Unlike static biometrics (e.g., faces and fingerprints), dynamic behavioral biometrics are believed to be more difficult to replicate. This paper investigates the security of behavioral biometrics considering the advancements in robotics and AI, particularly as humanoid robots, like Tesla Optimus, are expected to be mass-produced in the coming years. We find that general robotic arms have already gained the capability to reproduce human hand motion trajectories. However, using robots to replicate a user’s behavioral biometrics for attacks remains under-explored due to two long-standing challenges: 1) how to obtain the user’s complex behavioral biometrics through practical eavesdropping (not just trajectories); 2) how to replicate the user’s behavioral kinematics based on the eavesdropped data using a real robot. This work is the first to comprehensively address the two challenges. We develop the point-wise GAN-based Robot Replay Attack (GANRRA) to demonstrate a practical human behavioral replay attack using a hidden camera and a physical robot. GANRRA utilizes a hidden camera to eavesdrop on the user’s hand motions and employs a generative adversarial network to reconstruct the motion data, addressing the sensor discrepancies between the legitimate sensor and the hidden camera and maximizing the behavioral feature similarities. The reconstructed motion data is converted into velocity commands for a robot to execute point by point, replicating both hand movement trajectories and behavioral biometric features. For experiments, we implement an in-air signature system using two existing hand-tracking systems and fool them using a robotic arm attached with a fake hand. Results show that GANRRA reproduces in-air signatures with a 73.1% success rate. To address such robot-relay threats, a novel defense mechanism based on multi-joint behaviors is proposed.

Teleoperated robotics, which translates human behavior into robotic actions, remains a critical area of modern robotics. Although autonomous systems have advanced rapidly, they still struggle in complex and unstructured environments, making human-in-the-loop control indispensable for many real-world tasks. Teleoperation platforms commonly rely on motion-tracking technologies to capture detailed operator behavior, which is subsequently converted into robot control commands. However, these rich behavioral signals can also encode operator-specific biometrics, posing privacy risks such as user re-identification. While prior work shows that behavioral biometrics can be leveraged for reliable authentication, privacy leakage in teleoperation-centric motion streams has received comparatively less attention. To address this gap, we introduce a disentangled representation-learning framework based on a Variational Autoencoder (VAE) to suppress identity-revealing cues while retaining task-relevant motion patterns. We evaluate the proposed approach offline on reconstructed trajectories collected from a tele-robotic prototype, where multiple users perform a set of manipulation tasks. Our results demonstrate a substantial reduction in re-identification risk and a favorable privacy–utility trade-off in terms of task utility. More broadly, our findings highlight the need for robust privacy protections in future robotic teleoperation systems.

This paper explores the growing need for task-oriented communications in warehouse logistics, where traditional communication Key Performance Indicators (KPIs)-such as latency, reliability, and throughput-often do not fully meet task requirements. As the complexity of data flow management in large-scale device networks increases, there is also a pressing need for innovative cross-system designs that balance data compression, communication, and computation. To address these chal-lenges, we propose a task-oriented, edge-assisted framework for cooperative data compression, communication, and computing in Unmanned Ground Vehicle (UGV)-enhanced warehouse logistics. In this framework, two UGVs collaborate to transport cargo, with control functions-navigation for the front UGV and following/conveyance for the rear UGV-offloaded to the edge server to accommodate their limited on-board computing resources. We develop a Deep Reinforcement Learning (DRL)-based two-stage point cloud data compression algorithm that dynamically and collaboratively adjusts compression ratios according to task requirements, significantly reducing communication overhead. System-level simulations of our UGV logistics prototype demon-strate the framework's effectiveness and its potential for swift real-world implementation.

