Autonomous vehicles (AVs) rely heavily on cameras and artificial intelligence (AI) to make safe and accurate driving decisions. However, since AI is the core enabling technology, this raises serious cyber threats that hinder the large-scale adoption of AVs. Therefore, it is crucial to analyze the resilience of AV security systems against sophisticated attacks that manipulate camera inputs and deceive AI models. In this paper, we develop camera-camouflaged adversarial attacks and asses their impact on traffic sign recognition (TSR) in AVs. Specifically, the camera-camouflaged attacks are initiated by modifying the texture of a stop sign to fool the AV’s object detection system, thereby affecting the AV actuators. The attacks’ effectiveness is tested using the CARLA AV simulator, and the results show that such attacks can delay the auto-braking response to the stop sign, resulting in potentially catastrophic situations. We conduct extensive experiments under various conditions and on different CARLA maps, confirming that the proposed attacks are effective and robust. Then, two defense strategies are presented to mitigate the effect of camera-camouflaged attacks on the stop sign recognition. The proposed attack and defense methods are applicable to other end-to-end trained autonomous cyber-physical systems.

This paper studies wireless physical-layer impact on anonymity, considering information sniffing over the air as a threat model. Information can leak at the communication physical layer even though encryption is employed. A discrete memoryless channel as system model is used to reflect the fighting between the users and sniffer. Two anonymity metrics are defined and examined based on the probabilistic model. The sniffer is equipped with a classifier and a number of sniffing receivers in the field in order to figure out who is sending packages based on received signal strength (RSS). Four physical-layer anti-sniffing techniques are considered. Computer simulation is used to generate numerical results, considering a wireless uplink transmission system with time-division fashion in a lognormal fading environment. Performances for the best, worst and uniform probability distributions of users' presence are provided. The observations suggest that there could be a potential risk of privacy leakage at the physical layer and also a great potential to guard privacy using physical-layer means.

Signal reuse for multiple purposes is a way to increase spectrum utilization. In this paper, we leverage the WiFi signals of opportunity for the sensing purpose. The spectrograms derived from WiFi downlink (DL) signals reflected from cars are used as fingerprints to efficiently infer car density in parking lots. To achieve this, experimental measurements were conducted in a real outdoor environment to probe the reflected WiFi signals from targets (cars), and the collected datasets are employed for density estimation. The estimator combines hybrid convolutional neural network (CNN) and support vector machine (SVM) for classification, along with least-square estimate (LSE) for interpolation. The probed signals are influenced by many factors, such as the number of WiFi users and data traffic, thereby degrading the estimation accuracy. To address these challenges, we propose an uplink-downlink (UL-DL) WiFi identification and separation technique using the least absolute shrinkage and selection operator (LASSO) technique, without requiring coordination with WiFi access points. Compared to the estimation using a mixture of UL-DL WiFi signals, the simulation results demonstrate that the proposed method achieves significant improvement in estimation accuracy.

The quality of clock signal sources is essential to modern electronic systems, but it is subject to physical limitations, such as phase noise and timing jitter. In some applications a clock distribution circuit or network is required to drive multiple elements, at the cost of an increased level of total phase noise (TPN) (contributed by the phase noise and timing jitter). A windband receiver beamforming is used to study the impact of TPN. An analytical framework along with error vector magnitude (EVM) in closed-form is presented. In particular, this paper demonstrates how the phase noise correlations affect the beam pattern. Verified by the actual beamforming design, an important conclusion is drawn: the EVM decreases as the TPN correlation between channels increases; hence, increasing the clock phase correlation between channels is even more effective than purifying the clock source (suppressing the phase noise and jitter).

