques focus on the exploitation of spatial channel features to reduce their airtime overhead in stationary mmWave networks. In this article, we exploit the temporal correlation of wireless channels to facilitate beam selection in mobile mmWave networks. Specifically, we present a temporal beam prediction (TBP) scheme for a mobile mmWave device to predict its future beam direction based on its history beam selection profile. TBP has two challenges in its design: 1) nonuniform history data samples due to the bursty nature of data traffic and 2) nonsmooth beam angles over time due to the multipath effect of channels and the imperfect radiation pattern of phased-array antennas. TBP addresses these two challenges by employing a new mobility-aware LSTM model that takes data timestamp for its training, together with an adversarial learning model to exploit user-independent features for beam steering. We have evaluated TBP through over-the-air (OTA) experiments on a 60-GHz mmWave testbed. Experimental results show that the average prediction error of TBP is less than 7° and that TBP improves the throughput by 60% in representative mmWave networks.

The use of mobile phones while driving is a major source of distraction for vehicle drivers and has resulted in a large number of car accidents. While surveillance cameras can be used to detect the violation of phone use, they do not work well in some scenarios (e.g., darkness and blockage) and may raise privacy concerns. In this paper, we present PhoLoc, a roadside device to detect the violation of phone use in personal vehicles using the cellular signals emitted by cellphones. PhoLoc is equipped with two sensors: a multi-antenna radio receiver and a low-cost lidar. It jointly processes the multimodal data from the two sensors to estimate the relative location of a phone in a vehicle. The enabler of PhoLoc is a new near-field localization scheme, which is capable of estimating the location of a moving phone at a specific time moment by overhearing its cellular signals. We have built a prototype of PhoLoc and evaluated its performance in realistic scenarios. Experimental results show that PhoLoc achieves 4.2% false positive rate and 13.8% false negative rate in the detection of phone call violation.

We study the data packet transmission problem (mmDPT) in dense cell-free millimeter wave (mmWave) networks, i.e., users sending data packet requests to access points (APs) via uplinks and APs transmitting requested data packets to users via downlinks. Our objective is to minimize the average delay in the system due to APs’ limited service capacity and unreliable wireless channels between APs and users. This problem can be formulated as a restless multi-armed bandits problem with fairness constraint (RMAB-F). Since finding the optimal policy forRMAB-Fis intractable, existing learning algorithms are computationally expensive and not suitable for practical dynamic dense mmWave networks. In this paper, we propose a structured reinforcement learning (RL) solution formmDPTby exploiting the inherent structure encoded inRMAB-F. To achieve this, we first design a low-complexity and provably asymptotically optimal index policy forRMAB-F. Then, we leverage this structure information to develop a structured RL algorithm calledmmDPT-TS, which provably achieves an$\tilde {\mathcal {O}}(\sqrt {T})$Bayesian regret. More importantly,mmDPT-TSis computation-efficient and thus amenable to practical implementation, as it fully exploits the structure of index policy for making decisions. Extensive emulation based on data collected in realistic mmWave networks demonstrate significant gains ofmmDPT-TSover existing approaches.

UHF RFID tags have been widely used for contactless inventory and tracking applications. One fundamental problem with RFID readers is their limited tag reading rate. Existing RFID readers (e.g., Impinj Speedway) can read about 35 tags per second in a read zone, which is far from enough for many applications. In this paper, we present the first-of-its-kind RFID reader (mReader), which borrows the idea of multi-user MIMO (MU-MIMO) from cellular networks to enable concurrent multi-tag reading in passive RFID systems. mReader is equipped with multiple antennas for implicit beamforming in downlink transmissions. It is enabled by three key techniques: uplink collision recovery, transition-based channel estimation, and zero-overhead channel calibration. In addition, mReader employs a Q-value adaptation algorithm for medium access control to maximize its tag reading rate. We have built a prototype of mReader on USRP X310 and demonstrated for the first time that a two-antenna reader can read two commercial off-the-shelf (COTS) tags simultaneously. Numerical results further show that mReader can improve the tag reading rate by 45% compared to existing RFID readers.