A key component of dyadic spoken interactions is the contextually relevant non-verbal gestures, such as head movements that reflect a listener’s response to the interlocutor’s speech. Although significant progress has been made in the context of generating co-speech gestures, generating listener’s response has remained a challenge. We introduce the task of generating continuous head motion response of a listener in response to the speaker’s speech in real time. To this end, we propose a graph-based end-to-end crossmodal model that takes interlocutor’s speech audio as input and directly generates head pose angles (roll, pitch, yaw) of the listener in real time. Different from previous work, our approach is completely data-driven, does not require manual annotations or oversimplify head motion to merely nods and shakes. Extensive evaluation on the dyadic interaction sessions on the IEMOCAP dataset shows that our model produces a low overall error (4.5 degrees) and a high frame rate, thereby indicating its deployability in real-world human-robot interaction systems. Our code is available at - https://github. com/bigzen/Active-Listener

The integration of humanoid and animal-shaped robots into specialized domains, such as healthcare, multiterrain operations, and psychotherapy, necessitates a deep understanding of proxemics-the study of spatial behavior that governs effective human-robot interactions. Unlike traditional robots in manufacturing or logistics, these robots must navigate complex human environments where maintaining appropriate physical and psychological distances is crucial for seamless interaction. This study explores the application of proxemics in human-robot interactions, focusing specifically on quadruped robots, which present unique challenges and opportunities due to their lifelike movement and form. Utilizing a motion capture system, we examine how different interaction postures of a canine robot influence human participants' proxemic behavior in dynamic scenarios. By capturing and analyzing position and orientation data, this research aims to identify key factors that affect proxemic distances and inform the design of socially acceptable robots. The findings underscore the importance of adhering to human psychological and physical distancing norms in robot design, ensuring that autonomous systems can coexist harmoniously with humans.

Tele-operated robots rely on real-time user behavior mapping for remote tasks, but ensuring secure authentication remains a challenge. Traditional methods, such as passwords and static biometrics, are vulnerable to spoofing and replay attacks, particularly in high-stakes, continuous interactions. This paper presents a novel anti-spoofing and anti-replay authentication approach that leverages distinctive user behavioral features extracted from haptic feedback during human–robot interactions. To evaluate our authentication approach, we collected a time-series force feedback dataset from 15 participants performing seven distinct tasks. We then developed a transformer-based deep learning model to extract temporal features from the haptic signals. By analyzing user-specific force dynamics, our method achieves over 90% accuracy in both user identification and task classification, demonstrating its potential for enhancing access control and identity assurance in tele-robotic systems.

This paper proposes a task-oriented co-design framework that integrates communication, computing, and control to address the key challenges of bandwidth limitations, noise interference, and latency in mission-critical industrial Cyber-Physical Systems (CPS). To improve communication efficiency and robustness, we design a task-oriented Joint Source-Channel Coding (JSCC) using Information Bottleneck (IB) to enhance data transmission efficiency by prioritizing task-specific information. To mitigate the perceived End-to-End (E2E) delays, we develop a Delay-Aware Trajectory-Guided Control Prediction (DTCP) strategy that integrates trajectory planning with control prediction, predicting commands based on E2E delay. Moreover, the DTCP is co-designed with task-oriented JSCC, focusing on transmitting task-specific information for timely and reliable autonomous driving. Experimental results in the CARLA simulator demonstrate that, under an E2E delay of 1 second (20 time slots), the proposed framework achieves a driving score of 48.12, which is 31.59 points higher than using Better Portable Graphics (BPG) while reducing bandwidth usage by 99.19%.

Existing communication systems aim to reconstruct the information at the receiver side, and are known as reconstruction-oriented communications. This approach often falls short in meeting the real-time, task-specific demands of modern AI-driven applications such as autonomous driving and semantic segmentation. As a new design principle, task-oriented communications have been developed. However, it typically requires joint optimization of encoder, decoder, and modified inference neural networks, resulting in extensive cross-system redesigns and compatibility issues. This paper proposes a novel communication framework that aligns reconstruction-oriented and task-oriented communications for edge intelligence. The idea is to extend the Information Bottleneck (IB) theory to optimize data transmission by minimizing task-relevant loss function, while maintaining the structure of the original data by an information reshaper. Such an approach integrates taskoriented communications with reconstruction-oriented communications, where a variational approach is designed to handle the intractability of mutual information in high-dimensional neural network features. We also introduce a joint source-channel coding (JSCC) modulation scheme compatible with classical modulation techniques, enabling the deployment of AI technologies within existing digital infrastructures. The proposed framework is particularly effective in edge-based autonomous driving scenarios. Our evaluation in the Car Learning to Act (CARLA) simulator demonstrates that the proposed framework significantly reduces bits per service by 99.19% compared to existing methods, such as JPEG, JPEG2000, and BPG, without compromising the effectiveness of task execution.