Passive sensing leverages existing signals from illuminators of opportunity to perform target localization, detection, and tracking without the need for additional infrastructure. In this paper, we explore the use of WiFi signals of opportunity for estimating car density in parking lots. The aim is to develop an efficient and cost-effective parking occupancy estimation system to alleviate traffic congestion caused by drivers searching for parking spaces. To achieve this, experimental measurements were conducted in a real outdoor environment to collect reflected WiFi signals from targets (cars). These collected signals were then used for car density estimation using a combination of semi-supervised learning convolutional neural network (CNN) and weighted-centroid interpolation techniques, only requiring small size datasets and limited measurements. The proposed method overcomes the limitations of existing data-driven estimators by reducing the reliance on large labeled datasets and computational complexity associated with traditional supervised learning methods. In addition, it provides a cost-effective alternative to the traditional systems that rely on a large number of sensors. Simulations are performed to evaluate the performance of the estimation, and the results demonstrate that our scheme can effectively estimate car densities in the parking lot with reasonable estimation errors.

In this paper, we propose and study a particular use case capable of performing radio-based traffic density estimation for adaptive beam allocation. The proposed scheme explores the synergy between communication and sensing from an Integrated sensing and communication (ISAC) perspective. The traffic den-sity estimation is aided by communication functionality, which involves reusing communication waveforms and utilizing multi-beam forming and sweeping techniques. Meanwhile, the sensing outcomes assist in proactively allocating radio beams. There have been accurate traffic monitoring methods relying on a large number of detectors. However, these traditional techniques have some shortcomings, and it is necessary to explore alternative traffic density estimation approaches. In this regard, we exploit orthogonal frequency division multiplexing (OFDM) communication signals of opportunity reflected from targets (vehicles) to estimate the traffic density of a road section by using Jensen-Shannon (JS) divergence and weighted-centroid interpolation based on a few samples of density scenarios. Then, we present a millimeter-wave (mmWave) adaptive beam allocation protocol based on the traffic density estimation to enhance communication coverage for the vehicular users in the area of interest. The simulation results demonstrate that our traffic density estimation can handle a wide range of targets with a relatively low estimation error. In addition, the analysis of the adaptive beam allocation shows that it effectively improves the quality of service (QoS, in terms of outage probability) of the communication system.

This paper considers a use case of flying base station placement enabled by digital twin (DT), and demonstrates how DT can help reduce the impact of a non-stationary environment on reinforcement learning (RL). RL is able to learn the optimal policy via interacting with a specific environment. However, it has been observed that RL is very sensitive to environment change, mainly because the environment variation disturbs RL training/learning. A possible approach is to execute the RL process in the DT using snapshots of the environment (parameters), and update the parameters at a proper frequency. The DT-RL bundled approach takes advantage of computing resources in the DT, speeds up the process and saves battery energy of the flying base station, and more importantly, mitigates the nonstationary impact on RL. Specifically, the use case is about quickly connecting mobile users with an aerial bass station. The base station is autonomously and optimally placed according to a predefined criterion to connect scattered slow-movement users. Q-learning, a common type of RL, is employed as a solution to the optimization of base station placement. Tailored for this application, a two-stage base station placement algorithm is proposed and evaluated. For the configuration considered in this paper, numerical results suggest that 1) the Q-learning algorithm run solely in the physical space does not work due to intolerable time consuming and optimization divergence, and 2) the proposed scheme can catch up with random slow movement of mobile users, and tolerate certain measurement and model errors. With necessary modification and extension, the proposed framework could be applied to other DT-assisted cyber-physical systems.

This paper studies a use case for quickly connecting users with an aerial bass station (BS) in emergency by leveraging digital twin (DT) and robust reinforcement learning (RL). Scattered communities of users are assigned a limited number of channels, and the flying BS is autonomously and optimally placed according to predefined criteria. Q-learning, a common type of RL, is employed as a solution to optimization of BS placement. Two optimization objectives are considered to maximize the per-user data rate in the worst condition and minimize the total BS transmitted power, respectively. To overcome resource limitations of the aerial BS, the RL training with many iterations is done in the DT virtual space connected to the physical space via an aerial BS. In particular, a practically sound max-min technique is proposed to handle the measurement and prediction uncertainties. It is shown that, even if the models used in DT are imperfect and measurements are inaccurate, nearly-optimal results can be obtained in DT. Compared to RL training 100% in the physical space, a huge number of BS moves can be avoided and a significant amount of time and energy can be saved. The assessment results suggest that model and measurement errors, especially when applying DT, should be seriously considered, and the robust optimization technique has potential to handle the uncertainties.