Federated Learning (FL) is a promising technique to enhance the safety and efficiency of intelligent transportation systems. While FL has been extensively studied, the communication and networking challenges related to the operations of FL in dynamic yet dense vehicular networks remain under-explored. Limited storage and communication capacities of individual vehicles throttle the timely training of an FL model in distributed vehicular networks. In this paper, we present a communication framework for FL (CF4FL) in transportation systems. CF4FL aims to accelerate the convergence of FL training process through the innovation of two complementary networking components: (i) a deadline-driven vehicle scheduler (DDVS), and (ii) a concurrent vehicle polling scheme (CVPS). DDVS identifies a subset of vehicles for local model training in each iteration of FL, with the aim of minimizing data loss while respecting the deadline constraints derived from vehicles’ storage, computation, and energy budgets. CVPS takes advantage of multiple antennas on an edge server to enable concurrent local model transmissions in dynamic vehicular networks, thereby reducing the airtime overhead of each FL iteration. We have evaluated CF4FL through a blend of experimentation and simulation. Trace-driven simulation shows that, compared to existing scheduling and transmission schemes, CF4FL reduces the convergence time of FL training by 39%.

LoRa has emerged as a key wireless communication technology for a gateway to provide geographically-distributed IoT devices with low-rate, long-range connections. In this paper, we present MaLoRaGW, the first-of-its-kind Multi-antenna LoRa GateWay that enables multi-user MIMO (MU-MIMO) LoRa communications in both uplink and downlink. MaLoRaGW was inspired by the success of MU-MIMO in cellular and Wi-Fi networks. The key component of MaLoRaGW is a joint baseband PHY design for uplink packet detection and downlink beamforming. Its innovation lies in three modules: spatial signal projection, accurate channel estimation, and implicit beamforming, all of which reside only in a LoRa gateway and require no modification on LoRa client devices. We have built a prototype of two-antenna MaLoRaGW on a USRP device and extensively evaluated its performance with commercial LoRa dongles in three scenarios: lab, office building, and university campus. Our experimental results show that, compared to the state-of-the-art, the two-antenna MaLoRaGW increases uplink throughput by 10% and downlink throughput by 95%.

Wireless Internet of Things (IoT) applications have penetrated every aspect of our society and become increasingly important in smart homes, smart cities, and smart hospitals. However, many WiFi-based IoT devices (e.g., light switches, door/window open alert sensors, and Google Home) do not have input interfaces such as keypad or touchscreen due to their limits in physical size, power consumption, and/or manufacturing cost, making it inconvenient and onerous for end users to authenticate those IoT devices for wireless Internet access. In this article, we present AuthIoT, a learning-based authentication scheme for wireless IoT devices without input interfaces. The key component of AuthIoT is a channel state information (CSI)-based character classification algorithm for a WiFi access point (AP), which recognizes the passcode from an IoT device when an end user holds it in hand and writes the passcode over the air. AuthIoT has two salient features: 1) it is transferable for cross-environment applications and 2) it works in more realistic scenarios where AP is equipped with nonlinear antenna array. We have built a prototype of AuthIoT and evaluated its performance on two testbeds: 1) Intel 5300 WiFi card with three linear antennas and 2) USRP N310 with four nonlinear (square-shaped) antennas. The experimental results show that AuthIoT achieves 84% and 83% recognition accuracy on the two testbeds.

Current data-driven intelligent transportation systems are mainly reliant on IEEE 802.11p to collect and exchange information. Despite promising performance of IEEE 802.11p in providing low-latency communications, it is still vulnerable to jamming attacks due to the lack of a PHY-layer countermeasure technique in practice. In this paper, we propose JammingBird, a novel receiver design that tolerates strong constant jamming attacks. The enablers of JammingBird are two MIMO-based techniques: Jamming-resistant synchronizer and jamming suppressor. Collectively, these two new modules are able to detect, synchronize, and recover desired signals under jamming attacks, regardless of the PHY-layer technology employed by the jammers. We have implemented JammingBird on a vehicular testbed and conducted extensive experiments to evaluate its performance in three common vehicular scenarios: Parking lots (0~15 mph), local traffic areas (25~45 mph), and highways (60~70 mph). In our experiments, while the jamming attacks degrade the throughput of conventional 802.11p-based receivers by 86.7%, JammingBird maintains 83.0% of the throughput on average. Experimental results also show that JammingBird tolerates the jamming signals with 25 dB stronger power than the desired signals.