We introduce HARPER, a novel dataset for 3D body pose estimation and forecasting in dyadic interactions between users and Spot, the quadruped robot manufactured by Boston Dynamics. The key-novelty of HARPER is its focus on the robot’s perspective, i.e., on the data captured by the robot’s sensors. This makes 3D body pose analysis challenging, as being close to the ground results in only partial captures of humans. The scenario underlying HARPER includes 15 actions, of which 10 involve physical contact between the robot and users. The corpus contains recordings not only from Spot’s built-in stereo cameras but also from a 6-camera OptiTrack system, with all recordings synchronized. This setup leads to ground-truth skeletal representations with a precision of less than a millimeter. Additionally, the corpus includes reproducible benchmarks for 3D Human Pose Estimation, Human Pose Forecasting, and Collision Prediction, all based on publicly available baseline approaches. This enables future HARPER users to rigorously compare their results with those provided in this work.

How a robot moves is among the first things an observer notices when they encounter a robot. While considerable research has investigated the perception of robot body language, no studies yet, to our knowledge, have explored the social effects of how a robot moves through space (its gait) on people’s first impressions of a robot. To this end, here we performed two complementary experiments online (n = 98) and in-person (n = 26), with the objective of determining the extent to which a quadruped robot’s gait influences a) what animal people perceived it to be; and b) its social attributes in terms of warmth, competence, causing discomfort and zoomorphism. Results differed depending on whether participation was online or in-person, with gaits influencing participants’ perception more markedly when they encountered the robot in-person. Online, most participants saw the robot as a dog for every gait except one, while in-person participants reported more varied responses. Participants in both studies rated the more active, "bouncy" gait as warmer and less discomforting. In-person participants also consistently rated all gaits as warmer, more competent, less discomforting and more zoomorphic than did online participants. The study supports findings that in-person exposure and embodiment affect a robot’s social perception and further suggests that gait may have a limited effect as well.

Future robots will be very advanced with high flexibility and accurate control performance. They will have the ability to mimic human behaviours or even perform better, which raises the significant risk of robot attack. In this work, we study the robot mimic attack on the current keystroke-dynamic user authentication system. Specifically, we proposed a robot mimicry attack framework for keystroke-dynamics systems. We collected keyboard logging data and acoustical signal data from real users and extracted the timing pattern of keystrokes to understand victim's behaviour for robot imitation attacks. Furthermore, we develop a deep Q-Network (DQN) algorithm to control the velocity of robot which is one of the key challenges of forging the human typing timing features. We tested and evaluated our approach on the real-life robotic testbed. We presented our results considering user identification and user authentication performance. We achieved a 90.3% user identification accuracy with genuine keyboard logging data samples and 89.6% accuracy with robot-forged data samples. Furthermore, we achieved 11.1%, and 36.6% EER for user authentication performance with zero-effort attack, and robot mimicry attack, respectively.

Motion-controlled robotic arms allow a user to interact with a remote real world without physically reaching it. By connecting cyberspace to the physical world, such interactive teleoperations are promising to improve remote education, virtual social interactions, and online participatory activities. In this work, we build up a motion-controlled robotic arm framework comprising a robotic arm end and a user end, which are connected via a network and responsible for manipulator control and motion capture, respectively. To protect the system access, we propose to verify who is controlling the robotic arm by examining the robotic arm’s behavior, which adds a second security layer in addition to the system login credentials. We show that a robotic arm’s motion inherits its human controller’s behavioral biometric in interactive control scenarios. By extracting the angle readings of the robotic arm’s all joints, the proposed user authentication approach reconstructs the robotic arm’s end-effector movement trajectory that follows the user’s hand. Furthermore, we derive the unique robotic motion features to capture the user’s behavioral biometric embedded in the robot motions and develop learning-based algorithms to verify the robotic arm user to be one of the enrolled users or a nonuser. Extensive experiments show that our system achieves 94% accuracy to distinguish users while preventing user identity spoofing attacks with 95% accuracy.