Aimed at boosting spectrum utilization towards B5G/6G evolution, this paper studies an overlying network that is built by using “dummy” aerial relays and shares the frequency band with a primary network, where the low-cost relays without using sophisticated transmit beamforming generate interference to the primary network. The main objectives are, by using collaborative transmitter precoding, i) to increase Signal-to-Leakage Ratio (SLR) which is the ratio of signal power at the intended receiver to interference power at an unintended receiver, and ii) to reduce the overlying network’s total transmit power. A performance trade-off framework based on robust optimization is proposed to balance these two contradicting objectives, with consideration of the relays’ asynchronicity and channel measurement errors. It is found that the relays’ amplification is a sensitive parameter that can be tuned together with the transmitter precoder in optimization. To overcome the problem of unknown uncertainty set in the max-min optimization, a practical technique based on multiple measurements, semi-analytical robust design, is proposed and tested. Performance evaluation results suggest the proposed overlying network is promising toward the two objectives.

Foreseeing several potential use cases, this paper proposes a node-level edge intelligence (EI) based mobile pipeline computing concept in a Device-to-Device (D2D) communication setup and studies the related issues, where D2D is likely based on millimeter-wave (mmWave) in the 5G mobile communication. Different from traditional distributed computing systems, the proposed opportunistic system employs a chain of wirelessly pipelined resource-limited node-level edge devices (NEDs) on the move to handle real-time computation-intensive multi-stage processing for which current cloud computing technology may not be suitable. The feasibility of such a mobile pipeline can be anticipated as high-speed and low-latency wireless technologies get mature. We present a system model by defining the architecture and basic functions and introducing a possible optimal pipeline finding procedure. As part of the feasibility assessment, the impact of mmWave blockage on the pipeline stability is analyzed and examined for both single-pipeline and concurrent-multiple-pipeline scenarios. Our design and analysis results provide specific insight to guide system design and lay a foundation for further work along this line.

Interferometry is a powerful tool for estimating the incident angle of electromagnetic (EM) waves by calculating the correlation of received signals at different antennas. Motivated by very-long-baseline interfereometry (VLBI) in radio astronomy, an interferometry based sensing scheme is proposed as integrated sensing and communications (ISAC). It reuses the communication signal from base stations (BSs), similarly to passive radars, which improves the sensing precision and spectrum efficiency. Different from the almost-perfect synchronization in VLBI realized by atomic clocks, the synchronization in BSs of cellular commu-nication networks (usually based on GPS signals) could have significant errors. Therefore, algorithms for compensating for synchronization errors in both time and frequency are proposed. Numerical simulations demonstrate that the proposed algorithms can substantially alleviate the synchronization errors.

In Vehicular Ad Hoc Networks (VANETs), the roadside units (RSUs) need to know the number of vehicles in their vicinity to be used in traffic management. However, an attacker may launch a Sybil attack by pretending to be multiple simultaneous vehicles. This attack is severe when a vehicle colludes with others to use valid credentials to authenticate the Sybil vehicles. If RSUs are unable to identify such an attack, they will report wrong number of vehicles to the traffic management center, which may result in disseminating wrong traffic instructions to vehicles. In this paper, we propose a cross-layer scheme to enable the RSUs to identify such Sybil vehicles. Since Sybil vehicles do not exist in their claimed locations, our scheme is based on verifying the vehicles' locations. A challenge packet is sent the vehicle's claimed location using directional antenna to detect the presence of a vehicle. If the vehicle is at the expected location, it should be able to receive the challenge and send back a valid response packet. In order to reduce the overhead and instead of sending challenge packets to all the vehicles all the time, packets are sent only when there is a suspicion of Sybil attack. We also discuss several Sybil attack alarming techniques. The evaluation results demonstrate that our scheme can achieve high detection rate with low probability of false alarm. Additionally, the scheme requires acceptable communication and computation overhead.