Motion-controlled robots allow a user to interact with a remote real world without physically reaching it. By connecting cyberspace to the physical world, such interactive teleoperations are promising to improve remote education, virtual social interactions and online participatory activities. This work builds up a motion-controlled robotic arm framework and proposes to verify who is controlling the robotic arm by examining the robotic arm's behavior. We show that a robotic arm's motion inherits its human controller's behavioral biometric in interactive control scenarios. Furthermore, we derive the unique robotic motion features to capture the user's behavioral biometric embedded in the robot motions and develop learning-based algorithms to verify the robotic arm user. Extensive experiments show that our system achieves high accuracy to distinguish users while using the robot's behaviors.

Fueled by ubiquitous connectivity, packets are expected to be timely updated to the controller of interest in real-time control systems. Recently, the Age of Information (AoI) becomes a popular metric to quantify the packet importance, which improves the efficiency of communication resource utilization. However, it is analyzed only from the information freshness perspective, losing sight of considering feature data. In light of this, we first establish a feature-oriented teleoperation framework to quantify the packet importance, derived from the information bottleneck principle. Under this framework, a packet management method is proposed to increase the average feature quantity of the receiver. Finally, we build a prototype to deploy the proposed method, and the results show superiority in reducing both the communication traffic and the control error.

In industrial internet of things (IIoT), ultra-reliable and low-latency communication (URLLC) is proposed to guarantee the requirement of real-time wireless control systems in worst case, so as to maintain the system working in all cases. However, it is extremely challenging to maintain URLLC throughout the whole control process due to the scarcity of wireless resource. This paper develops an autonomous device-to-device (D2D) communication scheme by jointly considering reliability in URLLC and control requirement. In the proposed scheme, we consider the actual control requirement, i.e., control convergence rate, into communication design, where we find that it can be converted into a constraint on communication reliability. Then, the communication reliability constraint comes from control aspect, instead of URLLC, which leads to that the system does not need to guarantee worst case in URLLC. Second, the sensors autonomously decide whether to be activated with optimal probabilities to participate in the control process, which can maintain the communication reliability requirement with significantly less resource consumption. Simulation results show remarkable performance gain of our method. For instance, compared with fixed activation probability 40% only considering URLLC, the average power consumption of the proposed method can be reduced by at most about 100%.

In real-time feedback control systems, control performance, e.g., control cost and tracking error, is significantly affected by information freshness, which in turn relies heavily on the design of the feedback update policy. In this paper, we study the update policy for real-time feedback control systems. We first discuss the relationship between the age of information (AOI) and control performance, where AOI represents the level of "dissatisfaction" for information staleness. We find that minimizing AOI is not always equivalent to optimizing the control performance. Then, a new metric, called the value of information (VOI), is proposed to evaluate the timeliness of system update by linking AOI to the decay rate of the control system. By maximizing VOI, we design a new update policy, called the α - wait, which has superiorities in improving both control performance and communication cost. Finally, simulation results verify our method.

In real-time feedback control systems, the freshness of the packet is crucial to control performance, where packet management is vital to keep data fresh. Recently, age of information (AOI) has been used to measure the freshness of the update information, where minimizing AOI becomes popular in system designs. In this paper, we find that minimizing AOI is not always equivalent to maximizing the control performance. In particular, we define a metric, called the age of stale information (AOSI), to link the instability of the control system to AOI. By minimizing AOSI, we can maximize the control performance, and also reduce the communication cost.

Ultra-reliable and low-latency communication (URLLC) is promising to enable real-time wireless control systems for tactile internet. In such a system, it is difficult to maintain extremely high quality-of-service (QoS) in URLLC for real-time control. In this paper, we develop a probability-based device-to-device (D2D) scheme to deal with this issue, where communication and control are jointly considered. In our scheme, the transmitters autonomously decide whether to be active to participate in the control process of the receiver based on a certain probability, which can significant reduce the interacting communication latency between them, lower the transmission power consumption, and improve communication reliability. Compared with traditional D2D transmission method, simulation results show remarkable performance gain of our method.