Recent works on the universality of Empirical Spectral Density (E.S.D) of Random Matrices have provided a framework to study the asymptotic behavior of random network data. In this paper, the behavior of a wireless campus network with low received base-station transmit power is investigated. Mobile users are organized into 10 clusters of 25 mobile users each and wireless IMT-Advanced Macro channel standard is used for the channel model. The network is simulated using Optimized Network Engineering Tools (OPNET) Modeler 17.5 platform and packet drop data is collected to form the entries of a non-hermittian random matrix. Data is collected for a duration of 400s during downlink transmission and analyzed for three scenarios namely: static mobile users with shadowing vs no-shadowing, and network interference from a stealth jamming source. The result shows that for static nodes with shadowing and no-shadowing, the spectral density of packet drop covariance matrix follows Marcenko and Pastur (MP) distribution and the Ring law accurately, but deviates considerably from MP distribution when the network is jammed using a pulsed jamming source. Outliers are also observed in the Ring-Law together with a shrinkage of eigenvalue distribution towards the center of the inner-circle of the Ring-Law when jammers are introduced into the network.

This letter presents an experimental demonstration of software-defined-radio-based wireless tomography using computer-hosted radio devices called Universal Software Radio Peripheral (USRP). This experimental brief follows our vision and previous theoretical study of wireless tomography that combines wireless communication and RF tomography to provide a novel approach to remote sensing. Automatic data acquisition is performed inside an RF anechoic chamber. Semidefinite relaxation is used for phase retrieval, and the Born iterative method is utilized for imaging the target. Experimental results are presented, validating our vision of wireless tomography.

This paper investigates a multiple input single output (MISO) time reversal system for ultra-wideband (UWB) communication over inter-symbol interference (ISI) channels. Time reversal takes advantage of rich scattering environments to achieve signal focusing, which enables the use of simple receiver structures without sacrificing performance. On-off-keying (OOK) modulation and energy detection (square law) are considered for the purpose of low complexity at the receiver. This research is motivated by the need for high-data-rate wireless network with simple receive nodes. The discrete channel models and bit error rate (BER) formulas for the energy detector receiver over ISI channels are derived. Performance is evaluated based on measured data, considering practical signal waveforms. One reason to use our own measured data is that there is no proper UWB channel model for antenna array related study. Numerical results suggest that the proposed MISO time reversal system with extremely simple receiver is promising to support high data rate in severe multipath environments for robust communication in the UWB band.

This paper presents a reduced-complexity time reversal technique for ultra-wideband (UWB) communications. Time reversal takes advantage of rich scattering environments to achieve signal focusing via transmitter-side processing, which enables the use of simple receivers. The goal of this paper is to demonstrate a UWB time reversal system architecture based on experimental results and practical pulse waveform, taking into account some practical constraints, and to show feasibility of UWB time reversal. Pre-decorrelating in addition to time reversal processing is considered for a downlink multiuser configuration. Multiple transmit antennas are employed to improve the performance.

This paper first introduces per-path pulse distortion in multiuser detection for ultra-wideband communications. A new generalized RAKE structure that estimates and compensates for the pulse distortion is used. An finite-impulse response filter representation of the per-path impulse response is used and estimated. The new structure greatly improves the system performance. With four users considered in our simulations in a high-rise building environment, it is found that the average performance of the generalized RAKE using minimum mean-square error (MMSE) detection is improved over the conventional RAKE by 1.8 dB. Both synchronous and asynchronous transmission schemes for decorrelating detector and MMSE detector are considered.

The autocorrelation demodulation (ACD) has potential for use in medium-range low-cost UWB communications, however, its weak performance is a bottle neck. To improve ACD's performance, it is proposed to combine the multiple-symbol detection with ACD to form a multiple-symbol-based ACD receiver. In contrast to ordinary ACD scheme that uses symbol-by-symbol decision, the proposed structure uses block-based multiple-symbol joint decision. Performance of the new scheme is evaluated analytically for arbitrary block sizes employing Gaussian approximation approach. The accuracy of the method is verified by computer simulation. The evaluation shows that the proposed scheme offers several dB improvement in performances with only a moderate increase in complexity.