Co-channel interference is one of the most important issue in cache-enabled device-to-device (D2D) wireless networks. In this paper, we propose a D2D link scheduling algorithm to manage the interference, called Max-SIR scheduling algorithm. It consists of two steps, i.e., link scheduling and link removal. In the algorithm, D2D links are scheduled with the maximum signal-to-interference ratios (SIRs) considering the co-channel interference in a cell. Simulation results show that the proposed algorithm outperforms the existing ones in terms of the system throughput and the number of scheduled D2D links.

In this paper, we consider cache-enabled device-to- device (D2D) wireless networks and propose a content-centric link scheduling method, called CTLinQ, to maximize the number of overall D2D links that can be simultaneously activated. Then, interference among different D2D links can be reduced and the quality-of-service (QoS) of each activated link can be guaranteed. Simulation results show that the proposed method outperforms the existing ones in terms of power consumption, the number of activated links, and overall system throughput in particular in high signal-to- interference-plus-noise ratio (SINR) region.

In device-to-device (D2D) networks, co-channel interference is one of the main reasons that causes high power consumption, which further reduces the life time of mobile devices. In this paper, we adopt the superposition coding between macro and D2D users, which is expected to effectively eliminate the co-channel interference. In particular, we develop two power allocation methods to minimize the power consumption in cooperative and non-cooperative modes, respectively. Then, we use mode selection to obtain the minimum overall power consumption of the whole system. In power allocation, we model the average power consumption as a function of channel gain, power allocation factor, transmission rate, and noise power. Then, we obtain the close-form solution. Our results indicate that the proposed method outperforms the conventional non-cooperative methods in terms of power consumption, outage probability, and energy efficiency.

In cognitive radio networks, the channel gain between primary transceivers, namely, primary channel gain, is crucial for a cognitive transmitter (CT) to control the transmit power and achieve spectrum sharing. Conventionally, the primary channel gain is estimated in the primary system, and thus unavailable at the CT. To deal with the issue, two estimators are proposed by enabling the CT to sense primary signals. In particular, by adopting the maximum likelihood (ML) criterion to analyze the received primary signal, an ML estimator is first developed. To reduce the computational complexity of the ML estimator, a median-based (MB) estimator is then proposed. By comparing the ML estimator and the MB estimator from the aspects of the computational complexity as well as the estimation accuracy, both advantages and disadvantages of two estimators are revealed. Simulation results show that the estimation errors of both estimators can be as small as 0.015. Meanwhile, the ML estimator outperforms the MB estimator in terms of the estimation accuracy if the sensed primary signal at the CT is weak. Otherwise, the MB estimator is superior to the ML estimator from the aspects of both the computational complexity and the estimation accuracy.

In this paper, we propose a new estimation method to obtain the cross-channel gain, which avoids the severe interference to the primary receiver (PR) in existing relay-assisted estimation methods. In our method, we let the cognitive transmitter add a time delay when it conducts the full-duplex amplify-and-forward relaying. This forces the time-difference-of-arrival (TDOA) between the direct and relay signals to be large enough rather than randomly large or small. Then we develop our estimation method only in the large TDOA case and precisely control the interference to the PR. Simulation results indicate that the proposed method can significantly reduce the interference to the PR.

In frequency duplex division systems, it is very challenging to position the receiver (Rx) that belongs to the third-party system since the anchors usually do not know which frequency band the Rx uses for transmission. As a result, they can only use the received signal from the third-party transmitter (Tx) to position the Rx. In this paper, we propose a relay-based positioning method to estimate the location of the third-party Rx. In our method, we let each anchor alternatively act as a full-duplex amplify-and-forward (AF) relay for the Rx, which artificially creates a relay path. By estimating the time-difference-of-arrival (TDOA) between the direct and relay paths, the anchors can estimate the Rx location based on the received signal from the third-party Tx. Simulation results indicate the effectiveness of the proposed method.

In frequency division duplex (FDD) two-tier networks, the probing technique has recently been introduced into the cross-channel gain estimation, which requires the tier-two user to act as a relay for the tier-one user. Then the tier-two user can autonomously estimate the cross-channel gain. However, the improper location of the tier-two user, i.e., the relay, may cause severe interference to the tier-one user. In this paper, we analyze the impacts of the probing on the tier-one user and find two location regions, in which the probing does not cause interference. Based on that, we develop a detection method to let the tier-two user autonomously identify its located region. Then the interference caused by the probing can be avoided. Simulation results demonstrate that the proposed method has about 90% correct detection probability.

In cognitive radio networks, the location information of the primary receiver is critical for underlay spectrum sharing. However, positioning the primary receiver in frequency division duplex (FDD) systems is very challenging. In this paper, we propose a novel method to position the primary receiver using the full-duplex amplified-and-forward (AF) relay technique. Simulation results indicate that the proposed method can obtain the same level of estimation error compared with the conventional received signal strength (RSS)-based transmitter positioning methods.

In two-phase cooperative multicast communications, the unbalanced outage probabilities of the cell-center and cell-edge users are the main reason that caps the energy-efficiency of the system. In this paper, we consider the unbalanced outage probability and propose a probability-based relay selection and power control method to improve the energy-efficiency, in which each user can autonomously decide whether to participate in the relay transmission. In particular, we obtain the optimal solution that can minimize the user power consumption. In addition, since our method works in a distributed manner, it does not require any feedback either between the BS and users, or among the users. This saves the extra energy consumption caused by the feedback. Simulation results demonstrate that the proposed method can reduce the user energy consumption up to 54%.

In cognitive radio networks, the signal-to-noise ratio (SNR) of the primary receiver (PR), called primary-link SNR, is critical for underlay spectrum sharing. However, it is very difficult for the cognitive user (CU) to obtain the primary-link SNR. In this letter, we propose a new method to let the CU autonomously estimate the primary-link SNR, where the full-duplex relay technique is used. Then, the CU can conduct the underlay spectrum sharing more efficiently. Simulation results indicate the effectiveness of the proposed method.

In cognitive radio networks, the cross-channel gain from a cognitive transmitter (CT) to a primary receiver (PR) is critical for spectrum sharing and obtaining the cross-channel gain is very difficult. Even though proactive estimation allows the CT to autonomously estimate the cross-channel gain, it may cause severe interference to the PR. This raises a new issue for spectrum sensing, called spectrum sensing interference. In this paper, we deal with the sensing interference and propose a relay-assisted method to conduct the proactive estimation, which obtains the cross-channel gain with much less interference to the PR. In our method, we let the CT act as a full-duplex amplify-and-forward relay to probe the close-loop power control between primary transceivers. By measuring the power adjustment of the primary signal, the CT estimates the cross-channel gain. Simulation results indicate that our method can reduce the sensing interference to an extremely low level.

In wireless positioning, estimating the location of a receiver is very challenging since the receiver does not transmit signals. The difficulty is to obtain the distance between the anchors and the silent receiver. To deal with the issue, this paper proposes to use the full-duplex relay technique to estimate the anchor-receiver distance. Then, we can obtain the location of the silent receiver. With the proposed method, the location-aware applications are not limited to the conventional transmitter positioning. Instead, they can be extended to the receiver positioning. Simulation results demonstrate the performance of the proposed method.

In this letter, we propose a passive detection method to enable cognitive user to detect active primary receiver. With our method, cognitive user may conduct underlay spectrum sharing without the need of channel state information between primary and cognitive users. Simulation results show that our detector can provide about 100% to 300% more access probabilities in average than conventional energy detector.

In this paper, we investigate joint relay selection and power allocation to maximize system throughput with limited interference to licensed (primary) users in cognitive radio (CR) systems. As these two problems are coupled together, we first develop an optimal approach based on the dual method and then propose a suboptimal approach to reduce complexity while maintaining reasonable performance. From our simulation results, the proposed approaches can increase the system throughput by over 50%.

Cognitive radio (CR) networks are designed to utilize the licensed spectrum when it is not used by the primary (licensed) users. In this paper, we investigate how a CR user senses multiple channels and determine the optimal transmission duration and power allocation. When performing optimization, we take energy efficiency, throughput, and interference with the primary users into consideration and find a closed-form solution for transmission duration for chosen channels. It is shown that the proposed optimization approach significantly improves energy efficiency and throughput of CR networks.